We are IntechOpen, the world's leading publisher of Open Access books Built by scientists, for scientists

4,100+

Open access books available

116,000+

International authors and editors

120M+

Downloads

Our authors are among the

Top 1%

most cited scientists

12.2%

Contributors from top 500 universities

Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI)

## Interested in publishing with us? Contact book.department@intechopen.com

Numbers displayed above are based on latest data collected. For more information visit www.intechopen.com

## **Meet the editor**

Dr. Xihong Peng is an Assistant Professor in Applied Sciences and Mathematics at Arizona State University (ASU), Polytechnic Campus. Her research group performs first-principles electronic structure calculations to study properties of materials in multidisciplinary fields such as physics, chemistry, material science and engineering. Her recent research focuses on semiconducting nanostructure

materials, such as Si/Ge nanowires, group III-V and II-VI nanowires. Dr. Peng received Ph. D in Physics from Rensselaer Polytechnic Institute, New York, in 2007. She was a Visiting Assistant Professor in Skidmore College, New York, for the academic year 2007-2008 before joining ASU Polytechnic campus in fall 2008. She is a member of American Physical Society and American Chemical Society.

## Contents

#### **Preface XIII**


#### **Section 2 Silicon Nanowires 149**

Chapter 8 **Silicon Nanowire FinFETs 151** C. Mukherjee and C. K. Maiti

#### Chapter 9 **Hybrid Silicon Nanowires: From Basic Research to Applied Nanotechnology 177** Muhammad Y. Bashouti, Matthias Pietsch, Kasra Sardashti, Gerald Brönstrup, Sebastian W. Schmitt, Sanjay K. Srivastava, Jürgen Ristein, Jordi Arbiol, Hossam Haick and Silke Christiansen


Chapter 16 **Recent Advances in Mechanical Properties of Nanowires 371** Hui Li and Fengwei Sun

**Section 2 Silicon Nanowires 149**

**VI** Contents

Chapter 8 **Silicon Nanowire FinFETs 151**

C. Mukherjee and C. K. Maiti

**Nanotechnology 177**

and Silke Christiansen

**Section 3 Carbon Based Nanowires 261**

Chapter 9 **Hybrid Silicon Nanowires: From Basic Research to Applied**

Chapter 10 **Wet – Chemically Etched Silicon Nanowire Solar Cells:**

Chapter 11 **Thermal Transport in Semiconductor Nanowires 231** J. Anaya, J. Jiménez and T. Rodríguez

Chapter 12 **Carbon Nanotubes Filled With Ternary Chalcohalides 263**

Chapter 13 **Synthesis and Characterization of Advanced Carbon-Based**

**Containing These Nanowires as Fillers 295**

Chapter 14 **Cadmium Selenide-Gold (CdSe-Au) Hybrid Nanowires: Synthesis and Study of Charge Transfer 325**

Chapter 15 **A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical**

Daniela Munteanu and Jean-Luc Autran

**Nanowires – Study of Composites Actuation Capabilities**

V. Salles, L. Seveyrat, T. Fiorido, L. Hu, J. Galineau, C. Eid, B. Guiffard,

**Simulation of Radially-Symmetric Nanowire MOSFETs 341**

Marian Nowak and Marcin Jesionek

A. Brioude and D. Guyomar

Xiaowei Teng and Xiaoxiong Jia

**Section 4 Other Nanowires 323**

**Fabrication and Advanced Characterization 211** Björn Hoffmann, Vladimir Sivakov, Sebastian W. Schmitt,

Muhammad Y. Bashouti, Matthias Pietsch, Kasra Sardashti, Gerald Brönstrup, Sebastian W. Schmitt, Sanjay K. Srivastava, Jürgen Ristein, Jordi Arbiol, Hossam Haick and Silke Christiansen

Muhammad Y. Bashouti, Michael Latzel, Jiří Dluhoš, Jaroslav Jiruse

Chapter 17 **How and Why Alumina Matrix Architecture Influence the Shape and Composition of Nanowires Grown by AC Deposition? 395** Arūnas Jagminas

## Preface

Research on quasi-one dimensional nanowire structure has been a rapid growth field. While InTech scientific publisher has initialized a series books focusing on the fundamental re‐ search of nanowires, this book is intended to give an updated review about the applications of variety types of nanowires, and also associated advancements in synthesis and properties characterization.

Due to the shrinking size, nanowires have a large surface-to-volume ratio, compared with its bulk counterpart. This large ratio renders nanowires with very high sensitivity to surface effect, which enables a new generation of technologies in fields such as sensor and photovol‐ taic. The size miniature is shared by all nanostructures, such as nanowires and quantum dots/clusters. However, nanowires have unique geometrical advantages over other nano‐ structures, which facilitate the utilization of nanowires as electronic device bases, such as channel and interconnect. Many traditional techniques established in bulk devices are readi‐ ly applicable to nanowire devices, resulting in significantly boosted applications of nano‐ wires.

This book contains 17 chapters and is divided into four sections.

Section One (7 chapters) covers recent progress in the metal oxide nanowires, which include novel applications in gas sensor, wetting material, magnetoresistance device, water splitting, etc. Metal oxides, especially transition metal oxides, have many advanced properties, for ex‐ ample, strong catalyst properties and high magnetoresistance coefficient. These unique properties are essentially related to the unfilled d-electrons of transition metals. Combined with enhanced size effect and quantum confinement of nanowires, geometry and stoichiom‐ etry engineering, metal oxide nanowires can display a variety of new properties and are ex‐ pected to play an important role in many different fields.

Section Two (4 chapters) mainly focuses on Silicon nanowires. Si nanowires have been al‐ ways receiving particular interest, since Si is probably one of the most known bulk semicon‐ ductors and the logic device basis-CMOS structure is largely based on Si material. As the ever-reduced size of devices, the logic unit has long entered into nanometer territory. Recent fundamental progress is lying in the realization of a novel device structure - FinFET, in which short channel effect is mitigated and power consumption is tremendously reduced. Chapter 8 in Section Two dedicates a review for the advanced application of Si in FinFET devices. Si nanowires are also demonstrated for promising photovoltaic properties. The de‐ pletion of fossil fuels and global warming has dramatically increased the demands for re‐ newable and clean energy resources. There is no doubt that Si nanowires will be continuous‐ ly explored for its advantages in solar cell. At last, the thermal transport property of Si nano‐ wires is also reviewed in Section Two. This thermal transport property can be crucially im‐ portant for device function. In the mean time, the high thermoelectric coefficient renders Si nanowire a potential for superior thermoelectric devices.

Section Three (2 chapters) is focusing on the carbon based nanotubes and nanowires. A huge amount of applications of carbon nanotubes can be found extensively in literature. In this book, a ternary chalcohalides filled carbon nanotube structure is specifically reviewed, which has a potential in novel applications such as gas storage. Another topic in Section Three is regarding the electromechanical properties of electroactive polymer filled with car‐ bon nanowires.

The advancement of basic understanding of nanowire structures is critically important. For example, in Chapter 16 of the last Section, the mechanical properties of nanowires are inves‐ tigated. In this chapter, we can also see how the advanced metrology (AFM, etc.) and model‐ ing can synthetically boost our understanding. Other nanowire system, including hybrid and metal nanowires will be also reviewed in this section.

Finally, I would like to acknowledge all the authors for their hard work in each individual chapter. I would also greatly appreciate the opportunity provided by InTech publisher, so that such an interesting research theme can be reviewed and shared among the research so‐ ciety. At last, I would like to thank for all the great editing work done by the InTech staff.

> **Dr. Xihong Peng** Applied Sciences and Mathematics Arizona State University, Polytechnic Campus Arizona, United States

**Section 1**

**Oxide Nanowires**

newable and clean energy resources. There is no doubt that Si nanowires will be continuous‐ ly explored for its advantages in solar cell. At last, the thermal transport property of Si nano‐ wires is also reviewed in Section Two. This thermal transport property can be crucially im‐ portant for device function. In the mean time, the high thermoelectric coefficient renders Si

Section Three (2 chapters) is focusing on the carbon based nanotubes and nanowires. A huge amount of applications of carbon nanotubes can be found extensively in literature. In this book, a ternary chalcohalides filled carbon nanotube structure is specifically reviewed, which has a potential in novel applications such as gas storage. Another topic in Section Three is regarding the electromechanical properties of electroactive polymer filled with car‐

The advancement of basic understanding of nanowire structures is critically important. For example, in Chapter 16 of the last Section, the mechanical properties of nanowires are inves‐ tigated. In this chapter, we can also see how the advanced metrology (AFM, etc.) and model‐ ing can synthetically boost our understanding. Other nanowire system, including hybrid

Finally, I would like to acknowledge all the authors for their hard work in each individual chapter. I would also greatly appreciate the opportunity provided by InTech publisher, so that such an interesting research theme can be reviewed and shared among the research so‐ ciety. At last, I would like to thank for all the great editing work done by the InTech staff.

**Dr. Xihong Peng**

Arizona, United States

Applied Sciences and Mathematics

Arizona State University, Polytechnic Campus

nanowire a potential for superior thermoelectric devices.

and metal nanowires will be also reviewed in this section.

bon nanowires.

X Preface

## **Metal-Oxide Nanowires for Gas Sensors**

Supab Choopun, Niyom Hongsith and Ekasiddh Wongrat

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54385

## **1. Introduction**

In past decades, gas sensors based on the metal oxide semiconductors (MOSs) have been studied in diverse field for wide applications. A gas sensor is a device that can be used to detect various gas such as ethanol, LPG, CO2 and CO gases etc. The gas sensors based on MOSs such as SnO2, TiO2, WO3, ZnO, Fe2O3, and In2O3 have an important role in environmental monitor‐ ing, chemical process controlling, personal safety (Q. Wan et al., 2004), industrial process controls, for the detection of toxic environmental pollutants in human health, and for the prevention of hazardous gas leaks, which comes from the manufacturing processes (K. Arshak&I. Gaidan, 2005), wine quality monitoring, and traffic safety (X.F. Song et al., 2009).

The first generation of MOS gas sensors were based on thick films of SnO2 since 1960s which was firstly reported by Taguchi (E. Comini et al., 2009). The MOS gas sensors have some advantages such as small size, low-power-consumption (E. Comoni et al., 2009), simple construction, good sensing properties (K. Arshak&I. Gaidan, 2005), and high compatibility with microelectronic processing (E. Comini et al., 2002). So, they have rapidly gained attention over the years.

Recently, various morphologies of MOS nanostrcutures such as wire-like, belt-like, rod-like, and tetrapods have been widely investigated for gas sensor applications. It is well-known that the sensitivity characteristics of these sensors strongly depend on the morphology of MOS. Especially, one-dimensional nanostructures such as nanowires, nanobelts, nanoneedles have gained a lot of interest for nanodevice design and fabrication (Wang et al., 2008). The sensors based on MOS nanowires are promising due to feasibility for ultrahigh sensitive sensors or ppb-level sensors. These nanowires can be prepared by various techniques such as pulse laser deposition (PLD), chemical vapor deposition (CVD), thermal evaporation, metal-catalyzed

© 2012 Choopun et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Choopun et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

molecular beam epitaxy (CBE) and thermal oxidation technique. Moreover, there are many reports on gas sensor based on the nanowires.

The fundamental mechanism of gas sensing based on MOS depend on the reaction between the surface complexes such as O- , O2-, H+ , and OH reactive chemical species and the gas molecules (reducing/oxidizing gas) to be detected (E. Comini et al., 2009). Thus, it is important to understand the surface reactions between semiconductor surface and target gas for improving the sensing characteristics. Typically, the important parameters in sensor devel‐ opment are sensitivity, selectivity, and stability that called "3S" (E. Comini et al., 2009). However, in this chapter we will mostly discuss on only the sensitivity parameter.

The sensitivity of sensors based on bulk and thin film MOS is typically low. Thus, sensitivity improvement has been extensively studied by using several techniques. The two techniques that commonly used for sensitivity improvement are in the following:

#### **1.** *Using nanostructures*

The MOS sensors based on various nanostructures such as nanowires, nanobelts, nanoparti‐ cles, nanrods, and nanotubes etc. have been demonstrated to be excellent candidates for ultrahigh sensitivity due to their high surface-to-volume ratio. A large surface-to-volume ratio means that a significant fraction of atoms (or molecules) are much quantity on the surface. So, the reaction between target gas and reactive chemical species (O- , O2-, H+ , and OH- ) on the surface can extremely occur. A list of MOS sensors with the sensitivity for the different morphologies of ZnO nanostructures is summarized in Table 1. It can be seen that the sensor sensitivity strongly depends on size and morphology of ZnO nanostructures.

#### **2.** *Adding noble metal*

The noble metals such as Au, Pt, Pd, and Ag on the surface of MOSs can act as a catalyst to modify surface reactions of MOSs toward sensing gas and result in high sensitivity. The ZnO sensors with noble metal additive are also listed in Table 1.

Usually, the gas sensors based on MOS nanostructures exhibit high sensitivity and sometimes up to a few hundred folds over a conventional MOS sensor at a moderate concentration. On the other hand, the sensors based on a larger size of MOS such as in the form of thin film or micro-tetrapod shows lower sensitivity. Several models have been proposed to explain sensitivity characteristics of MOS sensors and still be a subject of discussion.

There are many reports about gas sensor model that are used to explain sensor response characteristics. Wang and co-workers have proposed a surface-depletion model and a contactcontrolled model that are used to describe the sensing mechanism of resistance-type metaloxide semiconductor sensors (Feng et al., 2005). The surface-depletion controlled model is used to explain the sensing mechanism of semiconducting oxide sensors based on nanocrystal/ nanowire/nanobelt structures, while the contact-controlled model is proposed to explain the contact between the outer ends of the rods. This leads to the formation of many junctions in the sensors that significantly modify the potential barrier of contact between rods. These barriers can control the transport of electrons between the rods resulting in the change of the sensor resistance. Chen and co-workers (Chen et al., 2006) have used space-charge model to explain the sensitivity improvement when the grain size is close to or smaller than the Debye length (2*L*d).

molecular beam epitaxy (CBE) and thermal oxidation technique. Moreover, there are many

The fundamental mechanism of gas sensing based on MOS depend on the reaction between

molecules (reducing/oxidizing gas) to be detected (E. Comini et al., 2009). Thus, it is important to understand the surface reactions between semiconductor surface and target gas for improving the sensing characteristics. Typically, the important parameters in sensor devel‐ opment are sensitivity, selectivity, and stability that called "3S" (E. Comini et al., 2009).

The sensitivity of sensors based on bulk and thin film MOS is typically low. Thus, sensitivity improvement has been extensively studied by using several techniques. The two techniques

The MOS sensors based on various nanostructures such as nanowires, nanobelts, nanoparti‐ cles, nanrods, and nanotubes etc. have been demonstrated to be excellent candidates for ultrahigh sensitivity due to their high surface-to-volume ratio. A large surface-to-volume ratio means that a significant fraction of atoms (or molecules) are much quantity on the surface. So,

surface can extremely occur. A list of MOS sensors with the sensitivity for the different morphologies of ZnO nanostructures is summarized in Table 1. It can be seen that the sensor

The noble metals such as Au, Pt, Pd, and Ag on the surface of MOSs can act as a catalyst to modify surface reactions of MOSs toward sensing gas and result in high sensitivity. The ZnO

Usually, the gas sensors based on MOS nanostructures exhibit high sensitivity and sometimes up to a few hundred folds over a conventional MOS sensor at a moderate concentration. On the other hand, the sensors based on a larger size of MOS such as in the form of thin film or micro-tetrapod shows lower sensitivity. Several models have been proposed to explain

There are many reports about gas sensor model that are used to explain sensor response characteristics. Wang and co-workers have proposed a surface-depletion model and a contactcontrolled model that are used to describe the sensing mechanism of resistance-type metaloxide semiconductor sensors (Feng et al., 2005). The surface-depletion controlled model is used to explain the sensing mechanism of semiconducting oxide sensors based on nanocrystal/ nanowire/nanobelt structures, while the contact-controlled model is proposed to explain the contact between the outer ends of the rods. This leads to the formation of many junctions in the sensors that significantly modify the potential barrier of contact between rods. These barriers can control the transport of electrons between the rods resulting in the change of the sensor resistance. Chen and co-workers (Chen et al., 2006) have used space-charge model to

, and OH-

reactive chemical species and the gas

, O2-, H+

, and OH-

) on the

, O2-, H+

However, in this chapter we will mostly discuss on only the sensitivity parameter.

that commonly used for sensitivity improvement are in the following:

the reaction between target gas and reactive chemical species (O-

sensors with noble metal additive are also listed in Table 1.

sensitivity strongly depends on size and morphology of ZnO nanostructures.

sensitivity characteristics of MOS sensors and still be a subject of discussion.

reports on gas sensor based on the nanowires.

the surface complexes such as O-

4 Nanowires - Recent Advances

**1.** *Using nanostructures*

**2.** *Adding noble metal*

In this chapter, we investigate the sensing characteristics of the MOS nanowire sensors and present the sensitivity formulas that are developed in order to explain all circumstances of gas sensors based on MOS nanowires. The size and morphology dependences on the sensitivity are explained in terms of the two important parameters including surface-to-volume ratio and depletion layer width. The developed formulas will be discussed and related to the experi‐ mental sensing characteristics of ZnO sensors.



**Table 1.** List of MOS sensors with the sensitivity for the different morphologies of ZnO nanostructures.

#### **2. Sensing mechanism and sensitivity parameters**

Normally, the gas sensor based on MOS has an optimum operating temperature at high temperature about 250-350°C. When the MOS is heated at lower temperature about 100-200°C, oxygen molecules in the atmosphere are adsorbed on its surface and form oxygen ion mole‐ cules by attracting an electron from the conduction band of MOS as shown in the Eq. (1).

$$O\_2(gas) + e^- \Leftrightarrow O\_2^-(adsorb) \tag{1}$$

At higher temperature, the oxygen ion molecules are dissociated into oxygen ion atoms with singly or doubly negative electric charges by attracting an electron again from the conduction band as shown in Eq. (2) and (3)

$$
\frac{1}{2}O\_2 + e^- \stackrel{k\_{\text{Oty}}}{\Leftrightarrow} O\_{\text{ads}}^- \tag{2}
$$

$$\frac{1}{2}O\_2 + 2\varepsilon^- \Leftrightarrow O\_{\text{ads}}^{2-} \tag{3}$$

where *kOxy* is the reaction rate constant. The oxygen ions on the surface of MOS are extremely active with the target gas molecule and give up the electrons from the surface back to the conduction band of MOS. The generally chemical reaction between gas molecule and oxygen ions is shown in Eq. (4)

$$\text{XX} + \text{O}^{b-}\_{ads} \xrightarrow{k\_{\text{gs}}} \text{X}' + b\text{e}^- \tag{4}$$

where *Χ* and *Χ* ′ is target gas and out gas, respectively. The *b* value is the number of electron and *k*gas is the reaction rate constant of the gas reaction.

The chemical reaction causes change of the carrier concentration in the conductivity and thus, change of sensor resistance. The change of sensor resistance depends on a type of MOSs. A schematic diagram for change of the sensor resistance upon exposure to the target gas (reducing gas) in the cases of n-type and p-type MOS sensors is illustrated in Fig 1 and will be discussed in the following.

#### **2.1. -type metal oxide gas sensor**

**Materials Morphology**

6 Nanowires - Recent Advances

SnO2-Pd (Lee et al., 2008)

ZnO-Au (Hongsith et al., 2008)

SnO2-Pt (Neri et

SnO2-Pd (Lee et al., 2008)

ZnO-Au (Li et al., 2007)

ZnO-Au (Wongrat et al., 2009)

band as shown in Eq. (2) and (3)

ions is shown in Eq. (4)

**Diameter (nm)**

**Target gas**

**Sensitivity (ppm) 1 10 25 50 100 150 200 300 500 1000 2000**

2 2 *O gas e O adsorb* () ( ) - - + Û (1)



nanorod <100 H2 ~33 - - ~700

nanorod <100 Ethanol ~15 - - ~125

nanorod 15 Ethanol 20.1 41.8 89.5 193.6 - -

nanowire 100-500 Ethanol 7 10 20 32

Normally, the gas sensor based on MOS has an optimum operating temperature at high temperature about 250-350°C. When the MOS is heated at lower temperature about 100-200°C, oxygen molecules in the atmosphere are adsorbed on its surface and form oxygen ion mole‐ cules by attracting an electron from the conduction band of MOS as shown in the Eq. (1).

At higher temperature, the oxygen ion molecules are dissociated into oxygen ion atoms with singly or doubly negative electric charges by attracting an electron again from the conduction

> *Oxy k Oe Oads*

*Oxy k OeOads*

where *kOxy* is the reaction rate constant. The oxygen ions on the surface of MOS are extremely active with the target gas molecule and give up the electrons from the surface back to the conduction band of MOS. The generally chemical reaction between gas molecule and oxygen

2

2 1 2

2 <sup>1</sup> <sup>2</sup> 2

al., 2006) nanopowder 6-100 Ethanol ~18 26.5 ~31 - -

**Table 1.** List of MOS sensors with the sensitivity for the different morphologies of ZnO nanostructures.

**2. Sensing mechanism and sensitivity parameters**

nanowire 60-180 Ethanol - 6 12 24 37

Since majority carriers in n-type MOSs are electron, the resistance of n-type MOS sensor decreases when the temperature increases due to their semiconducting properties. However under oxygen ambient, from Eq. (1)-(3), the electrons in the conduction band of n-type MOS are removed by the adsorbed oxygen ions. This causes a decrease of the carrier concentration and thus, an increase of resistance of n-type MOS sensor at operating temperature. When the n-type MOS sensor is under the target gas ambient (reducing gas), the electrons obtained from the chemical reaction as in Eq. (5) are given back to the conduction band leading to a decrease of the sensor resistance.

#### **2.2. p-type metal oxide gas sensor**

In another hand, the majority carriers in p-type MOSs are holes. Similar to n-type MOSs, the sensor resistance of p-type MOS decreases when the temperature increases. However, under oxygen ambient, p-type MOS generates holes when the oxygen ions are adsorbed on the surface via the excited electrons from valence band. This process results in raising the number of charge carriers, which leads to a decrease of the sensor resistance (opposite to n-type). When the p-type MOS sensor is under the target gas ambient (reducing gas), the electrons inject into the valence band and recombine with the holes and this method resulting in reducing the number of holes, which leads to an increase of the sensor resistance (opposite to n-type).

It should be noted that in the case of an oxidizing gas the change of resistance will be just opposite to the above discussions.

In addition, from Eq. (4), a rate equation of an electron density can be written as

$$\frac{dn}{dt} = k\_{\rm gas} [\mathbf{O}\_{\rm ads}]^b [\mathbf{X}]^b \tag{5}$$

where *n* is the electron density or the electron concentration under the gas atmosphere, and *kgas* is the reaction rate constant or reaction rate coefficient described as

**Figure 1.** Schematic diagram for change of the sensor resistance upon exposure to the target gas (reducing gas) in the cases of n-type and p-type MOS sensors.

$$k\_{\rm gas} = A \exp\left(-E\_a / k\_B T\right) \tag{6}$$

where *Ea* is the activation energy of a reaction, *kB* is the Boltzmann constant and *T* is absolute temperature. By integrating Eq. (5) at an equilibrium state under gas ambient and air ambient and using the carrier concentration defined as *n=α/R* (where *R* is a resistance and α is a proportional constant), a sensitivity relation can be obtained as (Hongsith et. al, 2010)

$$m = \Gamma\_t k\_{gas} [\mathbf{O}\_{ads}^{ion}]^b [\mathbf{X}]^b + n\_0 \tag{7}$$

$$S\_g = \frac{R\_a}{R\_g} = \frac{\Gamma\_t k\_{gas} [\mathbf{O}\_{ads}^{ion}]^b [\mathbf{X}]^b}{n\_0} + 1\tag{8}$$

Sometimes, a compact form of the sensitivity relation on gas concentration, *Χ*, can be rewritten as

$$S\_{\mathcal{g}} = aX^{b} + 1\tag{9}$$

where *a* is a controllable parameter.

**Figure 2.** a) sensors based on thin film structrue and (b) based on thin films with perpendicular spike nanowires. The sensitivity enhancement of nanowire sensors can be explained by Eq. (11) which is due to an increase of surface-tovolume ratio.

The sensitivity formula in Eq. (8) can be applied to explain very well in the case of thin film and bulk. In the case of nanostructure, however, two important parameters including surfaceto-volume ratio and depletion layer width need to consider in order to explain the sensing characteristics.

#### *2.2.1. Surface-to-volume ratio*

exp / ( ) *gas a B k A E kT* = - (6)

h+

Oxygen ambient

h+

p-type MOS sensor

x

e- e-

Resistance

Gas in Oxygen

target gas target gas

Adsorbed oxygen ion

*t gas ads n kO n* =G C + (7)

G C = = <sup>+</sup> (8)

*<sup>g</sup> S a* =C+ (9)

where *Ea* is the activation energy of a reaction, *kB* is the Boltzmann constant and *T* is absolute temperature. By integrating Eq. (5) at an equilibrium state under gas ambient and air ambient and using the carrier concentration defined as *n=α/R* (where *R* is a resistance and α is a

**Figure 1.** Schematic diagram for change of the sensor resistance upon exposure to the target gas (reducing gas) in the

<sup>0</sup> [ ][ ] *ion b b*

0

Sometimes, a compact form of the sensitivity relation on gas concentration, *Χ*, can be rewritten

1 *<sup>b</sup>*

[ ][ ] <sup>1</sup> *ion b b*

proportional constant), a sensitivity relation can be obtained as (Hongsith et. al, 2010)

n-type MOS sensor

Gas in

e- e-

*t gas ads a*

*R k O*

*R n*

*g*

*S*

Valence band

cases of n-type and p-type MOS sensors.

Eg

Resistance

ambient

Adsorbed oxygen ion

Conduction band

8 Nanowires - Recent Advances

where *a* is a controllable parameter.

as

*g*

Generally, this surface-to-volume ratio can be related to the density of the adsorbed oxygen ions. Thus, we have proposed that the density of adsorbed oxygen ions can be written in term of surface-to-volume ratio as

$$\mathbb{E}\left[\mathbf{C}\_{ads}^{ion}\right] = \frac{\sigma\_0 \Phi V\_m}{V\_s} \tag{10}$$

where *σ<sup>0</sup>* is a number of oxygen ions per unit area, *Φ* is a ratio of surface area per volume of material *Vm*, and *Vs* is a system volume. By substituting Eq. (10) onto Eq. (8) gives

$$\mathbf{S}\_{\mathcal{g}} = \frac{\Gamma\_t k\_{\mathcal{g}as} \left(\frac{\sigma\_0 \Phi V\_m}{V\_s}\right)^b}{n\_0} \mathbf{X}^b + \mathbf{1} \tag{11}$$

The Eq. (11) can be applied to explain very well in the case of sensors based on MOS nanowires. At the begining, it is quite amazed that sensors based on MOS nanowires exhibited the sensitivity higher than that of thin film or bulk even though the nanowires were not aligned along the conductive direction. For example, sensors based on thin films with perpendicular spike nanowires as shown in Fig.2 exhibit the sensitivity of higher than sensor based on thin film structure. The explanation of sensitivity enhancement is clear with using the Eq. (11) which is due to an increase of surface-to-volume ratio.

**Figure 3.** The sensitivity ratio as a function of diameter where L = 10 μm of a gas sensor based on vertical alignment ZnO nanorods with diamerter D and length L and a gas sensor based on ZnO thin films with an area of 1 cm2.

From Eq. (11), the sensitivity strongly depends on the surface-to-volume ratio (*Φ*) as discussed earlier. For example, let consider a thin film with an area of 1 cm2 as shown in Fig. 2 (a). Then, given cylindrical nanostructures of 10 μm in length with various diameters are grown on this area, as shown in Fig. 2 (b). The surface-to-volume ratio can be calculated and put in Eq. (11) for the sensitivity ratio as

$$\frac{S\_{\Phi}(B) - 1}{S\_{\Phi}(A) - 1} = \left(\frac{\Phi\_{B}}{\Phi\_{A}}\right)^{b} \tag{12}$$

The sensitivity ratio as a function of diameter is plotted as shown in Fig. 3. It can be seen that the sensor sensitivity enhances as decreasing the diameter due to an increase of surface-tovolume ratio and rapidly enhances at small diameter.

#### *2.2.2. Depletion layer width*

Typically, the sensor resistance is contributed from two parts including the resistance along a cylindrical nanowire, and the resistance between nanowires. The resistance along the nanowire is due to surface depletion layer and conductive channel. The resistance between nanowires is due to band bending (potential barrier between wires) which depend on a size of nano‐ structure.

First, let consider for only the resistance along the nanowire by neglecting the resistance between nanowires. According to the depletion layer or the space charge model, *Ld* (Debye length), can be expressed by

#### Metal-Oxide Nanowires for Gas Sensors http://dx.doi.org/10.5772/54385 11

$$L\_d = \left(\frac{\varepsilon k\_B T}{q^2 n}\right)^{1/2} \tag{13}$$

where *ε* is a static dielectric constant, *q* is an electrical charge of a carrier, and *n* is a carrier concentration. It can be seen that at steady operating temperature, the Debye length depends only on the carrier concentration. Let consider in Eq. (13), at the optimum operating temper‐ ature based on pure or undoped MOS nanowire sensor, the *Ld* can be regarded as a constant value and equals to a value of the depletion layer width.

In this model, a cylinder, which is one of the most basic geometric shapes in one-dimension, is considered and a conductive channel is assumed to be along the axis of the cylinder. At an operating temperature, the oxygen ions are adsorbed by attaching an electron on the surface of the cylinder. Therefore, the depletion layer is formed on the surface of cylinder with a thickness of *Ld*, and then a size of conductive channel is reduced along the radial direction as shown in Fig. 4.

When exposed to the reducing gas atmosphere, gas reacts with oxygen ions on the surface and gives back electrons to MOS sensors resulting in increasing conductive channel (decreasing depletion layer width). The conductive channel can be related to the carrier concentration, and can be written in term of the depletion layer width (*Ld*) as

$$m' = n\_0 \frac{\pi \left(D - 2L\_d\right)^2}{\pi D^2} \tag{14}$$

where *n0* is a carrier concentration of a intrinsic material, *n'*represents a carrier concentrationof the Debye length, and *D* is a diameter of the cylinder. Thus, the effect of the depletion layer on sensitivity based on cylindrical MOS nanowire are given by inserting Eq. (14) in Eq. (11) and obtained

$$\mathcal{S}\_{L\_d} = \left(\frac{\Gamma\_t k\_{\rm gas} (\sigma\_0 \Phi(V\_m / V\_s))^b}{n\_0}\right) \frac{D^2 \mathbf{X}^b}{\left(D - 2L\_d\right)^2} + 1\tag{15}$$

Let compare a diameter of cylinder *D* to the Debye length (*L*d). Since the Debye length is in the order of nanometer, it can be divided into three conditions.

#### **Under condition D >> 2Ld, Eq. (15) turns in to Eq. (11)**

From Eq. (11), the sensitivity strongly depends on the surface-to-volume ratio (*Φ*) as discussed earlier. For example, let consider a thin film with an area of 1 cm2 as shown in Fig. 2 (a). Then, given cylindrical nanostructures of 10 μm in length with various diameters are grown on this area, as shown in Fig. 2 (b). The surface-to-volume ratio can be calculated and put in Eq. (11)

**Figure 3.** The sensitivity ratio as a function of diameter where L = 10 μm of a gas sensor based on vertical alignment ZnO nanorods with diamerter D and length L and a gas sensor based on ZnO thin films with an area of 1 cm2.

**0 20 40 60 80 100 120 140 160 180 200**

**Diameter (nm)**


The sensitivity ratio as a function of diameter is plotted as shown in Fig. 3. It can be seen that the sensor sensitivity enhances as decreasing the diameter due to an increase of surface-to-

Typically, the sensor resistance is contributed from two parts including the resistance along a cylindrical nanowire, and the resistance between nanowires. The resistance along the nanowire is due to surface depletion layer and conductive channel. The resistance between nanowires is due to band bending (potential barrier between wires) which depend on a size of nano‐

First, let consider for only the resistance along the nanowire by neglecting the resistance between nanowires. According to the depletion layer or the space charge model, *Ld* (Debye

*b B A*

(12)

() 1 ()1

*S B S A* F F

**S**<sup>F</sup>

**(B)-1/S**F

**(A)-1**

volume ratio and rapidly enhances at small diameter.

for the sensitivity ratio as

10 Nanowires - Recent Advances

*2.2.2. Depletion layer width*

length), can be expressed by

structure.

In this condition, a diameter of cylinder is much larger than micrometer which is the case of microstructure or bulk materials. The depletion layer width is very small compared with the cylindrical diameter (*D >> 2Ld*) and Eq. (15) can be approximated to Eq. (11) which is an equation that has no depletion layer effect.

## **D> 2Ld, Eq. (15) can be approximated to Eq. (11)**

When a diameter of cylinder is in the order of nanometer but still larger than Debye length (*D > L*d), Eq. (15) again can be approximated to Eq. (11) with no depletion layer effect. However, the sensor sensitivity strongly depends on the oxygen ion density due to the surface-to-volume ratio, Φ, parameter.

### **D~2Ld, sensor sensitivity strongly depends on D**

When a cylindrical diameter decreases down to the order of nanometer and is comparable to the Debye length (*D ~ 2Ld*), the depletion layer has strong effect and the sensor sensitivity is strongly dependent on a cylindrical diameter. Thus, Eq. (15) can be used to explain sensitivity of all structural size ranging from nanometer to bulk and can be regarded as a general form of sensitivity formula in order to explain the sensing characteristics.

**Figure 4.** Schematic diagram of the depletion layer on the surface of cylinder with a thickness (or width) of Ld, under condition of D>>2Ld, D>>2Ld, and D~2Ld (not in scale).

Second, let consider for a case that sensor resistance is contributed from both along nanowire and between nanowires. So, the sensor resistance is due to both effects of surface deletion layer and band bending (potential barrier). It should be noted that band bending has also effect on the surface depletion layer. Since the sensor resistance strongly depend on a size of nanowire, let compare a diameter of a cylindrical nanowire *D* with a depletion layer width (*w*). It can be divided into three cases.

#### **Under** *D* **>>2** *w* **(large size)**

In this case, the depletion layer width is much narrow compared to a diameter of the cylinder so the depletion layer has small effect on the electron density in the wires. However, it does greatly affect the potential barrier of the contacts between the wires. Thus, the sensor resistance is dominantly controlled by the potential barrier in this case.

From the oxygen adsorption reaction as given in Eq. (1)-(3) which oxygen ion specie is O2-, at equilibrium condition it can be deduced by using the mass action law that

$$k\_a \left[ e \right]^2 P^{1/2} = k\_d \sigma. \tag{16}$$

$$
\sigma = k\_{\rm Oxy} \left[ e \right]^2 P^{1/2} \tag{17}
$$

where *kOxy* is an equilibrium constant of oxygen adsorption reaction which is defined as *kOxy* = *ka*/*kd*, where*ka* and *kd*is a reaction constant for oxygen adsorption and desorption

**D> 2Ld, Eq. (15) can be approximated to Eq. (11)**

**D~2Ld, sensor sensitivity strongly depends on D**

*Air ambient*

*Reducing gas ambient*

is dominantly controlled by the potential barrier in this case.

equilibrium condition it can be deduced by using the mass action law that

condition of D>>2Ld, D>>2Ld, and D~2Ld (not in scale).

divided into three cases.

**Under** *D* **>>2** *w* **(large size)**

sensitivity formula in order to explain the sensing characteristics.

ratio, Φ, parameter.

12 Nanowires - Recent Advances

When a diameter of cylinder is in the order of nanometer but still larger than Debye length (*D > L*d), Eq. (15) again can be approximated to Eq. (11) with no depletion layer effect. However, the sensor sensitivity strongly depends on the oxygen ion density due to the surface-to-volume

When a cylindrical diameter decreases down to the order of nanometer and is comparable to the Debye length (*D ~ 2Ld*), the depletion layer has strong effect and the sensor sensitivity is strongly dependent on a cylindrical diameter. Thus, Eq. (15) can be used to explain sensitivity of all structural size ranging from nanometer to bulk and can be regarded as a general form of

**Figure 4.** Schematic diagram of the depletion layer on the surface of cylinder with a thickness (or width) of Ld, under

Second, let consider for a case that sensor resistance is contributed from both along nanowire and between nanowires. So, the sensor resistance is due to both effects of surface deletion layer and band bending (potential barrier). It should be noted that band bending has also effect on the surface depletion layer. Since the sensor resistance strongly depend on a size of nanowire, let compare a diameter of a cylindrical nanowire *D* with a depletion layer width (*w*). It can be

In this case, the depletion layer width is much narrow compared to a diameter of the cylinder so the depletion layer has small effect on the electron density in the wires. However, it does greatly affect the potential barrier of the contacts between the wires. Thus, the sensor resistance

From the oxygen adsorption reaction as given in Eq. (1)-(3) which oxygen ion specie is O2-, at

s

(16)

<sup>2</sup> 1/2 . *a d keP k* é ù <sup>=</sup> ë û

D>>2*Ld* D>2*Ld* D~2*Ld*

Conducting channel Depletion layer

reactions, respectively. *σ* is a density of *O2-* per unit area, *e* is a density of conductionelectron at the surface, and *P* is a partial pressure of oxygen molecule. The density of conduction electrons at surface [e] can be expressed in terms of a surface potential as (Yamazoe & Shimanoe, 2008)

$$
\begin{bmatrix} e \\ \end{bmatrix} = n\_0 \exp\left(-\frac{qV\_B}{k\_B T}\right) \tag{18}
$$

where *n*0 is a density of donor electron, which is assumed to be constant throughout MOS and the relation between *VB* and *w* is shown as (Yamazoe & Shimanoe, 2008):

$$\sigma w = \left(\frac{2\varepsilon}{qn\_d}V\_B\right)^{1/2} = L\_d \left(2\beta V\_B\right)^{1/2}, \quad V\_B = \frac{\left(w \mid L\_d\right)^2}{2\beta}.\tag{19}$$

Here *Ld* is the Debye length defined by *LD* = (*εkT/q<sup>2</sup> Nd*) 1/2, and *β* = *q/kT*, where *k* and *T* are Boltzmann constant and temperature, respectively. In addition, *n0* is the density of conduction electrons and apparently is equal to the density of donors, *nd*. By using Eq. (18), we can rewrite it as (Yamazoe & Shimanoe, 2008):

$$\mathbb{E}\left[e\right] = n\_d \exp\left(-\frac{qV\_B}{k\_B T}\right). \tag{20}$$

Since the sensor resistance is dominantly controlled by the potential barrier in this case, the sensor resistance *R*, which is inversely proportional to the electron density is given as (Yamazoe & Shimanoe, 2008)

$$R = R\_0 \exp(m^2 / 2), \qquad R \propto \frac{1}{\left[\frac{e}{e}\right]} \tag{21}$$

where *R0* is a resistance under the flat band condition, *m* is the reduced depletion depth which is defined by *m=w/Ld*. Moreover, if there are no electron-trapping sites other than the adsorbed oxygen ion O2- on the surface, the depletion layer width (*w*) can be related to density of O2- per unit area (*σ*) as (Yamazoe & Shimanoe, 2008):

$$
\sigma = n\_d \text{\textquotedblleft} \tag{22}
$$

From Eq. (22), the larger depletion layer width results in higher density of O2- per unit area (*σ*). Finally, the density of O2- per unit area (*σ*) and [*e*]([*e*]=n0 in this case) can be inserted into Eq. (15) and give (Wongrat et al., 2012)

$$\mathbf{S}\_{\mathcal{g}} = \left( \frac{\Gamma\_t k\_{\rm gas} \left( \Phi n\_d \text{w} (V\_m / V\_s) \right)^b \exp(m^2 / 2)}{n\_d} \right) \mathbf{X}^b + 1 \tag{23}$$

*nd* is density of donor electron, which is assumed to be constant, *w* is the depletion layer width. From Eq. (23) (under *D*>>2*w*), it can be seen that the gas sensing mechanism is controlled by the potential barrier of the contact between wires. Furthermore, the potential barriers are independent of the cylinder size and therefore, the sensitivity is independent of the diameter of nanowire, *D*.

#### **Under** *D>2w*

In this case, the depletion layer has effect on both electron density in the wires and the potential barrier of the contacts between the wires. Thus, the sensor resistance is controlled by surface depletion layer and the potential barrier in this case.The conductance related to the nanowire after oxygen adsorption is given as:

$$G = \frac{e\mu\_0 \left[\underline{e}\right] \pi \left(D - 2w\right)^2}{4l} \tag{24}$$

where *μ0* is the crystal electron mobility, *l* is the distance between electrodes. It can be seen that the gas-sensing mechanism depends on both the width and the height of the contact potential barriers (Sysoev et al., 2009, Rothschild&Komem, 2004). Thus, the sensor resistance due to surface depletion layer and potential barrier is represented by

$$R = R\_0 \left(\frac{D}{D - 2\pi v}\right)^2 \exp\left(m^2 / 2\right). \tag{25}$$

From Eq. (12), (8) and using *e* =*n*0, the sensor response formula can be given as (Wongrat et al., 2012)

$$\mathbf{S}\_{\mathcal{S}} = \left( \frac{\Gamma\_{\text{f}} k\_{\text{gas}} \left( \Phi \eta\_{d} \text{w} \left( V\_{m} \,/\, V\_{s} \right) \right)^{b} \exp \left( m^{2} \,/\, 2 \right)}{n\_{d}} \right) \left( \frac{D}{D - 2w} \right)^{2} \mathbf{X}^{b} + 1 \tag{26}$$

From this equation, it can be seen that the sensor response depend not only on the potential barrier between wires but also on the size of nanowire.

#### **Under** *D≤2w*

*d*

From Eq. (22), the larger depletion layer width results in higher density of O2- per unit area (*σ*). Finally, the density of O2- per unit area (*σ*) and [*e*]([*e*]=n0 in this case) can be inserted into

> ( ) <sup>2</sup> ( / ) exp( / 2) <sup>1</sup> *b t gas d m s b*

*nd* is density of donor electron, which is assumed to be constant, *w* is the depletion layer width. From Eq. (23) (under *D*>>2*w*), it can be seen that the gas sensing mechanism is controlled by the potential barrier of the contact between wires. Furthermore, the potential barriers are independent of the cylinder size and therefore, the sensitivity is independent of the diameter

In this case, the depletion layer has effect on both electron density in the wires and the potential barrier of the contacts between the wires. Thus, the sensor resistance is controlled by surface depletion layer and the potential barrier in this case.The conductance related to the nanowire

( )

<sup>0</sup> 2 4 *e e Dw*

é ù - ë û <sup>=</sup> m

 p

*l*

where *μ0* is the crystal electron mobility, *l* is the distance between electrodes. It can be seen that the gas-sensing mechanism depends on both the width and the height of the contact potential barriers (Sysoev et al., 2009, Rothschild&Komem, 2004). Thus, the sensor resistance due to

2

*<sup>D</sup> R R <sup>m</sup> D w* æ ö <sup>=</sup> ç ÷

<sup>0</sup> exp / 2 . <sup>2</sup>

From Eq. (12), (8) and using *e* =*n*0, the sensor response formula can be given as (Wongrat et

( ( )) ( ) <sup>2</sup> <sup>2</sup> / exp / 2

ç ÷ G F æ ö <sup>=</sup> ç ÷ C +

*n D w*

*b t gas d m s b*

*d k nw V V m <sup>D</sup> <sup>S</sup>*

è ø

æ ö

*G*

surface depletion layer and potential barrier is represented by

2

( )

2

(24)

è ø - (25)

2

è ø -

1

(26)

*d*

è ø

*n* æ ö G F ç ÷ <sup>=</sup> C +

*k nwV V m*

= *n w* (22)

(23)

s

Eq. (15) and give (Wongrat et al., 2012)

*g*

*S*

after oxygen adsorption is given as:

of nanowire, *D*.

14 Nanowires - Recent Advances

**Under** *D>2w*

al., 2012)

*g*

In this case, the depletion layer has effect only on the electron density in the wires and can be regarded as fully depleted cylinders. The fully depleted cylinders are occurred when the Fermi level is totally controlled by surface states under nearly flat energy bands and so potential barrier can be neglected. Therefore, the sensor resistance is controlled only by the surface depletion in this case.

## **3. Experimental, results and discussions**

#### **3.1. Sensor fabrication**

The MOS nanowires have been synthesized by thermal oxidation technique. This technique has been successfully used for synthesizing ZnO or CuO by simply heating pure Zn and Cu material source, respectively. The process is usually conducted in a cylindrical furnace. For the synthesis of ZnO nanowire as shown in Fig. 5, Zn thin films or thick films were screened or evaporated on alumina substrate. Then, they were sintered at temperature in ranging of 500-700ºC to form the nanostructure. For the synthesis of CuO nanowire, the commercial grade copper plate with thickness of 0.1 mm was cut and cleaned by alcohol in an ultrasonic bath for 2 min and dried at room temperature. The copper plate was loaded into a center of a tube furnace at 600°C at normal atmosphere for oxidation reaction.

**Figure 5.** Schematic diagram of the synthesis of ZnO nanowire on alumina substrate by thermal oxidation technique.

The sensors were fabricated by putting gold paste as inter-digital electrodes on the top of the sensors surface and putting a heater underneath the alumina substrate as shown in Fig 6. The heater for the sensor was made from nickel-chromium which could control temperature in the range 280-380 ºC. The operating temperature was measured by a thermocouple placed on the middle of the sensor.

#### **3.2. Setup for ethanol sensing characteristic measurement**

A set up for the ethanol sensing characteristic measurement was shown in Fig. 7. The sensing response andrecoverycharacteristicsofMOS sensorswere studiedinagas flowchamberunder a dynamic equilibrium. Electrical measurements were performed using a volt-amperometric

**Figure 6.** Schematic diagram of the fabricated sensor structures.

technique.Aconstantbiasvoltageof 5Vwas appliedtothe sensor.Anothervoltagewas applied to the heater coil used for heating the sensor and maintaining at a desired operating tempera‐ ture. A thermocouple was employed to monitor the sensor temperature. The ammeter, voltme‐ ter,andthermocouplesignalsweremonitoredandrecordedviaaninterfacedpersonalcomputer. The ethanol sensing properties of the device were observed by the resistance change under an ethanol vapor atmosphere at different operating temperatures. Since our goal was to apply as an alcohol breath analyzer, the ethanol vapor at various concentrations was generated from ethanol solutions using alcohol simulator (GUTH Laboratories, Inc., Harrisburg, PA). The function of this alcohol simulator was to simulate alcohol concentration at conditions similar to exhaled human breath, being varied for ethanol concentration of 50–1,000 ppm for this experi‐ ment. A set up for the ethanol sensing characteristic measurement was illustrated in Fig. 7. The sensor sensitivity is defined by the ratio of the electrical resistance of the sensor in air and the electrical resistance of the sensor in ethanol vapor.

**Figure 7.** Schematic diagram of a set up for the ethanol sensing characteristic measurement.

#### **3.3. ZnO and CuO nanowires**

technique.Aconstantbiasvoltageof 5Vwas appliedtothe sensor.Anothervoltagewas applied to the heater coil used for heating the sensor and maintaining at a desired operating tempera‐ ture. A thermocouple was employed to monitor the sensor temperature. The ammeter, voltme‐ ter,andthermocouplesignalsweremonitoredandrecordedviaaninterfacedpersonalcomputer. The ethanol sensing properties of the device were observed by the resistance change under an ethanol vapor atmosphere at different operating temperatures. Since our goal was to apply as an alcohol breath analyzer, the ethanol vapor at various concentrations was generated from ethanol solutions using alcohol simulator (GUTH Laboratories, Inc., Harrisburg, PA). The function of this alcohol simulator was to simulate alcohol concentration at conditions similar to exhaled human breath, being varied for ethanol concentration of 50–1,000 ppm for this experi‐ ment. A set up for the ethanol sensing characteristic measurement was illustrated in Fig. 7. The sensor sensitivity is defined by the ratio of the electrical resistance of the sensor in air and the

Alcohol simulator Gas chamber

flow out

Gold wire

I-V measurement

Computer

*Alcohol solution*

**Figure 7.** Schematic diagram of a set up for the ethanol sensing characteristic measurement.

Heater

Substrate

MOS

Electrode

16 Nanowires - Recent Advances

Substrate

MOS nanowire

electrical resistance of the sensor in ethanol vapor.

**Figure 6.** Schematic diagram of the fabricated sensor structures.

Air flow in

For the synthesis of ZnO nanowires, it was found that the nanowires exhibit the belt-like or wire-like structures with sharp tip. The diameter and length were in the range of 100-500 nm and 2-7 μm, respectively. For the synthesis of CuO nanowires, the nanowires having a diameter of 100–400 nm and the length of around several micrometers were observed. The different morphologies of ZnO and CuO nanowires were shown in Fig 8. It can be seen that Fig 8 (a) and (b) were for the thick film morphologies. No nanowires were observed in this case representing as the bulk case. Fig 8 (d) and (f) were for vertical aligned and non-vertical aligned ZnO nanowires, respectively. The vertical aligned ZnO nanowires were prepared by zinc screening technique but non-vertical alignment of ZnO nanowires were prepared by zinc evaporation technique. And finally Fig 8 (g) was for the CuO nanowires (Raksa et al., 2005).

**Figure 8.** FE-SEM images of bulk ZnO (a, b), ZnO nanowires (c-f), and CuO nanowires (g).

#### **3.4. Response and recovery curve**

The resistance response and recovery characteristics of ZnO and CuO gas sensor that exposed to ethanol concentration in range of 50-1,000 ppm at the optimum operating are shown in Fig. 9 (a) and (b), respectively. At the beginning, the measured resistance is steady in air atmosphere and then, the resistance of ZnO sensor drops sharply in ethanol atmosphere and recovers to the initial value after removing ethanol vapor. While the resistance of CuO sensor rapidly rises in ethanol atmosphere and falls back after removing ethanol. The decrease and increase of resistance under the ethanol atmosphere is due to the n-type and p-type conducting of ZnO and CuO, respectively as discussed earlier.

The sensitivity of ZnO sensors with different sizes and morphologies at different ethanol concentration are summarized in Table 2. The sensitivity of sensor is defined as *Ra/Rg* (reverse for CuO case) when *Ra* was the resistance of sensor in air and *Rg* was the steady resistance of sensor in ethanol. The sensor sensitivity strongly depends on size and morphology of the sensors. This size and morphology dependence will be discussed in a later section.

**Figure 9.** The response and recovery curve of n-type and p-type MOS sensor for ZnO nanowires (a) and CuO nano‐ wires (Raksa et al., 2005) (b), respectively.


**Table 2.** Sensor sensitivity of sensor with different ZnO sizes and morphologies.

#### **3.5. Linear relation and Sensitivity enhancement based on metal oxide nanowires**

Sometimes, a compact form of the sensitivity relation on gas concentration can be rewritten as

$$S\_{\mathcal{g}} = aX^{b} + 1\tag{27}$$

$$a = \frac{\Gamma\_t k\_{\rm gas} \left(\frac{\sigma\_0 \Phi V\_m}{V\_s}\right)^b}{n\_0} \tag{28}$$

where *a* is a controllable parameter and X is a concentration of gas. At the optimum operating temperature condition, the relation between the sensitivity and gas concentration can be rewritten as:

$$
\log(S\_{\mathcal{g}} - 1) = \log a + b \log X. \tag{29}
$$

It can be seen that log(Sg-1) has a linear relation with log X having a slope of *b* value. Thus, *b* value which represents oxygen ion species on the surface of MOS sensors can be obtained from a slope of a plot between log(Sg-1) and log X.The value of the constant b is normally around either 0.5or 1, depending on the charge state of the surface ion. The adsorbed oxygen ion is O2-for *b* of 0.5, the adsorbed oxygen ion is O¯ for *b* of 1.

The plots between log (S*g*-1) and log Χ of some works as listed in Table 1 was displayed in Fig 10. It was found that the value *b* of all sensors is close to 0.5 suggesting that the adsorbed surface oxygen species on ZnO sensor is *O2-*. In addition, this suggests that the oxygen adsorption species on the surface is independent on the size and morphology of ZnO regardless of bulk, microstructure or nanostructure.

**Figure 10.** Plot of sensor response (*Sg-1*) and ethanol concentration in log scale for ethanol sensor based on different ZnO morphologies. The linear line in the graph has a slope of 0.5.

#### **3.6. Sensor sensitivity**

resistance under the ethanol atmosphere is due to the n-type and p-type conducting of ZnO

The sensitivity of ZnO sensors with different sizes and morphologies at different ethanol concentration are summarized in Table 2. The sensitivity of sensor is defined as *Ra/Rg* (reverse for CuO case) when *Ra* was the resistance of sensor in air and *Rg* was the steady resistance of sensor in ethanol. The sensor sensitivity strongly depends on size and morphology of the

sensors. This size and morphology dependence will be discussed in a later section.

1000 ppm 500 ppm 100 ppm 50 ppm

*Ra>Rg*

11k

*Ra*

**Figure 9.** The response and recovery curve of n-type and p-type MOS sensor for ZnO nanowires (a) and CuO nano‐

**(50 ppm)**

*Nanowire (screening)* 100-500 3.1 3.9 6.4 8.7 **Nanowire(evaporation)** 11-104 17 24 30 40

Sometimes, a compact form of the sensitivity relation on gas concentration can be rewritten as

1 *<sup>b</sup>*

*b*

0

æ ö F G ç ÷ è ø <sup>=</sup> s

*<sup>m</sup> t gas s*

*n*

*<sup>V</sup> <sup>k</sup> V*

0

**3.5. Linear relation and Sensitivity enhancement based on metal oxide nanowires**

*bulk* - 1.3 1.6 2.0 2.2

12k

13k

*Rg*

Time (s) 0 200 400 600 800 1000

> **Sensitivity (500 ppm)**

14k (b)

Gas out

*Ra<Rg*

**Sensitivity (1000 ppm)**

(28)

Gas in

**Sensitivity (100 ppm)**

*<sup>g</sup> S a* =C+ (27)

0 200 400 600 800 1000

Gas out

Time (s)

*Morphology Diameter (nm)* **Sensitivity**

**Table 2.** Sensor sensitivity of sensor with different ZnO sizes and morphologies.

*a*

and CuO, respectively as discussed earlier.

Gas in

Ethanol

*Ra* (a)

1M

*Rg*

wires (Raksa et al., 2005) (b), respectively.

10M

18 Nanowires - Recent Advances

Resistance ()

In the section 2 we have explained the sensing formula for n-type and p-type MOS sensor. In this section, we will explain the sensing parameters for understanding in the mechanism of resistance change under air atmosphere and gas atmosphere and will discuss in term of the size and morphology dependence on the sensor sensitivity.

**Figure 11.** Sensor sensitivity for different ZnO sizes and morphologies with ethanol concentration of 1,000 ppm.

Let consider a sensitivity of MOS gas sensor based on the different sizes and morphologies as seen in Fig. 11, it can be divided into three cases of MOS ethanol gas sensor.

#### *3.6.1. Bulk ZnO*

From Table 2, the sensitivity of a bulk ZnO sensor shows the lowest value compare with that of nanostructures. It can be simply explained by using Eq. (8). The sensing parameter of the surface-to-volume ratio is low for the bulk sensor and thus, the oxygen ion surface density is low resulting in low sensitivity.

#### *3.6.2. Large size nanowires*

From Table 2, it can be seen that the sensitivity of MOS sensor based on vertical aligned nanowires (zinc screening nanowires) is higher than that of bulk. The sensitivity is 8.7 for vertical aligned nanowires sensor at 1,000 ppm of ethanol compared to that of about 2 for bulk ZnO gas sensor. The higher sensitivity of MOS sensor based on vertical aligned nanowires can be explained in terms of surface-to-volume ratio parameter (*Φ)* by using Eq. (11) or (15). Since *Φ* is proportional to sensitivity, the larger *Φ* due to nanostructure gives the higher sensitivity for the case of vertical aligned nanowires even though the nanowires are not aligned along the conducting direction.

#### *3.6.3. Small size nanowires*

It can be seen in Table 2 that the sensitivity of MOS sensor based on non-vertical aligned nanowires (zinc evaporation nanowires) is even higher than that of zinc screening nanowires (the sensitivity increases from 8.7 to 40). This is due to the combination effects of the surfaceto-volume ratio and the depletion layer and can be explained by using Eq. (15). The non-vertical aligned nanowires exhibit smaller diameter (11-104 nm) indicating higher surface-to-volume ratio. Also, the diameter of nanowires is comparable to the depletion layer width (two times of Debye length; about 10 nm for ZnO) indicating strong effect of the depletion layer. Thus, the surface-to-volume ratio parameters (*Φ)* increases and the difference between diameter and the depletion layer width (*D-2L*d) decreases resulting in the increase of the sensor sensitivity as indicated in Eq. (15).

#### **3.7. Ultra high gas sensing by adding novel metal**

Sensitivity

*3.6.1. Bulk ZnO*

20 Nanowires - Recent Advances

low resulting in low sensitivity.

*3.6.2. Large size nanowires*

conducting direction.

*3.6.3. Small size nanowires*

I. Bulk II. Nanowire

Let consider a sensitivity of MOS gas sensor based on the different sizes and morphologies as

From Table 2, the sensitivity of a bulk ZnO sensor shows the lowest value compare with that of nanostructures. It can be simply explained by using Eq. (8). The sensing parameter of the surface-to-volume ratio is low for the bulk sensor and thus, the oxygen ion surface density is

From Table 2, it can be seen that the sensitivity of MOS sensor based on vertical aligned nanowires (zinc screening nanowires) is higher than that of bulk. The sensitivity is 8.7 for vertical aligned nanowires sensor at 1,000 ppm of ethanol compared to that of about 2 for bulk ZnO gas sensor. The higher sensitivity of MOS sensor based on vertical aligned nanowires can be explained in terms of surface-to-volume ratio parameter (*Φ)* by using Eq. (11) or (15). Since *Φ* is proportional to sensitivity, the larger *Φ* due to nanostructure gives the higher sensitivity for the case of vertical aligned nanowires even though the nanowires are not aligned along the

It can be seen in Table 2 that the sensitivity of MOS sensor based on non-vertical aligned nanowires (zinc evaporation nanowires) is even higher than that of zinc screening nanowires (the sensitivity increases from 8.7 to 40). This is due to the combination effects of the surfaceto-volume ratio and the depletion layer and can be explained by using Eq. (15). The non-vertical aligned nanowires exhibit smaller diameter (11-104 nm) indicating higher surface-to-volume

**Figure 11.** Sensor sensitivity for different ZnO sizes and morphologies with ethanol concentration of 1,000 ppm.

seen in Fig. 11, it can be divided into three cases of MOS ethanol gas sensor.

*SI < SII <SIII*

F*<sup>I</sup> <<* F*II <*F*III*

*I. II. III.*

III. Nanowire

In addition, the sensor sensitivity can be further enhanced up to several folds by using adding novel metal technique.There have been several reports on sensitivity enhancement due to metal doping or metal adding effect. We have also reported on the sensitivity enhancement due to gold nanoparticles (AuNPs) on the surface of ZnO (Wongrat et al., 2012).As shown in Fig 12, the sensitivity of ZnO:AuNPs exhibits maximum value of 478 at ethanol concentration of 1,000 ppm while the sensitivity of pure ZnO nanostructure sensor is about 40.

**Figure 12.** Ultra high sensitivity of ZnO sensor withAuNPs on the surface at ethanol concentration of 1,000 ppm (Wongrat et al., 2012).

The metal adding effect can be simply explained in our model by using Eq. (15), (17) and (26). It can be seen that the sensitivity is proportional to the reaction rate constant*, kgas* and *kOxy*, through the oxygen density. Basically, the reaction rate constant can be affected by noble metals (such as Au, Pd, Pt etc.) in MOS sensors due to the catalytic effect. Thus, metal adding causes an increasing of reaction rate constant and resulting in enhancement of the sensor sensitivity.

The evidence for having higher *kgas* can be observed from the sensor resistance in the ethanol ambient. The sensor resistance under ethanol ambient of ZnO:AuNPs sensor is about 200 kΩ which is less than of pure ZnO nanostructure sensor(500 kΩ) suggesting higher *kgas*. In addition, the evidence for having higher *kOxy*, can be observed from the sensor resistance in air. The highest resistance in air of ZnO:AuNPs sensor is about 100 MΩ which is more than of resistance of pure ZnO nanostructure sensor (20 MΩ).The increase of resistance in air suggests that the gold metal catalytically activates the dissociation of molecular oxygen due to higher reaction rate that results in increasing an amount of the oxygen adsorption molecules. Therefore, electrons in ZnO nanowire are captured by oxygen adsorption to form oxygen ion and hence ZnO nanowire loses more electrons and causes the larger depletion layer at ZnO surface resulting in higher resistance.

Therefore, the sensitivity enhancement of ZnO sensors with noble metal is due to catalytic effect in both ethanol adsorption reaction *kgas* and oxygen adsorption reaction *kOxy* through the oxygen density. In the other words, the ultra-high sensitivity of ZnO sensors with noble metal can be explained by an increase of both parameters of *kgas* and *w* (depletion layer width) as described in Eq. (26).

### **4. Summary**

The MOS nanostructures, especially nanowires have been demonstrated to be an excellent candidate as a gas sensing device. The gas sensors based on MOS nanowires are promising due to feasibility for ultrahigh sensitive sensors or ppb-level sensors. The simple and low-cost technique of thermal oxidation technique is effectively used to prepare the MOSs nanowires. The sensing characteristics of the MOS sensors strongly depend on sizes and morphologies of ZnO nanostructures. The sensitivity formulas have been developed in order to explain all circumstances of gas sensors based on MOS nanowires. The size and morphology dependences on the sensitivity have been explained in terms of the two important parameters including surface-to-volume ratio and depletion layer width. The experimental sensing characteristics of ZnO sensors are in good accordance to the developed formulas. This suggests that the sensing formulas are a powerful tool in sensor design and also can be applied for sensors based on other MOSs such as SnO2, TiO2, MoO3 or WO3, etc. In addition, the sensor sensitivity can be further enhanced up to several folds by using adding novel metal technique and can be simply used the developed sensitivity formula for explanation in term of the depletion layer parameter.

### **Symbols used**

*k* Reaction rate *Χ* Density, Concentration, of X *n*, *n0* Electron carrier density *Γ* Time constant *Ra* Resistance in air *Rg* Resistance in active gas *Sg* Sensitivity *Φ* Surface to volume ratio *σ, σ0* Oxygen ion surface density *Vm* Volume of material *Vs* Volume of system *Cg* Ethanol concentration *Ld* Debye length *ε* Static dielectric constant *q* Electrical charge of carrier *kB* Boltzmann constant *T* Temperature (K)

and hence ZnO nanowire loses more electrons and causes the larger depletion layer at ZnO

Therefore, the sensitivity enhancement of ZnO sensors with noble metal is due to catalytic effect in both ethanol adsorption reaction *kgas* and oxygen adsorption reaction *kOxy* through the oxygen density. In the other words, the ultra-high sensitivity of ZnO sensors with noble metal can be explained by an increase of both parameters of *kgas* and *w* (depletion layer width) as

The MOS nanostructures, especially nanowires have been demonstrated to be an excellent candidate as a gas sensing device. The gas sensors based on MOS nanowires are promising due to feasibility for ultrahigh sensitive sensors or ppb-level sensors. The simple and low-cost technique of thermal oxidation technique is effectively used to prepare the MOSs nanowires. The sensing characteristics of the MOS sensors strongly depend on sizes and morphologies of ZnO nanostructures. The sensitivity formulas have been developed in order to explain all circumstances of gas sensors based on MOS nanowires. The size and morphology dependences on the sensitivity have been explained in terms of the two important parameters including surface-to-volume ratio and depletion layer width. The experimental sensing characteristics of ZnO sensors are in good accordance to the developed formulas. This suggests that the sensing formulas are a powerful tool in sensor design and also can be applied for sensors based on other MOSs such as SnO2, TiO2, MoO3 or WO3, etc. In addition, the sensor sensitivity can be further enhanced up to several folds by using adding novel metal technique and can be simply used the developed sensitivity formula for explanation in term of the depletion layer

surface resulting in higher resistance.

described in Eq. (26).

22 Nanowires - Recent Advances

**4. Summary**

parameter.

**Symbols used**

*Χ* Density, Concentration, of X

*n*, *n0* Electron carrier density

*Rg* Resistance in active gas

*Φ* Surface to volume ratio

*σ, σ0* Oxygen ion surface density

*k* Reaction rate

*Γ* Time constant

*Sg* Sensitivity

*Ra* Resistance in air

## **Acknowledgements**

This work was supported by Thailand Research Fund (TRF). Also, we would like to thank P. Raksa for useful discussion.

## **Author details**

Supab Choopun1,3, Niyom Hongsith2,3 and Ekasiddh Wongrat2

1 Department of Physics and Materials Science, Faculty of Science, Chiang Mai University Chiang Mai, Thailand

2 School of Science, University of Phayao, Phayao, Thailand

3 ThEP Center, CHE, Bangkok, Thailand

### **References**


## **Synthesis of ZnO Nanowire by MOCVD Technique: Effect of Substrate and Growth Parameter**

Sachindra Nath Das, Jyoti Prakash Kar, Junjie Xiong and Jae-Min Myoung

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54577

## **1. Introduction**

[6] Wongrat, E, Pimpang, P, & Choopun, S. Appl. Surr. Sci., (2009).

Benedetti, B. D. Sensor. Actuator B, (2006).

[12] Arshak, K, & Gaidan, I. Mater. Sci. Eng. B, (2005).

Wang, C. Sensor. Actuator B, (2010).

Actuators B, (2012).

24 Nanowires - Recent Advances

S. (2008).

[7] Wongrat, E, Hongsith, N, Wongratanaphisan, D, Gardchareon, A, & Choopun, S. Sens.

[8] Neri, G, Bonavita, A, Micali, G, Donato, N, Deorsola, F. A, Mossino, P, Amato, I, &

[9] Zhang, H. N, Li, Z. Y, Liu, L, Xu, X. R, Wang, Z. J, Wang, W, Zheng, W, Dong, B, &

[10] Son, J. Y, Lim, S. J, Cho, J. H, & Seong, W. K. and Hyungjun Kim,Appl. Phys.Lett., (2008).

[11] Xu, J. Q, Zhang, Y, Chen, Y. P, Xiang, Q, Pan, Q. Y, & Shi, L. Y. Mater. Sci. Eng. B, (2008).

[14] Hongsith, N, Viriyaworasakul, C, Mangkorntong, P, Mangkorntong, N, & Choopun,

[15] Hongsith, N, Wongrat, E, Kerdcharoen, T, & Choopun, S. Sens. Actuators B, (2010).

[19] Raksa, P, Gardchareon, A, Chairuangsri, T, Mangkorntong, P, Mangkorntong, N, &

[21] Choopun, S, Hongsith, N, Mangkorntong, P, Mangkorntong, N, & Physica, E. (2007).

[22] Sysoev, V. V, Schneider, T, Goschnick, J, Kiselev, I, Habicht, W, Hahn, H, Strelcov, E,

[26] Liu, Y. X, Gao, C. T, Pan, X. J, An, X. Y, Xie, Y. Z, Zhou, M, Song, J, & Zhang, H. L. Z.Y.

[28] Song, X. F, Wang, Z. J, Liu, Y. B, Wang, C, & Li, L. J. Nanotechnology, (2009).

[29] Yang, Z, Li, L. M, Wan, Q, Liu, Q. H, & Wang, T. H. Sens. Actuators B, (2008).

[13] Bie, L. J, Yan, X. N, Yin, J, Duan, Y. Q, & Yuan, Z. h. Sens. Actuators B, (2007).

[16] Wang, N, Cai, Y, & Zhang, R. Q. Mater. Sci. Eng. R-Rep., (2008).

[20] Wan, Q, Li, Q. H, Chen, Y. J, & Wang, T. H. Appl. Phys.Lett., (2004).

[23] Lee, Y. C. Hui Huang, O.K. Tan, and M.S. Tse, Sens. Actuators B, (2008).

[17] Yamazoe, N, & Shimanoe, K. Sens. Actuators B, (2008).

[18] Feng, P, Wan, Q, & Wang, T. H. Appl. Phys.Lett., (2005).

Choopun, S. Ceramics International, (2009).

& Kolmakov, A. Sens. Actuators B, (2009).

[24] Chen, Y. J, Zhu, C. L, & Xiao, G. Sens. Actuators B, (2008).

[25] Chen, Y. J, Zhu, C. L, & Xiao, G. Nanotechnology, (2006).

[27] Liu, Q, Zhao, Y. H, & Zhang, E. Q. Xie, Appl. Surr. Sci., (2010).

ZnO nanostructures have been studied extensively in the past few years due to their funda‐ mental and technological importance [1-5]. In particular, it is a wide-direct-band gap (~3.37 eV), II–VI semiconductor with many potential applications such as nanolaser arrays [2,6], gas sensors [7-10], field emission devices [11-12] and luminescent materials [13-16]. A great deal of attention has been focused on one-dimensional nanostructures because of their supe‐ rior properties such as high surface-to-volume ratio, high crystalline quality, and quantum confinement effects, which are suitable for various electronics and optoelectronics applica‐ tions. Up to now, many ZnO configurations have been reported such as nanobelts, nano‐ wires [10–11], nanoneedles [3], nanotetrapods [9], nanocombs [17] and so on. It has also been suggested that ZnO nanostructures are probably the most abundant forms of any known materials. Therefore, proper control of processing parameters is essential to reproduce the desired nanostructures. Synthesis of aligned one-dimensional ZnO nanowire arrays has at‐ tracted much attention, which includes the techniques of metal-organic chemical vapour deposition (MOCVD), vapor-liquid-solid (VLS), pulsed laser deposition method (PLD) and solution based chemical technique [21-24]. A solution-based technique is an attractive ap‐ proach for the growth of ZnO nanostructures because of its simplicity and low cost [25, 26]. In addition, this method allows large-scale growth of aligned ZnO nanowires on an arbitra‐ ry substrate, which promotes the integration of ZnO nanomaterials for various applications. But the main drawback of chemical process is the poor quality of nanowire. On the other hand, physical process are of particular interest since they have many advantages such as the ability to fabricate nanostructures of better quality, well controllable configurations, and good reproducibility. From a physical point of view, the synthesis methods for one-dimen‐

© 2012 Das et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Das et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

sional ZnO nanostructures are categorized into two groups, metal catalyst-assisted growth and catalyst-free growth which corresponded to vapor–liquid–solid and vapor–solid mecha‐ nisms, respectively. The latter is better than the former in terms of no metal catalytic impuri‐ ty. On the other hand, MOCVD is superior because, it has abilities to produce large area; catalytic free, long, vertically aligned ZnO NWs, which have excellent morphological, struc‐ tural and optical properties [27]. As the growth of ZnO nanowires by MOCVD is a bottom– up technique, the nature of substrates has a vital role for the determination of nanowire di‐ mension and alignment. According to lattice misfit, the most suitable substrate for ZnO growth is ScAlMgO4, which is expensive and technologically inconvenient [28]. Interesting‐ ly, C-plane sapphire overcomes some of the limitations. On the other hand, ZnO nanostruc‐ tures on R-plane sapphire have an advantage for the fabrication of the next generation sensors and detectors in the form of surface acoustic wave (SAW) devices due to higher elec‐ tromechanical coupling [29-31]. However, in the pursuit of next generation ZnO-based nano devices, it would be highly preferred if well-ordered ZnO nanowires with high aspect ratio could be aligned onto cheap and CMOS compatible substrates, such as silicon. In addition, silicon is the most popular substrate for microelectronics and micro electro mechanical sys‐ tems (MEMS), and it serves as a good integration platform. But, it is difficult to obtain wellaligned ZnO nanowires on silicon substrates, because the formation of an interfacial layer (SiO2) and large lattice misfit [32,33]. On the other hand, microchannel-based solid-state de‐ vices on Si have played an important role for chemical, gas and flow sensors with compact size and superior performance [34-36]. The growth of vertically aligned ZnO microtubes on a silicon wafer without any catalyst are of particular interests for monolithic integration of micro/nano devices.

As far as the growth of ZnO nanowires on glass or SiO2 layer is concerned, deterioration in the quality and alignment of the nanowires come into picture because of the lack of specific epitax‐ ial arrangements. In such cases the nanowires become quasi-aligned instead of vertical aligned. Since, MOCVD is a bottom up technique, the introduction of an appropriate interfacial layer may also solve the above problem. It is more preferable if the interfacial layer would be trans‐ parent and conductive for its integration with optoelectronic devices. Taking these issues into consideration, Ga doped ZnO (GZO) film may be a suitable interfacial layer for the growth of vertical aligned ZnO nanowires. In addition, GZO film can be used as a transparent conduct‐ ing layer during the integration of the devices. Furthermore, the cost of GZO is much lower than conventionally used indium doped transparent conductive oxide layer. The nature of the substrate with a interfacial layer also take a significant role on the shape of nanowire. The con‐ trol over size and morphology of nanometer and micrometer-sized semiconductor materials represents a great challenge to realizing the design of novel functional devices. Several efforts have been directed towards the synthesis of nanostructures to control the shape, aspect ratio (length/width), growth site, and growth direction. One salient feature of the ZnO-based nano‐ structures is a gradual transition between different morphologies with different substrate which leads to the investigation for the effects of substrate and growth parameter on MOCVD technique*.*

## **2. Experimental**

sional ZnO nanostructures are categorized into two groups, metal catalyst-assisted growth and catalyst-free growth which corresponded to vapor–liquid–solid and vapor–solid mecha‐ nisms, respectively. The latter is better than the former in terms of no metal catalytic impuri‐ ty. On the other hand, MOCVD is superior because, it has abilities to produce large area; catalytic free, long, vertically aligned ZnO NWs, which have excellent morphological, struc‐ tural and optical properties [27]. As the growth of ZnO nanowires by MOCVD is a bottom– up technique, the nature of substrates has a vital role for the determination of nanowire di‐ mension and alignment. According to lattice misfit, the most suitable substrate for ZnO growth is ScAlMgO4, which is expensive and technologically inconvenient [28]. Interesting‐ ly, C-plane sapphire overcomes some of the limitations. On the other hand, ZnO nanostruc‐ tures on R-plane sapphire have an advantage for the fabrication of the next generation sensors and detectors in the form of surface acoustic wave (SAW) devices due to higher elec‐ tromechanical coupling [29-31]. However, in the pursuit of next generation ZnO-based nano devices, it would be highly preferred if well-ordered ZnO nanowires with high aspect ratio could be aligned onto cheap and CMOS compatible substrates, such as silicon. In addition, silicon is the most popular substrate for microelectronics and micro electro mechanical sys‐ tems (MEMS), and it serves as a good integration platform. But, it is difficult to obtain wellaligned ZnO nanowires on silicon substrates, because the formation of an interfacial layer (SiO2) and large lattice misfit [32,33]. On the other hand, microchannel-based solid-state de‐ vices on Si have played an important role for chemical, gas and flow sensors with compact size and superior performance [34-36]. The growth of vertically aligned ZnO microtubes on a silicon wafer without any catalyst are of particular interests for monolithic integration of

As far as the growth of ZnO nanowires on glass or SiO2 layer is concerned, deterioration in the quality and alignment of the nanowires come into picture because of the lack of specific epitax‐ ial arrangements. In such cases the nanowires become quasi-aligned instead of vertical aligned. Since, MOCVD is a bottom up technique, the introduction of an appropriate interfacial layer may also solve the above problem. It is more preferable if the interfacial layer would be trans‐ parent and conductive for its integration with optoelectronic devices. Taking these issues into consideration, Ga doped ZnO (GZO) film may be a suitable interfacial layer for the growth of vertical aligned ZnO nanowires. In addition, GZO film can be used as a transparent conduct‐ ing layer during the integration of the devices. Furthermore, the cost of GZO is much lower than conventionally used indium doped transparent conductive oxide layer. The nature of the substrate with a interfacial layer also take a significant role on the shape of nanowire. The con‐ trol over size and morphology of nanometer and micrometer-sized semiconductor materials represents a great challenge to realizing the design of novel functional devices. Several efforts have been directed towards the synthesis of nanostructures to control the shape, aspect ratio (length/width), growth site, and growth direction. One salient feature of the ZnO-based nano‐ structures is a gradual transition between different morphologies with different substrate which leads to the investigation for the effects of substrate and growth parameter on MOCVD

micro/nano devices.

26 Nanowires - Recent Advances

technique*.*

ZnO nanowires were grown on different substrates (e.g. different types of sapphire, Si, Si mi‐ crochanel, GZO coated glass, ZnO microcreators) using MOCVD technique with diethylzinc (DEZn) as the zinc precursor. Argon and oxygen (5N) were used as carrier and reactive gases, respectively. The base pressure of the reactor chamber was 10−6 Torr. During the growth, the substrate temperature, working pressure and the nozzle-substrate spacing were fixed, but they were varied for different substrates to synthesize desired nanostructure. Morphological inves‐ tigations of the nanowires were carried out using field-emission scanning electron Microscope (FESEM). The crystallinity of the nanowires was studied by X-Ray diffractometer. In addition, the microstructures of ZnO NWs were observed using high-resolution transmission electron microscope (HRTEM). For the cross-sectional TEM and energy dispersive X-ray (EDX) study, the samples were prepared by focused ion beam (FIB) system.

## **3. Result and discussion**

#### **3.1. Effect of sapphire substrate**

The morphology of nanostructures greatly depends on the growth parameters as well as the underlying substrates. c-plane sapphire substrates were used for the growth of ZnO nanowires by MOCVD technique with substrate temperature and substrate to nozzle distance at 630 °C and 3 cm, respectively. The experimental details are given in table 1. Fig. 1(a–c) is typical FES‐ EM images of ZnO nanowires arrays prepared with different growth conditions. The samples constitute uniform and densely packed nanowires arrays with different surface morphologies, namely nanoneedles (Fig. 1(a)), nanonails (Fig. 1(b)) and nanowires with rounded tip (Fig. 1(c)). Cross-sectional view of the samples (inset of Fig 1.) revealed that the nanowires were grown almost perpendicularly on the substrates and their lengths were about 3 μm. It is known that each nucleation site usually supported growth of only one nanowires, but when the nucle‐ ation sites were close together, two nanowires combined into one with a larger diameter [37] and different surface morphologies were obtained. From XRD pattern (Fig. 2), a prominent ZnO (0002) peak is observed, which corresponds to the growth of ZnO nanowires along c-axis. No characteristic peaks indicating other impurities are observed in the samples. This is attrib‐ uted to the hexagonal wurtzite structure of nanowires.

In order to investigate the structural characteristics of ZnO nanowires, TEM study of individual nanowires is a very important characterization technique. The bright field TEM micrographs of the ZnO nanowires and the corresponding HRTEM image (inset) are shown in Fig. 3. The HRTEM images indicated that the nanowires were structurally uniform and did not exhibit any noticeable defects. The HRTEM image showed well resolved lattice with an inter-planar spacing of 0.52 nm, which corresponds to the (0001) plane of wurtzite structure. Hence, one may infer that the nanowires possessed a single-crystal hexagonal structure with good crystalline quality along the [0001] axis.

**Figure 1.** Surface morphologies of vertically aligned ZnO nanowires: (a) nanoneedles, (b) nanonails and (c) nanowires with rounded tip.

**Figure 2.** XRD pattern of a representative ZnO nanowire sample

Synthesis of ZnO Nanowire by MOCVD Technique: Effect of Substrate and Growth Parameter http://dx.doi.org/10.5772/54577 29

**Figure 3.** TEM images of ZnO NRs: (a) nanoneedles, (b) nanonails and (c) nanorods with rounded tip. Inset shows the corresponding HRTEM images


**Table 1.** Growth parameter of ZnO nanorod arrays.

**Figure 1.** Surface morphologies of vertically aligned ZnO nanowires: (a) nanoneedles, (b) nanonails and (c) nanowires

with rounded tip.

28 Nanowires - Recent Advances

**Figure 2.** XRD pattern of a representative ZnO nanowire sample

To investigate the effect of lattice mismatch on the growth mechanism, ZnO nanowires were grown on A (112¯0), C (0001) and R (1 ¯012) plane sapphires using MOCVD system. Here, the working pressure of the reactor was maintained at 6 Torr and the distance be‐ tween the susceptor and the nozzle was fixed at 5 cm. Fig. 3 depicts the FESEM images of ZnO nanostructures on various sapphire substrates. Vertically aligned nanowires of width 100–150 nm and length 3–4 μm are grown on A- and C-plane sapphires (Fig. 4(a) and (b)), but nanopencil/tip like ZnO nanostructures of length 1.5–2.5 μm is observed for the R-sapphire substrate (Fig. 4(c)). The average basal diameter of the nanopencils is around 250–300 nm and its shape becomes narrow towards top-end. Sharp tip of the nanostructures may be due to the residual amount of precursors at the end of the growth process. For better understanding of the growth process, nanostructures were grown for a longer time (90 min). For both A- and C-plane sapphires, the nanowires retain their verti‐ cal nature and are uniform throughout its length, which clearly indicates the growth is along c-axis (Fig. 4(d)). On the other hand, Fig. 4(e) shows a typical image of ZnO nano‐ structures on R-plane sapphire, in which lateral growth has been taken place indicating a deterioration of c-axis growth. This may be due to the generation of stress during the ini‐ tial stage of growth and afterwards these nanostructures join together to make thicker structures. The magnified image (inset of Fig. 4(e)) revealed that the thicker nanostruc‐ tures are randomly oriented and their surface is appeared to be comparatively rough. The bright-field (BF) TEM image of single ZnO nanowire is shown in Fig. 4 (e). This shows the diameter of the nanowires grown on A, C sapphire are around 100 nm and where as it is around 300 nm for the nanowire grown on R sapphire. The tip of the nanowire was gradually tapered down from several hundred of nanometers to 50 nm. This is due to the lack of the fresh reactant supply towards the final stage of growth process. At the initial stage of the growth, the width of the nanowires were thin and the individual thin nano‐ wires laterally merged together to give thicker one with increase in growth time. General‐ ly, the formation mechanism of nanostructures consists of two stages: nucleation and growth. At first, nanosized crystalline nuclei are formed on the substrate and afterwards the nanowires start to grow along a specific direction. The diameter of the nanowires de‐ pends on the size and density of the initial ZnO nuclei. In case of C-plane sapphire, ZnO has a partial ionic character, which originates due to the opposite charge on the (0001) and (0001¯) planes being Zn and O termination, respectively [38]. Thus there is a net di‐ pole moment when the crystal is terminated by the basal plane, which causes the surface energy to diverge. On the other hand, the (112¯0) plane is nonpolar as it comprises an equal number of O and Zn atoms, and hence has a lower surface energy [39-40]. There‐ fore the growth rates of ZnO nanostructures on C- and R-plane sapphires are different, which is confirmed from the variation in aspect ratio (length/width) of the nanostruc‐ tures. Hence, it is expected that the condition of supersaturation is different for the growth of ZnO nanostructures on C and R-plane sapphires. At present, it is believed that there is a formation of an unintentional thin interfacial ZnO layer during the initial stage of growth on R-plane sapphire. Furthermore, the shape of the nanowires was change to nanopencil using an intentional thin ZnO film as an interfacial layer [41]. This indicates that the different shape, size and density of the nanowires can be attributed to the inplane stress during the initial stage of growth, which comes from thermal and lattice mis‐ matches between ZnO nanostructures and the substrates [42-43]. According to crystallographic point of view, Miyauchi et al. [44] reported that the probability of oxy‐ gen vacancies at the terminated surfaces of ZnO nanostructures grown on R-plane sap‐ phire is higher than others. The surface morphology as well as crystallinity of ZnO nanostructures, grown on R-plane sapphire, is quite different from A- and C-plane sap‐ phires. For R-plane sapphire, it is expected that simultaneous existence of O and Zn atoms in (112¯0) plane plays a vital role on the property of nanostructure.

To investigate the effect of lattice mismatch on the growth mechanism, ZnO nanowires

Here, the working pressure of the reactor was maintained at 6 Torr and the distance be‐ tween the susceptor and the nozzle was fixed at 5 cm. Fig. 3 depicts the FESEM images of ZnO nanostructures on various sapphire substrates. Vertically aligned nanowires of width 100–150 nm and length 3–4 μm are grown on A- and C-plane sapphires (Fig. 4(a) and (b)), but nanopencil/tip like ZnO nanostructures of length 1.5–2.5 μm is observed for the R-sapphire substrate (Fig. 4(c)). The average basal diameter of the nanopencils is around 250–300 nm and its shape becomes narrow towards top-end. Sharp tip of the nanostructures may be due to the residual amount of precursors at the end of the growth process. For better understanding of the growth process, nanostructures were grown for a longer time (90 min). For both A- and C-plane sapphires, the nanowires retain their verti‐ cal nature and are uniform throughout its length, which clearly indicates the growth is along c-axis (Fig. 4(d)). On the other hand, Fig. 4(e) shows a typical image of ZnO nano‐ structures on R-plane sapphire, in which lateral growth has been taken place indicating a deterioration of c-axis growth. This may be due to the generation of stress during the ini‐ tial stage of growth and afterwards these nanostructures join together to make thicker structures. The magnified image (inset of Fig. 4(e)) revealed that the thicker nanostruc‐ tures are randomly oriented and their surface is appeared to be comparatively rough. The bright-field (BF) TEM image of single ZnO nanowire is shown in Fig. 4 (e). This shows the diameter of the nanowires grown on A, C sapphire are around 100 nm and where as it is around 300 nm for the nanowire grown on R sapphire. The tip of the nanowire was gradually tapered down from several hundred of nanometers to 50 nm. This is due to the lack of the fresh reactant supply towards the final stage of growth process. At the initial stage of the growth, the width of the nanowires were thin and the individual thin nano‐ wires laterally merged together to give thicker one with increase in growth time. General‐ ly, the formation mechanism of nanostructures consists of two stages: nucleation and growth. At first, nanosized crystalline nuclei are formed on the substrate and afterwards the nanowires start to grow along a specific direction. The diameter of the nanowires de‐ pends on the size and density of the initial ZnO nuclei. In case of C-plane sapphire, ZnO has a partial ionic character, which originates due to the opposite charge on the (0001) and (0001¯) planes being Zn and O termination, respectively [38]. Thus there is a net di‐ pole moment when the crystal is terminated by the basal plane, which causes the surface energy to diverge. On the other hand, the (112¯0) plane is nonpolar as it comprises an equal number of O and Zn atoms, and hence has a lower surface energy [39-40]. There‐ fore the growth rates of ZnO nanostructures on C- and R-plane sapphires are different, which is confirmed from the variation in aspect ratio (length/width) of the nanostruc‐ tures. Hence, it is expected that the condition of supersaturation is different for the growth of ZnO nanostructures on C and R-plane sapphires. At present, it is believed that there is a formation of an unintentional thin interfacial ZnO layer during the initial stage of growth on R-plane sapphire. Furthermore, the shape of the nanowires was change to nanopencil using an intentional thin ZnO film as an interfacial layer [41]. This indicates that the different shape, size and density of the nanowires can be attributed to the in-

¯012) plane sapphires using MOCVD system.

were grown on A (112¯0), C (0001) and R (1

30 Nanowires - Recent Advances

**Figure 4.** Micrographs of ZnO nanostructures on (a) A-, (b) C-, (c) R-plane sapphires for 15 min growth, (d) C-, (e) Rplane sapphires for 90 min growth, and (f) bright-filed TEM images [J. P. Kar et al. Applied Surface Science 256 (2010) 4995].

#### **3.2. Effect of Si Substrate**

To investigate the effect of substrate, ZnO nanostructures were grown on Si substrate and Si with ZnO interfacial layer. Pencil like ZnO nanostructures were obtained on Si sub‐ strate with ZnO thin film as an interfacial layer (Fig. 5). Here, the substrate temperature, working pressure and substrate to nozzle distance were kept at 600 °C, 3 Torr and 3 cm, respectively. With the decrease in substrate to nozzle distance (~1 cm) and increase in the substrate temperature 700 OC, a hallow microtube like structures were grown. FE-SEM images of the vertically aligned ZnO microtubes with a hollow tubular shape are shown in Fig. 6. The magnified SEM images (insets) show that all of the microtubes have a hol‐ low cavity and a faceted hexagonal shape. The tubular wall-thickness is in the range of micrometers towards base and it reduced to nanometers range towards top-end. The average outer basal diameter of the microtubes is around 10 μm and its shape becomes narrow (5 μm) towards top end. The length of the ZnO microtubes is about 40 μm and the diameter is ~4 μm towards the open tip whereas it becomes 1–1.5 μm at the basal part. Typical crystal morphology and a representative cross-sectional view of a single ZnO microtube is shown in Fig. 6(c). Hollow basal part of the tubular structure (inset of Fig. 6(c)) consists of many nanospikes (width ~50–120 nm and length ~500 nm). We be‐ lieve that these nanospikes grow further during the course of the growth process and merge together to make solid surface, which in fact reduces the inner diameter. From XRD pattern (not shown here), a prominent ZnO (0002) peak is observed, which corre‐ sponds to the growth of ZnO microtubes along c-axis. No characteristic peaks indicating other impurities are observed in the samples. This is attributed to the hexagonal wurtzite structure of microtubes. Vertically aligned ZnO microtubes display inerratic or perfect hexagon-shaped cross-sectional structure and open tips. At the initial stage of growth, several vertically aligned nanowires are generated due to the high growth rate of ZnO crystal along [0001] direction, which is thermodynamically stable. As the growth time in‐ creases, the ZnO nanowires may grow laterally even though the lateral growth rate is much lower than that of the vertical direction. Moreover, a high reaction temperature may accelerate the lateral growth of the ZnO nanowires. Therefore, neighboring ZnO nanowires gradually start to coalesce together to form hexagonal circles. As the reaction time increases, the complete prismatic ZnO microtubes are finally formed by the coales‐ cence of ZnO nanowires. Furthermore, the substrate was very close (1 cm) to the nozzle of the precursor, which favored the reasonable supply of precursors. It is clear that the morphology changed from the nanowires to microtubes with the in increase growth tem‐ perature and decrease in nozzle-substrate spacing. Hence, the higher growth temperature and lower nozzle-substrate spacing favor the "oriented attachment" process of the nano‐ wires to form microtubes. At present, it is believed that substrate temperature has more influence for the quality of the microtubes. On the other hand, it has also been reported that catalysts like carbon and metastable zinc-rich oxide layer played a vital role for the generation of tubular structures [45,46].

Synthesis of ZnO Nanowire by MOCVD Technique: Effect of Substrate and Growth Parameter http://dx.doi.org/10.5772/54577 33

**3.2. Effect of Si Substrate**

32 Nanowires - Recent Advances

generation of tubular structures [45,46].

To investigate the effect of substrate, ZnO nanostructures were grown on Si substrate and Si with ZnO interfacial layer. Pencil like ZnO nanostructures were obtained on Si sub‐ strate with ZnO thin film as an interfacial layer (Fig. 5). Here, the substrate temperature, working pressure and substrate to nozzle distance were kept at 600 °C, 3 Torr and 3 cm, respectively. With the decrease in substrate to nozzle distance (~1 cm) and increase in the substrate temperature 700 OC, a hallow microtube like structures were grown. FE-SEM images of the vertically aligned ZnO microtubes with a hollow tubular shape are shown in Fig. 6. The magnified SEM images (insets) show that all of the microtubes have a hol‐ low cavity and a faceted hexagonal shape. The tubular wall-thickness is in the range of micrometers towards base and it reduced to nanometers range towards top-end. The average outer basal diameter of the microtubes is around 10 μm and its shape becomes narrow (5 μm) towards top end. The length of the ZnO microtubes is about 40 μm and the diameter is ~4 μm towards the open tip whereas it becomes 1–1.5 μm at the basal part. Typical crystal morphology and a representative cross-sectional view of a single ZnO microtube is shown in Fig. 6(c). Hollow basal part of the tubular structure (inset of Fig. 6(c)) consists of many nanospikes (width ~50–120 nm and length ~500 nm). We be‐ lieve that these nanospikes grow further during the course of the growth process and merge together to make solid surface, which in fact reduces the inner diameter. From XRD pattern (not shown here), a prominent ZnO (0002) peak is observed, which corre‐ sponds to the growth of ZnO microtubes along c-axis. No characteristic peaks indicating other impurities are observed in the samples. This is attributed to the hexagonal wurtzite structure of microtubes. Vertically aligned ZnO microtubes display inerratic or perfect hexagon-shaped cross-sectional structure and open tips. At the initial stage of growth, several vertically aligned nanowires are generated due to the high growth rate of ZnO crystal along [0001] direction, which is thermodynamically stable. As the growth time in‐ creases, the ZnO nanowires may grow laterally even though the lateral growth rate is much lower than that of the vertical direction. Moreover, a high reaction temperature may accelerate the lateral growth of the ZnO nanowires. Therefore, neighboring ZnO nanowires gradually start to coalesce together to form hexagonal circles. As the reaction time increases, the complete prismatic ZnO microtubes are finally formed by the coales‐ cence of ZnO nanowires. Furthermore, the substrate was very close (1 cm) to the nozzle of the precursor, which favored the reasonable supply of precursors. It is clear that the morphology changed from the nanowires to microtubes with the in increase growth tem‐ perature and decrease in nozzle-substrate spacing. Hence, the higher growth temperature and lower nozzle-substrate spacing favor the "oriented attachment" process of the nano‐ wires to form microtubes. At present, it is believed that substrate temperature has more influence for the quality of the microtubes. On the other hand, it has also been reported that catalysts like carbon and metastable zinc-rich oxide layer played a vital role for the

**Figure 5.** FE-SEM images of ZnO nanowires grown on (a) Si substrate, (b) Si substrate with ZnO interfacial layer.

**Figure 6.** FE-SEM (a) 45º tilted, (b) top, and (c) cross-sectional images of ZnO microtubes grown on Si substrates [J. P. Kar et al. Materials Letters 63 (2009) 2327].

To investigate the effect of the shape of Si substrate, ZnO nanostructures were grown by MOCVD technique on Si microchanel. Before ZnO growth, Si substrates were patterned by wet chemical etching process to form microcavity. At first, 200-nm-thick silicon nitride masking layer was grown by low-pressure chemical vapor deposition (LPCVD) at 800 0 C in the mixture of di-chloro silane (SiH2Cl2) and ammonia (NH3) with 240 mTorr pres‐ sure.Thereafter, standard photolithography and etching techniques were adopted in order to fabricate micropatterns. Reactive ion etcher (RIE) was used to remove silicon nitride layer from the micro- patterned windows and the etch rate was 350 nm/min. Afterwards, the sam‐ ples were immersed in KOH solution for 10 hour to fabricate the truncated pyramid-shaped etch patterns. Before the growth of ZnO nanowires, no passivation layer was intentionally used. The substrate temperature was kept at 700 0 C during the nanowire growth. The work‐ ing pressure of the reactor was maintained at 3 Torr and the distance between the susceptor and the nozzle was fixed at 1.5cm. Fig. 7 and Fig. 8 show the crosssectional image (FESEM) of the different microchannel in silicon substrate, which contains ZnO nanowires. Three parts (top surface [Fig. 7(a), (b) and Fig. 8(a) and (b)], side wall [Fig. 7(c) and Fig. 8(c)] and bottom surface [Fig. 7(d), (e) and Fig. 8(d) and (e)]) have ZnO nanowires with different di‐ mensions and surface morphologies. On the top of silicon substrate the nanowires were al‐ most uniform along the length (2 mm) and width of 200–250 nm. The aspect ratio, packing fraction and the number density of nanowires on top surface are around 10, 0.8 and 107 per mm2 , respectively. The packing fractions of nanowires were determined by measuring the area occupied by the nanowires on a top view micrograph. Interestingly, vertical nanopen‐ cil /tip like ZnO nanostructure was observed towards the bottom of the trench. The ZnO nanowires have preferentially grown along a particular direction, which his normal to the substrate surface. It is expected that at the bottom surface the nanowires would have a high perpendicular anisotropy because of the high aspect ratio and the low packing fraction. Figs. 7(e) show the microstructures of ZnO nanowires at the bottom corner of the microchannel. Furthermore, randomly distributed few nanoflowers are also observed at side wall sand bot‐ tom of the microchannel. It may be generated from the defects of Si surface, arrived during etching of silicon. The flower has several wings, which have grown from the common ori‐ gin. These happen due to the different directional growth of ZnO nanowires at the defect points. The possible growth mechanisms of ZnO nanowires are stated below. At the top sur‐ face, high-density island were nucleated on the substrate and grew preferentially along the c-axis orientation. The continuous growth led to the coalescence of these islands and as a re‐ sult wider nanowires, with a low density and a reduced aspect ratio, were grown. But at the inclined surface, the growth direction (c-axis) and the direction of incoming precursor mole‐ cules are not same.

So, the width of nanowires is less than the top surface. Now at the bottom surface of the mi‐ crochannel, number of incoming fresh precursor molecule is less than the top surface due to stagnant of gaseous species at the cavity. It is assumed that the sharpness of the tips origi‐ nates due to the lack of fresh reactant supply to the bottom of cavity during the growth of nanostructures. Thus, the probability of coalescence of grain is less, which leads to the for‐ mation of thin nanowires compared to the top surface.

To investigate the effect of the shape of Si substrate, ZnO nanostructures were grown by MOCVD technique on Si microchanel. Before ZnO growth, Si substrates were patterned by wet chemical etching process to form microcavity. At first, 200-nm-thick silicon nitride masking layer was grown by low-pressure chemical vapor deposition (LPCVD) at 800 0

the mixture of di-chloro silane (SiH2Cl2) and ammonia (NH3) with 240 mTorr pres‐ sure.Thereafter, standard photolithography and etching techniques were adopted in order to fabricate micropatterns. Reactive ion etcher (RIE) was used to remove silicon nitride layer from the micro- patterned windows and the etch rate was 350 nm/min. Afterwards, the sam‐ ples were immersed in KOH solution for 10 hour to fabricate the truncated pyramid-shaped etch patterns. Before the growth of ZnO nanowires, no passivation layer was intentionally

ing pressure of the reactor was maintained at 3 Torr and the distance between the susceptor and the nozzle was fixed at 1.5cm. Fig. 7 and Fig. 8 show the crosssectional image (FESEM) of the different microchannel in silicon substrate, which contains ZnO nanowires. Three parts (top surface [Fig. 7(a), (b) and Fig. 8(a) and (b)], side wall [Fig. 7(c) and Fig. 8(c)] and bottom surface [Fig. 7(d), (e) and Fig. 8(d) and (e)]) have ZnO nanowires with different di‐ mensions and surface morphologies. On the top of silicon substrate the nanowires were al‐ most uniform along the length (2 mm) and width of 200–250 nm. The aspect ratio, packing fraction and the number density of nanowires on top surface are around 10, 0.8 and 107

, respectively. The packing fractions of nanowires were determined by measuring the area occupied by the nanowires on a top view micrograph. Interestingly, vertical nanopen‐ cil /tip like ZnO nanostructure was observed towards the bottom of the trench. The ZnO nanowires have preferentially grown along a particular direction, which his normal to the substrate surface. It is expected that at the bottom surface the nanowires would have a high perpendicular anisotropy because of the high aspect ratio and the low packing fraction. Figs. 7(e) show the microstructures of ZnO nanowires at the bottom corner of the microchannel. Furthermore, randomly distributed few nanoflowers are also observed at side wall sand bot‐ tom of the microchannel. It may be generated from the defects of Si surface, arrived during etching of silicon. The flower has several wings, which have grown from the common ori‐ gin. These happen due to the different directional growth of ZnO nanowires at the defect points. The possible growth mechanisms of ZnO nanowires are stated below. At the top sur‐ face, high-density island were nucleated on the substrate and grew preferentially along the c-axis orientation. The continuous growth led to the coalescence of these islands and as a re‐ sult wider nanowires, with a low density and a reduced aspect ratio, were grown. But at the inclined surface, the growth direction (c-axis) and the direction of incoming precursor mole‐

So, the width of nanowires is less than the top surface. Now at the bottom surface of the mi‐ crochannel, number of incoming fresh precursor molecule is less than the top surface due to stagnant of gaseous species at the cavity. It is assumed that the sharpness of the tips origi‐ nates due to the lack of fresh reactant supply to the bottom of cavity during the growth of nanostructures. Thus, the probability of coalescence of grain is less, which leads to the for‐

used. The substrate temperature was kept at 700 0

mm2

34 Nanowires - Recent Advances

cules are not same.

mation of thin nanowires compared to the top surface.

C in

per

C during the nanowire growth. The work‐

**Figure 7.** FESEM images of ZnO nanowires on (a) microchannel, (b) Top surface, (c-e) magnified image of bottom sur‐ face.

**Figure 8.** FESEM images of ZnO nanowires on (a) microchannel, (b) Top surface, (c) bottom surface, (d) magnified im‐ age of bottom surface, (e) bottom corner.

#### **3.3. Effect of interfacial layer**

To investigate the effect of interfacial layer on ZnO growth by MOCVD technique, Ga dop‐ ed ZnO coated glass and ZnO microcators were used as substrates. Ga-doped ZnO films were deposited by DC sputtering of GZO target on corning glass (1737) substrates. The sub‐ strate temperature, working pressure and sputtering power were fixed around 280 ºC, 3×10-3 torr and 1.2 kW, respectively. The substrate temperature and working pressure were fixed at 600 ºC, 3 Torr and 1.5 cm, respectively. The evolution of columnar NWs were possible when the flow rate of O2 (20 sccm) was lower than the carrier Ar gas (50 sccm).

**(a) (b)** 

**(c) (d)** 

**Figure 8.** FESEM images of ZnO nanowires on (a) microchannel, (b) Top surface, (c) bottom surface, (d) magnified im‐

To investigate the effect of interfacial layer on ZnO growth by MOCVD technique, Ga dop‐ ed ZnO coated glass and ZnO microcators were used as substrates. Ga-doped ZnO films were deposited by DC sputtering of GZO target on corning glass (1737) substrates. The sub‐

**(e)** 

36 Nanowires - Recent Advances

age of bottom surface, (e) bottom corner.

**3.3. Effect of interfacial layer**

Vertical aligned long ZnO nanowires (width 400-500 nm) were then grown on GZO coated glass substrates and the FESEM image is shown in Fig. 9(b). In order to compare, ZnO nano‐ wires were also grown on c-sapphire substrates keeping all of the parameters same (Fig. 9(a)). This observation depicts the lengths (~ 25 μm) of the nanowires are similar for both the substrates. These kind of long nanowires are suitable for the single nanowire based devices, where the larger length is feasible for various electrical contacts. XRD patterns (not shown here) of ZnO nanowires, grown on GZO/glass substrate shows the presence of (0002) and (0004) peaks. The absence of additional peaks in the XRD pattern excludes the possibility of any extra phases. The nanowires exhibited a prominent (0002) peak with full width at half maximum (FWHM) values around 0.06, indicating high crystallinity of the ZnO nanowires. These data reveal that the ZnO nanowires are all c-axis oriented with wurtzite structure. The bright-field (BF) TEM image of single ZnO nanowire is shown in Fig. 10 (a). This shows the diameter of the nanowires is 460 nm without any side branches. The tip of the nanowire was gradually tapered down from several hundred of nanometers to 50 nm. This is due to the lack of the fresh reactant supply towards the final stage of growth process. The dopant re‐ distribution at high process temperature is a usual phenomenon during the fabrication of semiconductor devices. Herein, the nanowires were grown at 600 ºC and hence there may be a chance of dopant diffusion at the interfacial layer. In this regard, cross-sectional TEM as well as EDX spectroscopic measurements of the ZnO nanowires on GZO film are carried out. Figure 10 (a) shows the enlarged image of the interface between the ZnO nanowires GZO films. This enlarged image indicates that the sputtered film has columnar grains and the nanowires are grown along the columnar grains of the GZO film.

**Figure 9.** FESEM of ZnO NWs on (a) c-plane sapphire. (b) GZO/glass substrates

**Figure 10.** a) Cross-sectional TEM image across the interface of ZnO NWs and GZO film, and EDX spectroscopic map‐ ping images of (a) Zn, (b) O and (c) Ga across the interface.

The mapping images Fig. 10(b-c), of Zn and O atoms were distributed uniformly at the inter‐ face, whereas nonuniform distributions of the Ga atoms are clearly observed (Fig.10 (d)). The Ga atoms are appeared not only in the underlying film, but also seen towards the bot‐ tom of the nanowires. The Ga concentration in the nanowires decreases with increase in dis‐ tance from the interface. This observation indicates that the growth temperature provided sufficient thermal energy to Ga atoms in the interfacial film to be diffused into the lower part of nanowires. The concentration of Ga is estimated as 2.95 at.% and 0.32 at.% for GZO film and the bottom of the ZnO nanowires, respectively. The lightly incorporation of Ga to‐ wards the bottom of the nanowires, during the initial stage of growth, does not alter the col‐ umnar growth due to the similar atomic radius of Ga and Zn as well as the comparable covalent bond lengths of Ga-O and Zn-O. Yan et al. observed the evolution of nanostruc‐ tures with Ga doping in contrast to undoped ZnO, where Stranski-Krastanov (SK) model was used to explain the phenomena [47]. We have already observed that undoped ZnO film on Si substrate favours the formation of low density, thick nano-pencil structure by MOCVD technique. Thus Ga has an ability to scale down the grain size of the GZO film, which stimu‐ lates the formation of large number of homogeneous nucleation sites at the initial stage of nanowires growth. The c-axis orientation in ZnO nanowires can be explained by the ''sur‐ vival of the fastest'', where the nuclei were formed with various orientations at the initial stage of the deposition and each nucleus competed to grow but only the nuclei in c-axis ori‐ entations could survive due to the fastest growth rates. Once the nanowires have grown, the high-speed laminar gas flow under our specific growth parameters induces turbulent flow between the nanostructures, which results in adsorption of fresh reactant gases only on nanowires tips. Hence the growth of very long nanowires is achieved. It has been reported that beyond a certain doping level ZnO nanowires retains its vertical shape and above this the tip of the nanostructures bends due to the generation of stacking faults and the surface becomes very rough. In the present case, such kind of abnormality has not been seen be‐ cause of the mild doping of Ga at the bottom of the nanowires during the initial growth.

**Ga-ZnO Film** 

38 Nanowires - Recent Advances

**Glass Substrate** 

**500 nm** 

ping images of (a) Zn, (b) O and (c) Ga across the interface.

**ZnO nanowires** 

**(a) (b)** 

**(c) (d)** 

**Figure 10.** a) Cross-sectional TEM image across the interface of ZnO NWs and GZO film, and EDX spectroscopic map‐

The mapping images Fig. 10(b-c), of Zn and O atoms were distributed uniformly at the inter‐ face, whereas nonuniform distributions of the Ga atoms are clearly observed (Fig.10 (d)). The Ga atoms are appeared not only in the underlying film, but also seen towards the bot‐ tom of the nanowires. The Ga concentration in the nanowires decreases with increase in dis‐ tance from the interface. This observation indicates that the growth temperature provided sufficient thermal energy to Ga atoms in the interfacial film to be diffused into the lower part of nanowires. The concentration of Ga is estimated as 2.95 at.% and 0.32 at.% for GZO

To study the effect of interfacial layer as well as the shape of the substrate, ZnO micro-craters were fabricated. To do so, AZO thin films were deposited on Corning glass (1737) by pulsed DC sputtering of an AZO ceramic target (Al ~ 2 wt.%) with Ar (5 N) as the ambient gas. The sputtering chamber was evacuated to a base pressure of about 5x10-5 Torr. Prior to the sputter‐ ing, the target was pre-sputtered to remove any residual contamination from the target sur‐ face. The deposition parameters, pulse frequency (2 kHz), DC power (2 KW), substrate temperature (200 0 C), working pressure (3 mTorr) and deposition time (10 min), were opti‐ mized and kept constant for all the films. Then the AZO films were anisotropically etched for 90 s by dilute hydrochloric acid (0.5% HCl in H2O) in order to create textured surfaces. Fig. 11(a–b) shows the 450 tilted FESEM images of the as-deposited film and the AZO films etched for 90 s. Before etching, the film is very smooth (Fig. 11(a)). After etching, hexagonal, conical craters with diameters of 0.5–2.5 lm are created on the film surface (Fig. 11(b)). The formation of craters mainly occurs at positions of O-terminated ZnO columns, as it is highly sensitive to acid attack. The exposed facets of the etched surfaces may mainly be (1 0 1) plane [48]. There‐ after, the etched AZO films were put into a metal–organic chemical vapor deposition (MOCVD) chamber for growth of ZnO nanowires. During the growth period, the substrate temperature, working pressure and the nozzle-substrate spacing were fixed at 550 0 C, 3 Torr and 1 cm, respectively. The deposition time was 5 min. ZnO nanowires onto its surface to fab‐ ricate the hierarchical structure. Fig. 11(c) shows the SEM images of the hierarchical structure of the ZnO nanowires. All surfaces of the craters are covered by nanowires with cone-shaped tips, 50–100 nm in diameter and 150–300 nm in length. The structure as shown is very similar to that of lotus leaf. The lotus leaf surface is a good example of how, the combination of a mi‐ cro-/nano-hierarchical structure and low surface energy wax layer, exhibits high water con‐

**Figure 11.** FESEM images (45° view) of the AZO films: (a) as-deposited, (b) etched for 90 seconds (c) nanowires on etched sample. (d) AFM image with cross section profile (J. Xiong et al J. Colloid and Interface Science 350 (2010) 344.).

tact angle (CA) and low CA hysteresis. A superhydrophobic and self-cleaning surface mimics the surface of a lotus leaf. The combination of both self-cleaning and antireflective properties is very desirable for applications in outdoor photovoltaic and display devices, self-cleaning windows and car windshields. A two step method, using a combination of top down and bot‐ tom-up approaches, for fabrication of ZnO based hierarchical structures with nanowires on microcraters helps to mimic the lotus leaf structure. Fig. 11(d) shows the AFM image with the corresponding typical cross sectional profile of the etched films after the growth of nano‐ wires. The microcraters can be easily recognized by the hexagonal holes on the surface. Due to the co-existence of micro- and nano-scale features and the non-vertical geometric alignment, the structure can induce multiple scattering of the incident light. The co-existence of microand nano-structures can interact with light of different wavelengths, causing reflective sup‐ pression over a wide wavelength range [49]. Another advantage of the hierarchical structure over flat surface film having an antireflection property that is less angle dependent. This indi‐ cates a potential application in developing Si thin film solar cell, where it can act as a front electrode (replacing ITO, etched AZO or textured FTO film) to achieve improved light trap‐ ping. Furthermore, it can also be used in the dye-sensitized solar cell as an electrode for the collection of electrons. The hierarchical structure can trap light, improve dye absorption and introduce a nano-scale pathway for carrier transportation, similar with the flower-like ZnO structure. Similarly, this structure can also be used for enhancing the extraction efficiency of LED, which is usually realized by using photonic crystals, conductive omnidirectional reflec‐ tors, or surface roughening of transparent oxide electrodes (normally ITO).

## **4. Conclusion**

Among the different synthesis process, MOCVD is superior to reproduce large area; catalyt‐ ic free, long, vertically aligned ZnO NWs with excellent morphological, structural and opti‐ cal properties. As the growth of ZnO nanowires by MOCVD is a bottom–up technique, the nature of substrate, shape of the substrate and interfacial layer on the substrate have vital role for the determination of nanowire dimension and alignment. According to lattice misfit, the most suitable substrate for ZnO growth is ScAlMgO4, which is expensive and technolog‐ ically inconvenient. Generally, the formation mechanism of nanostructures consists of two stages: nucleation and growth. At first, nanosized crystalline nuclei are formed on the sub‐ strate and afterwards the nanowires start to grow along a specific direction. The c-axis orien‐ tation in ZnO nanowires can be explained by the ''survival of the fastest'', where depending on the nature of the substrate, the nuclei were formed with various orientations at the initial stage of the deposition and each nucleus competed to grow but only the nuclei in c-axis ori‐ entations could survive due to the fastest growth rates. Once the nanowires have grown, the high-speed laminar gas flow under our specific growth parameters induces turbulent flow between the nanostructures, which results in adsorption of fresh reactant gases only on nanowires tips. Hence the growth of very long nanowires is achieved. At the initial stage of the growth, the width of the nanowires were thin and the individual thin nanowires lateral‐ ly merged together to give thicker one with increase in growth time. Thus, the diameter of the nanowires depends on the size and density of the initial ZnO nuclei. Also the supply of the fresh reactant during the growth process takes a vital role on the nanostructure. Among different substrates, nanowires on C-plane sapphire show the excellent optical and structur‐ al properties. On the other hand, by changing the deposition condition of MOCVD and type of substrate one can synthesize ZnO nanowire of desired structure for different application.

## **Acknowledgements**

tact angle (CA) and low CA hysteresis. A superhydrophobic and self-cleaning surface mimics the surface of a lotus leaf. The combination of both self-cleaning and antireflective properties is very desirable for applications in outdoor photovoltaic and display devices, self-cleaning windows and car windshields. A two step method, using a combination of top down and bot‐ tom-up approaches, for fabrication of ZnO based hierarchical structures with nanowires on microcraters helps to mimic the lotus leaf structure. Fig. 11(d) shows the AFM image with the corresponding typical cross sectional profile of the etched films after the growth of nano‐ wires. The microcraters can be easily recognized by the hexagonal holes on the surface. Due to the co-existence of micro- and nano-scale features and the non-vertical geometric alignment, the structure can induce multiple scattering of the incident light. The co-existence of microand nano-structures can interact with light of different wavelengths, causing reflective sup‐ pression over a wide wavelength range [49]. Another advantage of the hierarchical structure over flat surface film having an antireflection property that is less angle dependent. This indi‐ cates a potential application in developing Si thin film solar cell, where it can act as a front

**Figure 11.** FESEM images (45° view) of the AZO films: (a) as-deposited, (b) etched for 90 seconds (c) nanowires on etched sample. (d) AFM image with cross section profile (J. Xiong et al J. Colloid and Interface Science 350 (2010)

344.).

40 Nanowires - Recent Advances

**(c)** 

**(a) (b)** 

**(d)** 

This research was supported by WCU (World Class University) program through the Na‐ tional Research Foundation of Korea funded by the Ministry of Education, Science and Technology (R32-20031).

#### **Author details**

Sachindra Nath Das1,3, Jyoti Prakash Kar2,3, Junjie Xiong3 and Jae-Min Myoung3

1 Department of Physics, Burdwan Raj College, Burdwan, West Bengal, India


#### **References**


[18] Li Y., Liu F., Sun J., Chem. Commun. 2009; 2730.

**Author details**

42 Nanowires - Recent Advances

**References**

2005; 86: 203115.

2004; 84: 3654.

2007;18: 225.

nology 2007; 18: 035606.

Phys. Lett. 2004; 394: 452.

Sachindra Nath Das1,3, Jyoti Prakash Kar2,3, Junjie Xiong3

[1] Pan Z.W., Dai Z.R., Wang Z.L. Science 2001; 291: 1947.

[4] Xu C.X., Sun X.W. Appl. Phys. Lett. 2003; 83: 3806.

[7] Xu J., Pan Q., Shun Y., Tian Z., Sens. Actuators B 2000; 66: 277.

[12] Huang Y., Yu K., Zhu Z., Curr. Appl. Phys. 2007; 7: 702.

[13] Park S., Kim S., Han S., Nanotechnology 2007; 18: 055608.

[14] Wu L., Wu Y., Pan X., Kong F., Optical Mater. 2006; 28: 418.

[5] Gao P.X., Wang Z.L. Small 2005; 1: 945.

1 Department of Physics, Burdwan Raj College, Burdwan, West Bengal, India

2 Department of Physics, National Institute of Technology, Rourkela, Odisha, India

3 Department of Material Science and Engineering, Yonsei University, Seoul, South Korea

[2] Lee C.J., Lee T.J., Lyu S.C., Zhang Y., Ruh H., Lee H.J. Appl. Phys. Lett 2002; 81: 3648. [3] Zhao Q., Zhang H.Z., Zhu Y.W., Feng S.Q., Sun X.C., Xu J., Yu D.P. Appl. Phys. Lett.

[6] Yan H., He R., Johnson J., Law M., Saykally R.J., Yang P. J. Am. Chem. Soc. 2003; 125: 4728.

[8] Cheng X.L., Zhao H., Huo L.H., Gao S., Zhao J.G., Sens. Actuators B 2004;102: 248. [9] Xiangfeng C., Dongli J., Djurišic A.B., Leung Y.H., Chem. Phys. Lett. 2005; 401: 426.

[10] Wan Q., Li Q.H., Chen Y.J., Wang T.H., He X.L., Li J.P., Lin C.L., Appl. Phys. Lett.

[11] Liao L., Zhang W.F., Lu H.B., Li J.C., Wang D.F., Liu C., Fu D.J., Nanotechnology,

[15] Mahalingam T., Lee K.M., Park K.H., Lee S., Ahn Y., Park J.Y., Koh K.H., Nanotech‐

[17] Leung Y.H., Djurišić A.B., Gao J., Xie M.H., Wei Z.F., Xu S.J., Chan W.K., Chem.

[16] Kang H.S., Kang J.S., Kim J.W., Lee S.Y., Appl. Phys. Lett. 2004; 95: 1246.

and Jae-Min Myoung3


**Provisional chapter**

## **Ferromagnetic ZnO Nanowires for Spintronic Applications Applications**

**Ferromagnetic ZnO Nanowires for Spintronic**

Usha Philipose and Gopal Sapkota Additional information is available at the end of the chapter

Usha Philipose and Gopal Sapkota

Additional information is available at the end of the chapter 10.5772/52825

http://dx.doi.org/10.5772/52825

## **1. Introduction**

[42] Kar J.P., Lee S.W, Lee W., Myoung J.M., Appl. Surf. Sci. 2008; 254: 6677.

[45] Miyauchi M., Shimai A., Tsuru Y., J. Phys. Chem. B 2005; 109: 13307.

[48] Li M.K, Wang D.Z, Shi F., Ding S., Hong J. Chin Phys Lett 2007;24:236–9.

[44] Zhu Q.S., Wang Z.G., Appl. Phys. Lett. 2005; 87: 231903

[46] Sun T, Qiu J. Mater Lett 2008; 62:1528.

2005;87:093110.

44 Nanowires - Recent Advances

[43] Kar J. P., Das S. N., Choi J. H., Lee T. I., Myoung J. M., Appl. Surf. Sci. 2010; 256: 4995.

[47] Xu W.Z, Ye Z.Z, Ma D.W, Lu H.M, Zhu L.P, Zhao B.H, Appl Phys Lett

[49] Ellmer K., Klein A., Rech B., Transparent Conductive Zinc Oxide, Springer, 2007.

[50] Zhang L., Zhang X., Dai Z., Wu J., Zhao N., Xu J., J. Colloid Interface Sci. 2010; 345: 116.

Dilute Magnetic Semiconductors (DMS) are expected to be the main building blocks for the realization of semiconductor spin based electronics–spintronics. The giant magnetoresistance of these materials, i.e. large resistivity changes in response to an applied magnetic field, make them suitable for applications in magnetic sensors, magnetic random access memories and spintronic devices. DMS materials have been experimentally synthesized [1, 2] and theoretically modeled [3–5] for several years, focused on the transition metal (TM) doped group III-V and II-VI semiconductors. In order to develop spintronics as a practical technology, it is necessary to explore materials whose magnetic property can be controlled by changing either the external field or the carrier concentration. From an industrial point of view, it is essential to fabricate ferromagnetic DMS with Curie temperature (Tc) above room temperature and a ferromagnetism that is intrinsic, rather than due to the presence of dopant clusters. The synthesis of DMS as nanowires is particularly attractive for the fabrication of low dimensional, spin-based electronic and optoelectronic devices. Transition metals with partially filled d states (Sc, Ti, V, Cr, Mn, Fe, Co, Ni, and Cu) have been used as magnetic atoms in DMS. The electron spin of the partially filled d states are responsible for their magnetic behavior.

The field of spintronics gained prominence when *Ga*1<sup>−</sup>*xMnxAs* was found to be ferromagnetic up to 110 K [1] and up to 173 K with a Mn concentration of 6.8% [6]. Ferromagnetism in this DMS was explained in terms of carrier-mediated exchange coupling between the localized spins of the transition metal ions. It is based on the theory that doping GaAs with Mn introduces shallow acceptors and hence holes into the valence band of GaAs. These holes are anti-ferromagnetically coupled to the local Mn moments. Dietl calculated the Curie temperature of various DMS doped with Mn [3] and found this temperature to be proportional to the concentration of impurities and to the square root of the p-type charge

the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Philipose and Sapkota; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Philipose and Sapkota, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

©2012 Philipose and Sapkota, licensee InTech. This is an open access chapter distributed under the terms of

carrier concentration. Ferromagnetism has been predicted and confirmed in several III-V semiconductors doped with transition metal ions such as Mn, Cr, V, and Co. However, III-V semiconductors require complex processing steps and are expensive to synthesize. Moreover, obtaining *Tc* higher than room temperature poses a major challenge for development of III-V-based DMS.

The quest for materials with a high magnetic transition temperature continues and in this regard, TM doped II-VI semiconductors like ZnO has emerged as an attractive candidate as it is abundant, non-toxic and inexpensive to synthesize. ZnO is also a transparent, semiconducting oxide exhibiting piezoelectric and possibly ferromagnetic behavior. Dietl predicted room temperature ferromagnetism for 5% Mn in p-type wide band gap semiconductors like ZnO and GaN [3]. Sharma and coworkers reported room temperature ferromagnetism in Mn-doped ZnO thin films [2]. The author reported the first observation of room temperature ferromagnetism in Mn doped ZnO nanowires [7], and since then there have been works presenting evidence for intrinsic ferromagnetism in individual ZnO nanoparticles doped with TM ions [8]. However, recent experiments reveal inconsistent ferromagnetic properties for TM doped ZnO [9–13], leaving the real mechanism for ferromagnetism in ZnO unclear. There are several controversial and conflicting reports on TM doped ZnO; including claims that the magnetic signature arises from clustering, segregated phases [13, 14], or defect mediated ferromagnetic coupling [15, 16]. The controversy in understanding the origin of ferromagnetism in TM doped II-VI semiconductors like ZnO is because in II-VI materials, dopants like Mn behave as an isovalent impurity. When *Mn*2<sup>+</sup> substitutes for *Zn*2<sup>+</sup> in the ZnO lattice, it provides no additional carriers. Therefore, charge carriers would need to be provided by additional doping in order to render Mn-doped ZnO ferromagnetic. Based on the predictions by Dietl et al. [3], room temperature ferromagnetism is expected in Mn-doped ZnO in the presence of a fairly high hole concentration of ≈ 1020*cm*<sup>−</sup>3. Such high hole concentrations seem rather unrealistic, especially, since it is difficult to achieve any reasonable concentration of holes in ZnO. The origin of experimentally observed ferromagnetism in TM doped II-VI semiconductors thus remains unclear. Kolesnik and coworkers [14] argue that the magnetic signature may arise from segregated phases of spinel oxides such as *Mn*3*O*<sup>4</sup> that is ferromagnetic. Theoretical calculations predict that *Zn*1<sup>−</sup>*xMnxO* nanowires favor the ferromagnetic state if Zn vacancies are introduced, while they are anti-ferromagnetic unless additional carriers are incorporated [17].

In this chapter, we will review experimental results of observed room temperature ferromagnetism in TM doped II-VI semiconductors. Studies drawn primarily from the author's research along with similar experimental findings will be presented. The choice of nanowires (as opposed to thin films) for studying the origin of ferromagnetism in II-VI semiconductors is justified because they allow: (a) clean and smooth surfaces with no precipitates or attachments on the nanowire surface, and (b) a unique geometry for realizing spintronic devices with possible one dimensional charge transport.

### **2. Theory of ferromagnetism in ZnO based DMS**

ZnO, a II-VI semiconductor, is a wide band gap (3.37 eV at room temperature) semi-conducting metal oxide with a large exciton binding energy of 60 meV, which makes it a promising material for optoelectronic applications. It has attracted increasing attention and stimulated intense research efforts since theoretical calculations predict Mn-doped ZnO can show ferromagnetism with a *Tc* value above room temperature making it a promising candidate for spintronic applications too. ZnO crystallizes in the wurtzite structure with tetrahedral *sp*<sup>3</sup> bonding. When doped with Mn to form *Zn*1−*xMnxO*, Mn atoms substitute the Zn atoms in the crystal lattice. The electronic configuration of Mn is 1*s*2, 2*s*2, 2*p*6, 3*s*2, 3*p*6, 3*d*5, 4*s*<sup>2</sup> and during substitution, Mn donates its 4*s*<sup>2</sup> electrons to the *sp*<sup>3</sup> bonding and exists as *Mn*2<sup>+</sup> ions. The half filled 3d band of *Mn*2<sup>+</sup> has 5 electrons and it hybridizes with the p-bands of oxygen in ZnO. This results in an exchange interaction between the localized magnetic moments (3d spins) and the spins of itinerant carriers in the host ZnO. Several mechanisms have been proposed to explain ferromagnetism in DMS. These include the Zener model [18], carrier mediated exchange [3], double exchange [5, 19] and coupling via a bound magnetic polaron [20, 21].

2 Nanowires - Recent Advances

III-V-based DMS.

[17].

carrier concentration. Ferromagnetism has been predicted and confirmed in several III-V semiconductors doped with transition metal ions such as Mn, Cr, V, and Co. However, III-V semiconductors require complex processing steps and are expensive to synthesize. Moreover, obtaining *Tc* higher than room temperature poses a major challenge for development of

The quest for materials with a high magnetic transition temperature continues and in this regard, TM doped II-VI semiconductors like ZnO has emerged as an attractive candidate as it is abundant, non-toxic and inexpensive to synthesize. ZnO is also a transparent, semiconducting oxide exhibiting piezoelectric and possibly ferromagnetic behavior. Dietl predicted room temperature ferromagnetism for 5% Mn in p-type wide band gap semiconductors like ZnO and GaN [3]. Sharma and coworkers reported room temperature ferromagnetism in Mn-doped ZnO thin films [2]. The author reported the first observation of room temperature ferromagnetism in Mn doped ZnO nanowires [7], and since then there have been works presenting evidence for intrinsic ferromagnetism in individual ZnO nanoparticles doped with TM ions [8]. However, recent experiments reveal inconsistent ferromagnetic properties for TM doped ZnO [9–13], leaving the real mechanism for ferromagnetism in ZnO unclear. There are several controversial and conflicting reports on TM doped ZnO; including claims that the magnetic signature arises from clustering, segregated phases [13, 14], or defect mediated ferromagnetic coupling [15, 16]. The controversy in understanding the origin of ferromagnetism in TM doped II-VI semiconductors like ZnO is because in II-VI materials, dopants like Mn behave as an isovalent impurity. When *Mn*2<sup>+</sup> substitutes for *Zn*2<sup>+</sup> in the ZnO lattice, it provides no additional carriers. Therefore, charge carriers would need to be provided by additional doping in order to render Mn-doped ZnO ferromagnetic. Based on the predictions by Dietl et al. [3], room temperature ferromagnetism is expected in Mn-doped ZnO in the presence of a fairly high hole concentration of ≈ 1020*cm*<sup>−</sup>3. Such high hole concentrations seem rather unrealistic, especially, since it is difficult to achieve any reasonable concentration of holes in ZnO. The origin of experimentally observed ferromagnetism in TM doped II-VI semiconductors thus remains unclear. Kolesnik and coworkers [14] argue that the magnetic signature may arise from segregated phases of spinel oxides such as *Mn*3*O*<sup>4</sup> that is ferromagnetic. Theoretical calculations predict that *Zn*1<sup>−</sup>*xMnxO* nanowires favor the ferromagnetic state if Zn vacancies are introduced, while they are anti-ferromagnetic unless additional carriers are incorporated

In this chapter, we will review experimental results of observed room temperature ferromagnetism in TM doped II-VI semiconductors. Studies drawn primarily from the author's research along with similar experimental findings will be presented. The choice of nanowires (as opposed to thin films) for studying the origin of ferromagnetism in II-VI semiconductors is justified because they allow: (a) clean and smooth surfaces with no precipitates or attachments on the nanowire surface, and (b) a unique geometry for realizing

ZnO, a II-VI semiconductor, is a wide band gap (3.37 eV at room temperature) semi-conducting metal oxide with a large exciton binding energy of 60 meV, which makes it a promising material for optoelectronic applications. It has attracted increasing attention

spintronic devices with possible one dimensional charge transport.

**2. Theory of ferromagnetism in ZnO based DMS**

According to the Zener model, since the Mn-d shell is only half filled, the direct interaction between d shells of adjacent Mn atoms leads to an antiferromagnetic state. However, the indirect coupling of spins through itinerant carriers tends to align the spins of the d shell electrons in a ferromagnetic manner. Ferromagnetism is therefore assumed to occur due to interactions between the local moments of the Mn atoms, mediated by free holes in the material. Based on this model, TM-doped p-type ZnO is predicted to be a promising candidates for ferromagnetic DMS with high Curie temperature. However, these predictions are based on the incorporation of ≈ 5% TM element into the host and hole concentrations of above 1020*cm*<sup>−</sup>3. Considering that p-type doping has always been a challenge in II-VI semiconductors, the achievement of such high hole concentrations is not realistic.

The Zener double exchange was used to explain ferromagnetism in TM doped InAs [19] and is an indirect coupling mechanism between two neighboring TM ions. According to this theory, if neighboring TM magnetic moments are in the same direction, there is a widening of the TM-d band due to hybridization of the spin states. The band energy of the ferromagnetic state can be lowered below that of the antiferromagnetic state if a sufficient number of holes (or electrons) exist. The 3d electrons of the TM ions hop to the 3d orbitals of neighboring TM ions with parallel magnetic moments, thus lowering its energy by hopping in the ferromagnetic state.

In all of the proposed mechanisms, the impact of charge state and the involved electronic 3d-shell configuration of TM ions are crucial points in understanding the magnetic mechanism within the diluted magnetic semiconductors. Experimental investigations have shown that Mn-substituted ZnO samples exhibit distinct magnetic properties including spin-glass behavior [22]; paramagnetism [23]; ferromagnetism with *Tc* of 45 K [24]; ferromagnetism with *Tc* of 425 K in thin films of Mn doped ZnO [2] and *Tc* > 400 K for Mn doped ZnO nanowires [7]. Structural defects have also been reported to play an important role in the occurrence and stability of ferromagnetism [25]. To interpret the origin of the magnetic properties, many magnetic interaction mechanisms have been proposed, including direct super-exchange, indirect superexchange, and carrier-mediated exchange. In spite of all these works, the local structure and microscopic origins of the high-*Tc* ferromagnetism in ZnO, GaN, and Si-based DMSs are still under debate. This debate has emerged as one of the most important challenges in understanding the science of DMS.

## **3. Ferromagnetic II-VI nanowires**

The attractive features in using II-VI semiconductors for fabricating DMS are several: interesting opto-electronic properties, high solubility of transition metal atoms in the host lattice (implies that the amount of injected spins can be large), and also because II-VI materials like ZnO are cheap, abundant in nature and are also environmentally friendly. The ternary nature of the *AI I* <sup>1</sup>−*xMnxBV I* alloys, allows the possibility of 'tuning' the lattice constant and band parameters by varying the composition of the material. The substitutional Mn atoms in the *AI IBV I* lattice are also characterized by highly efficient electroluminescence, which makes dilute *AI I* <sup>1</sup>−*xMnxBV I* alloys important in optical flat panel display applications. However, since Mn atoms do not provide any carriers when they are substituted in the insulating host *AI I* <sup>1</sup>−*xMnxBV I* lattice, it was believed for a long time that the magnetic interaction in II-VI DMS is dominated by antiferromagnetic exchange among the Mn spins, which would result in paramagnetic or antiferromagnetic behavior of the material. The prediction of room temperature ferromagnetism in TM doped II-VI semiconductors initiated intense theoretical and experimental work on II-VI materials, with several contradicting results. Of the II-VI semiconductors, ZnO is particularly interesting because it has a smaller lattice constant compared to other IIUVI semiconductors, which will serve to increase the ˝ magnetic coupling strength among the magnetic ions.

## **3.1. Synthesis and structural characterization of** *Zn*1<sup>−</sup>*xMnxO* **nanowires**

The *Zn*1<sup>−</sup>*xMnxO* nanowires were grown on Si substrates by the vapor phase growth, using the VLS mechanism [7]. The growth temperature was 700◦*C*. Mn doping was achieved by using *MnO*<sup>2</sup> as the dopant material, which was evaporated at 700◦*C*. A mixture of ZnO and graphite was used as the source material, evaporated at 900◦*C*. After growth for an hour, the Si surface was found coated with a grayish colored substance. The morphologies of this product was investigated by a scanning electron microscope (SEM) using Hitachi SEM S-5200, and a high resolution transmission electron microscope (HRTEM) using Hitachi STEM HD-2000. Magnetic properties of these samples were studied by superconducting quantum interference device (SQUID) magnetometer (MPMS MultiVu Application). The Mn content was determined by Energy Dispersive X-ray (EDX) spectroscopy and Ion Coupled Plasma Mass Spectrometry (ICPMS) using Perkin Elmer Sciex ELAN DRC II. The morphology of the *Zn*1<sup>−</sup>*xMnxO* nanowires was observed using SEM, which revealed that the nanowires have lengths in the range of 8–10 *µm* and diameter between 40–150 nm. The crystallinity of the *Zn*1<sup>−</sup>*xMnxO* nanowires as evidenced by HRTEM is shown in Figure 2. The image clearly reveals the lattice fringes of ZnO {0001} planes with an interplanar spacing of about 0.52 nm, indicating that the nanowire is a single crystal and grows along the [001] direction, as shown in Figure 2. No observable change in structure was noted between pure and Mn doped nanowires. All the wires were straight and had a smooth surface, which also confirms that Mn is doped into the lattice structure instead of being precipitated as a second phase.

The composition of the nanowires was characterized by EDX at nanometer resolution. All samples show uniform distribution of Zn, O and no other impurities other than Mn (Figure 3. The signal from Si is due to the substrate. Due to the limited sensitivity of EDX at high resolution, Mn atoms were detected only in samples with more than 2% Mn concentration although ferromagnetism was observed in all Mn-doped samples as discussed below. The elemental mapping (EDX) spectra show a highly uniform distribution of Mn atoms with

**Figure 1.** SEM image of *Zn*1<sup>−</sup>*xMnxO* nanowires growing in a 'weed-like' manner on Si substrate.

4 Nanowires - Recent Advances

**3. Ferromagnetic II-VI nanowires**

magnetic coupling strength among the magnetic ions.

The ternary nature of the *AI I*

which makes dilute *AI I*

insulating host *AI I*

The attractive features in using II-VI semiconductors for fabricating DMS are several: interesting opto-electronic properties, high solubility of transition metal atoms in the host lattice (implies that the amount of injected spins can be large), and also because II-VI materials like ZnO are cheap, abundant in nature and are also environmentally friendly.

constant and band parameters by varying the composition of the material. The substitutional Mn atoms in the *AI IBV I* lattice are also characterized by highly efficient electroluminescence,

However, since Mn atoms do not provide any carriers when they are substituted in the

interaction in II-VI DMS is dominated by antiferromagnetic exchange among the Mn spins, which would result in paramagnetic or antiferromagnetic behavior of the material. The prediction of room temperature ferromagnetism in TM doped II-VI semiconductors initiated intense theoretical and experimental work on II-VI materials, with several contradicting results. Of the II-VI semiconductors, ZnO is particularly interesting because it has a smaller lattice constant compared to other IIUVI semiconductors, which will serve to increase the ˝

The *Zn*1<sup>−</sup>*xMnxO* nanowires were grown on Si substrates by the vapor phase growth, using the VLS mechanism [7]. The growth temperature was 700◦*C*. Mn doping was achieved by using *MnO*<sup>2</sup> as the dopant material, which was evaporated at 700◦*C*. A mixture of ZnO and graphite was used as the source material, evaporated at 900◦*C*. After growth for an hour, the Si surface was found coated with a grayish colored substance. The morphologies of this product was investigated by a scanning electron microscope (SEM) using Hitachi SEM S-5200, and a high resolution transmission electron microscope (HRTEM) using Hitachi STEM HD-2000. Magnetic properties of these samples were studied by superconducting quantum interference device (SQUID) magnetometer (MPMS MultiVu Application). The Mn content was determined by Energy Dispersive X-ray (EDX) spectroscopy and Ion Coupled Plasma Mass Spectrometry (ICPMS) using Perkin Elmer Sciex ELAN DRC II. The morphology of the *Zn*1<sup>−</sup>*xMnxO* nanowires was observed using SEM, which revealed that the nanowires have lengths in the range of 8–10 *µm* and diameter between 40–150 nm. The crystallinity of the *Zn*1<sup>−</sup>*xMnxO* nanowires as evidenced by HRTEM is shown in Figure 2. The image clearly reveals the lattice fringes of ZnO {0001} planes with an interplanar spacing of about 0.52 nm, indicating that the nanowire is a single crystal and grows along the [001] direction, as shown in Figure 2. No observable change in structure was noted between pure and Mn doped nanowires. All the wires were straight and had a smooth surface, which also confirms that Mn is doped into the lattice structure instead of being precipitated as a second phase. The composition of the nanowires was characterized by EDX at nanometer resolution. All samples show uniform distribution of Zn, O and no other impurities other than Mn (Figure 3. The signal from Si is due to the substrate. Due to the limited sensitivity of EDX at high resolution, Mn atoms were detected only in samples with more than 2% Mn concentration although ferromagnetism was observed in all Mn-doped samples as discussed below. The elemental mapping (EDX) spectra show a highly uniform distribution of Mn atoms with

**3.1. Synthesis and structural characterization of** *Zn*1<sup>−</sup>*xMnxO* **nanowires**

<sup>1</sup>−*xMnxBV I* alloys, allows the possibility of 'tuning' the lattice

<sup>1</sup>−*xMnxBV I* alloys important in optical flat panel display applications.

<sup>1</sup>−*xMnxBV I* lattice, it was believed for a long time that the magnetic

**Figure 2.** HRTEM image of a crystalline *Zn*1<sup>−</sup>*xMnxO* nanowire. The nanowire has smooth surface and there is no evidence of any secondary phases or defects on these nanowires.

**Figure 3.** EDX spectra for *Zn*1<sup>−</sup>*xMnxO* nanowires, with ∼ 2 *at*% Mn.

no clustering or segregated Mn-rich phases. As the growth method does not allow us to precisely control the Mn concentration and the sensitivity of EDX is limited, ICPMS (a destructive chemical analysis method) was done to estimate the Mn concentration. The nanowires were first dissolved in *HNO*<sup>3</sup> and subsequently introduced into an extremely hot (7,000 to 10,000 K) radio frequency induced argon plasma. The plume of the argon plasma is sampled directly into the entrance orifice of a quadrupole mass spectrometer. Analyte ions produced by the plasma are accelerated, magnetically separated and counted using an electron multiplier. In all samples the Mn concentration was measured to be in the range of 1% to 4%.

#### *3.1.1. SQUID measurements of magnetization on Zn*1−*xMnxO nanowires*

Magnetization measurements, using a SQUID magnetometer (MPMS MultiVu Application) showed a distinct hysteresis loop superimposed on a diamagnetic contribution from the substrate (Figure 4).

Figure 5 shows the variation of magnetization (M) with temperature T, measured during cooling in an applied magnetic field of 100 Oe. The magnetization decays very slowly with T over the range studied. *Tc* is well above room temperature, but it is hard to determine the exact value, since the value of *Tc* is rather high, exceeding the range of measurements. The same conclusion can be reached from the dependence of remnant magnetization (*Mr*) and coercive field (*Hc*) on temperature (Figure 5). *Mr* and *Hc* are 2.44 <sup>×</sup> <sup>10</sup>−<sup>6</sup> e.m.u and 37.05 Oe respectively at 300 K.

### *3.1.2. Estimation of magnetic moment per Mn atom in Zn*1−*xMnxO nanowires*

The *Zn*1−*xMnxO* nanowires were grown on Si substrates of area 9 *mm*2. The nanowires had an average length of 10 *µm* and hence the maximum volume of nanowires is estimated to be 9 × <sup>10</sup>−<sup>5</sup> *cm*3. For Mn concentration varying from 1 *at*% to 4 *at*%, the number of Mn atoms (*NMn*) is estimated to vary from 2 ± 0.8 × 1015 *atoms*/*cm*<sup>3</sup> to 8 ± 3.2 × 1015 *atoms*/*cm*3. The magnetic moment per Mn atom (in terms of the Bohr magneton) was determined by the equation:

6 Nanowires - Recent Advances

1% to 4%.

substrate (Figure 4).

respectively at 300 K.

equation:

**Figure 3.** EDX spectra for *Zn*1<sup>−</sup>*xMnxO* nanowires, with ∼ 2 *at*% Mn.

*3.1.1. SQUID measurements of magnetization on Zn*1−*xMnxO nanowires*

*3.1.2. Estimation of magnetic moment per Mn atom in Zn*1−*xMnxO nanowires*

no clustering or segregated Mn-rich phases. As the growth method does not allow us to precisely control the Mn concentration and the sensitivity of EDX is limited, ICPMS (a destructive chemical analysis method) was done to estimate the Mn concentration. The nanowires were first dissolved in *HNO*<sup>3</sup> and subsequently introduced into an extremely hot (7,000 to 10,000 K) radio frequency induced argon plasma. The plume of the argon plasma is sampled directly into the entrance orifice of a quadrupole mass spectrometer. Analyte ions produced by the plasma are accelerated, magnetically separated and counted using an electron multiplier. In all samples the Mn concentration was measured to be in the range of

Magnetization measurements, using a SQUID magnetometer (MPMS MultiVu Application) showed a distinct hysteresis loop superimposed on a diamagnetic contribution from the

Figure 5 shows the variation of magnetization (M) with temperature T, measured during cooling in an applied magnetic field of 100 Oe. The magnetization decays very slowly with T over the range studied. *Tc* is well above room temperature, but it is hard to determine the exact value, since the value of *Tc* is rather high, exceeding the range of measurements. The same conclusion can be reached from the dependence of remnant magnetization (*Mr*) and coercive field (*Hc*) on temperature (Figure 5). *Mr* and *Hc* are 2.44 <sup>×</sup> <sup>10</sup>−<sup>6</sup> e.m.u and 37.05 Oe

The *Zn*1−*xMnxO* nanowires were grown on Si substrates of area 9 *mm*2. The nanowires had an average length of 10 *µm* and hence the maximum volume of nanowires is estimated to be 9 × <sup>10</sup>−<sup>5</sup> *cm*3. For Mn concentration varying from 1 *at*% to 4 *at*%, the number of Mn atoms (*NMn*) is estimated to vary from 2 ± 0.8 × 1015 *atoms*/*cm*<sup>3</sup> to 8 ± 3.2 × 1015 *atoms*/*cm*3. The magnetic moment per Mn atom (in terms of the Bohr magneton) was determined by the

**Figure 4.** Hysteresis loop (M-H curve) of *Zn*1<sup>−</sup>*xMnxO* nanowires at a temperature of 300 K; inset: the as-obtained data from the SQUID measurements including the diamagnetic contribution arising from the Si substrate.

$$\frac{m}{\mu\_{\text{B}}} = \frac{M\_{\text{s}}}{N\_{\text{Mn}}(9.27 \times 10^{-24})} \tag{1}$$

where *Ms* is the saturation magnetization. From the observed saturation magnetization (Figure 4), *Ms* is of the order of 2.0 <sup>−</sup> 3.0 <sup>×</sup> <sup>10</sup>−<sup>5</sup> e.m.u (2.0 <sup>−</sup> 3.0 <sup>×</sup> <sup>10</sup>−<sup>8</sup> *<sup>A</sup>*.*m*2), and the magnetic moment per Mn atom is estimated to be in the range of 0.30 *µ<sup>B</sup>* to 1.2 *µB*, which is smaller than the maximum spin moment (5 *<sup>µ</sup>B*) per *Mn*2<sup>+</sup> with S = 5/2 and g=2; but is larger than those reported in previous reports [2].

#### *3.1.3. Optical characterization of Mn doped and undoped ZnO nanowires*

The luminescence spectra of ZnO exhibits strong mid-gap green emission, which has been occasionally attributed to Zn interstitials [26] and Zn vacancies [27], the most widely accepted origin of this emission is oxygen vacancies [28, 29]. The *Zn*1−*xMnxO* and undoped ZnO nanowires were characterized by room temperature photoluminescence (PL) measurements.

The wires were excited by a 355 nm wavelength laser at room temperature. The photoluminescence spectrum shown in Figure 6 shows a very weak band edge luminescence from exciton (UV emission) at about 380 nm and a strong mid-gap emission (green emission) at about 525 nm. The ZnO band edge emission is extremely weak possibly because of strong

**Figure 5.** Dependence of remnant magnetization (*Mr*), coercive field (*Hc* ) and field cooled magnetization (M-T at 100 Oe) of *Zn*1<sup>−</sup>*xMnxO* nanowires on temperature.

non-radiative processes due to the large density of surface states in a nanowire. When doped with Mn, a strong suppression of the mid-gap emission is observed followed by a substantial increase in the band-edge luminescence. Although most Mn-doped II-VI materials show a manganese specific orange emission no such luminescence was observed in Mn doped ZnO nanowires. This is consistent with previous reports on Mn-doped ZnO [30, 31], but the reason for this absence of Mn-related emission is not clear. Although, the strong suppression of the oxygen vacancy related emission could be due to Forester type dipole-dipole energy transfer, the small oscillator strength of Mn absorption would make such a process very inefficient. The suppression of mid-gap emission in Mn-doped II-VI materials implies a more efficient direct capture of photo excited carriers before getting trapped by states responsible for mid-gap emission. However, such a process would lead to suppression of band-edge luminescence as well, contrary to the observation of enhanced band-edge emission. These results therefore indicate that there is an interaction between Mn doping and native defects like oxygen vacancies. The mechanism of such an interaction is not clear and deserves further study.

#### *3.1.4. Discussion*

Unlike previous reports, the observed ferromagnetism can be attributed only to the Mn doped into the ZnO lattice structure since HRTEM revealed a clean and smooth surface of the nanowire and no precipitates or attachments to the wire surface were observed. No ferromagnetic behaviour was observed for pure ZnO nanowires. Moreover, ferromagnetic impurity phases responsible for the observed magnetic ordering can be ruled out since there

**Figure 6.** Room temperature photoluminescence spectra of undoped and *Zn*1<sup>−</sup>*xMnxO* nanowires.

8 Nanowires - Recent Advances

*Zn*1<sup>−</sup>*xMnxO* nanowires on temperature.

study.

*3.1.4. Discussion*

**Figure 5.** Dependence of remnant magnetization (*Mr*), coercive field (*Hc* ) and field cooled magnetization (M-T at 100 Oe) of

non-radiative processes due to the large density of surface states in a nanowire. When doped with Mn, a strong suppression of the mid-gap emission is observed followed by a substantial increase in the band-edge luminescence. Although most Mn-doped II-VI materials show a manganese specific orange emission no such luminescence was observed in Mn doped ZnO nanowires. This is consistent with previous reports on Mn-doped ZnO [30, 31], but the reason for this absence of Mn-related emission is not clear. Although, the strong suppression of the oxygen vacancy related emission could be due to Forester type dipole-dipole energy transfer, the small oscillator strength of Mn absorption would make such a process very inefficient. The suppression of mid-gap emission in Mn-doped II-VI materials implies a more efficient direct capture of photo excited carriers before getting trapped by states responsible for mid-gap emission. However, such a process would lead to suppression of band-edge luminescence as well, contrary to the observation of enhanced band-edge emission. These results therefore indicate that there is an interaction between Mn doping and native defects like oxygen vacancies. The mechanism of such an interaction is not clear and deserves further

Unlike previous reports, the observed ferromagnetism can be attributed only to the Mn doped into the ZnO lattice structure since HRTEM revealed a clean and smooth surface of the nanowire and no precipitates or attachments to the wire surface were observed. No ferromagnetic behaviour was observed for pure ZnO nanowires. Moreover, ferromagnetic impurity phases responsible for the observed magnetic ordering can be ruled out since there is no known phase of manganese oxide with above room temperature *Tc*. All the possible *MnOx* phases that could be forming in our growth process are antiferromagnetic, with the exception of *Mn*3*O*4, which is ferromagnetic below 45 K. In contrast to the bulk and thin film samples studied in the past we have single crystal nanowires and no evidence for multiple phases. The nanowires studied here are not intentionally doped. In general, undoped ZnO tends to be n-type due to intrinsic defects. Although the large density of surface states and the VLS mechanism of nanowire growth can potentially make them p-type, we do not have any evidence for substantial hole concentration in these samples. In order to estimate the conductivity of the nanowire array, a sample for conductivity measurements was prepared following the procedure described in Reference [32]. Conductivity measurements on ZnO nanowire arrays show that these wires are conducting but the observed resistance is more than 1 GΩ per wire implying very low mobile carrier concentration. However, carrier mediated exchange interaction only requires itinerant carriers that hybridize with the Mn 3-d band so that hopping of localized spins between Mn ions can stabilize the parallel spin configuration. The material may still remain semi-insulating or show hopping conductivity [33]. Further, conduction through nanowires are known to be very sensitive to charge inhomogeneities along the length of the wire [32]. Even though ferromagnetism is predicted and expected in p-type Mn doped ZnO through strong p-d hybridization [3], Sato et al. [5] predicted that the ferromagnetic state of *Co*2+(d7) in Co-doped ZnO can be stabilized by *s* − *d* hybridization, thus making it possible to obtain high Curie temperature ferromagnetic materials in n-type ZnO.

## **3.2. Synthesis and characterization of** *Zn*1<sup>−</sup>*xFexO* **nanotubes**

Sato et al.,[34] used first principles calculations to predict that the ferromagnetic state is stable in V-, Cr-, Fe-, Co- or Ni-doped ZnO without any additional carrier doping treatments. It is believed that (Zn,Fe)O is a promising candidates for high *Tc* ferromagnets, and there are reports of ferromagnetism in Fe-doped ZnO nanocrystals [35], and with co-doping [36, 37]. In this section, a recent work in the author's research group[38] on the synthesis of *Zn*1<sup>−</sup>*xFexO* nanotubes with clear evidence of Fe ions substituting for *Zn*2<sup>+</sup> ions in the ZnO crystal lattice is presented. The Fe-doped ZnO nanotubes were synthesized by a simple electrochemical process at 75◦*<sup>C</sup>* using *ZnCl*<sup>2</sup> and *KCl* mixed with Fe as an electrolyte. The undoped ZnO compound has near-zero or negligible magnetic moment, since all its electrons are paired. However, when ZnO is doped with Fe (which has an incomplete d subshell), it exhibits a magnetic moment that is determined by the number of unpaired electrons in the doped compound. The determination of the magnetic moment was done using a technique similar to the Gouy method [39]. In this technique, an Evans balance was employed to measure the change in current required to balance suspended magnets when their magnetic field interacts with a magnetic sample. This value was then used to determine the effective magnetic moment. Magnetization measurements on the *Zn*1<sup>−</sup>*xFexO* nanotubes have not been completed and so no claims can be made on the ferromagnetic property of these nanostructures.

Synthesis of ZnO nanotubes was done in a three electrode standard electrochemical cell with a Si(111) substrate coated with 300 nm Au film as the cathode and the working electrode. The counter electrode was made of Pt. The electrolyte comprised of a mixture of *FeCl*<sup>3</sup> (0.1 mM) + *ZnCl*<sup>2</sup> (6 mM) + *KCl* (0.2 M)]. Synthesis was carried out in a temperature-controlled water bath maintained at 75◦*<sup>C</sup>* . Oxygen (20 sccm) was continuously bubbled through the electrolyte during the growth process. A constant bias of -1 V was maintained between the working and reference electrodes during the growth process. Post growth, the nanostructures were characterized by SEM (NOVA dual-beam SEM/FIB, FEI Nova 200 Nano-Lab), HRTEM (FEI Tecnai G2 F20 S-Twin 200 keV), and selected-area electron microscopy (SAED) pattern analysis. Composition and structural analysis were further done by EDX and x-ray diffraction (XRD) (Rigaku Ultima III). The hexagonal ZnO nanotubes had wall thickness of 20 nm and pore diameter in the range of 80 nm to 120 nm, as shown in Figure 7(a) and (b). EDX analysis of the nanotubes showed a Fe peak, confirming a dilute concentration of Fe (< 4*wt*.%) in the nanotubes. Electrolyte concentration, growth time, and growth temperature play a vital role in determining the morphology of the nanostructures. The first stage of the synthesis process was the growth of nanorods, which was followed by etching of the nanorods along a specific plane. The etching process was facilitated by the high concentration of *KCl* in the electrolyte.

### *3.2.1. XRD measurements on Zn*1<sup>−</sup>*xFexO nanotubes*

Figure 8 shows results of XRD measurements on undoped and Fe doped ZnO nanotubes. The diffraction peaks have been indexed to the hexagonal ZnO wurtzite structure. The strong peak at 2*<sup>θ</sup>* <sup>=</sup> 34.36*<sup>o</sup>* corresponds to the (0002) c-axis of ZnO, which indicates that the nanotubes preferentially grow along (0001) direction. No additional peak attributed to Fe or its oxide could be detected in the XRD spectrum of Fe-doped ZnO nanotubes, though a small shift of the (0002) peak was observed. This shift of the (0002) peak position for

**Figure 7.** SEM image of nanotubes: (a) *Zn*1<sup>−</sup>*xFexO* nanotubes with wall thickness of about 20 nm and pore diameter of about 80 nm to 120 nm. Inset shows a magnified SEM image of a single Fe-doped ZnO nanotube. (b) EDX spectrum confirming the presence of Fe, Zn, and O. The Fe concentration is estimated to be about 3-4 wt.%.

the Fe-doped ZnO indicates that there is an increase in the c-axis lattice parameter for the Fe-doped sample, indicative of the fact that Fe ions substitute for *Zn*2<sup>+</sup> ions in the ZnO lattice, without changing its wurtzite crystal structure. The ionic radius of *Fe*2<sup>+</sup> is larger than that of *Zn*2<sup>+</sup> by about 5%. Hence, when *Fe*2<sup>+</sup> ions occupy the *Zn*2<sup>+</sup> sites of ZnO, the in-plane atomic arrangement with closed atomic packing will be strained due to the substitution of Zn by the larger Fe atom. This cause an increase in the c-lattice constant and a shift of the (0002) peak to smaller angle in the XRD pattern of the Fe-doped ZnO nanotubes. It is also possible that *Fe*2<sup>+</sup> and *Fe*3<sup>+</sup> ions are both present and occur as substitutional as well as interstitial dopants. In this case, the atomic arrangement is disturbed and the number of interstitials will increase, which will result in lattice disorder and strain. A comparison of the XRD spectrum of the undoped and doped nanotubes (inset of Figure 8) shows that the full-width at half maximum (FWHM) of the (002) diffraction peak is broadened (about 3% to 4%) and its intensity decreased in the Fe-doped ZnO nanotubes. This indicates that the dilute (<4 wt.%) Fe doping influences the crystalline quality of the nanotubes by causing lattice defects. This result was further validated by comparison of the PL spectra before and after doping.

### *3.2.2. Optical characterization of Zn*1−*xFexO nanotubes*

10 Nanowires - Recent Advances

nanostructures.

electrolyte.

*3.2.1. XRD measurements on Zn*1<sup>−</sup>*xFexO nanotubes*

**3.2. Synthesis and characterization of** *Zn*1<sup>−</sup>*xFexO* **nanotubes**

Sato et al.,[34] used first principles calculations to predict that the ferromagnetic state is stable in V-, Cr-, Fe-, Co- or Ni-doped ZnO without any additional carrier doping treatments. It is believed that (Zn,Fe)O is a promising candidates for high *Tc* ferromagnets, and there are reports of ferromagnetism in Fe-doped ZnO nanocrystals [35], and with co-doping [36, 37]. In this section, a recent work in the author's research group[38] on the synthesis of *Zn*1<sup>−</sup>*xFexO* nanotubes with clear evidence of Fe ions substituting for *Zn*2<sup>+</sup> ions in the ZnO crystal lattice is presented. The Fe-doped ZnO nanotubes were synthesized by a simple electrochemical process at 75◦*<sup>C</sup>* using *ZnCl*<sup>2</sup> and *KCl* mixed with Fe as an electrolyte. The undoped ZnO compound has near-zero or negligible magnetic moment, since all its electrons are paired. However, when ZnO is doped with Fe (which has an incomplete d subshell), it exhibits a magnetic moment that is determined by the number of unpaired electrons in the doped compound. The determination of the magnetic moment was done using a technique similar to the Gouy method [39]. In this technique, an Evans balance was employed to measure the change in current required to balance suspended magnets when their magnetic field interacts with a magnetic sample. This value was then used to determine the effective magnetic moment. Magnetization measurements on the *Zn*1<sup>−</sup>*xFexO* nanotubes have not been completed and so no claims can be made on the ferromagnetic property of these

Synthesis of ZnO nanotubes was done in a three electrode standard electrochemical cell with a Si(111) substrate coated with 300 nm Au film as the cathode and the working electrode. The counter electrode was made of Pt. The electrolyte comprised of a mixture of *FeCl*<sup>3</sup> (0.1 mM) + *ZnCl*<sup>2</sup> (6 mM) + *KCl* (0.2 M)]. Synthesis was carried out in a temperature-controlled water bath maintained at 75◦*<sup>C</sup>* . Oxygen (20 sccm) was continuously bubbled through the electrolyte during the growth process. A constant bias of -1 V was maintained between the working and reference electrodes during the growth process. Post growth, the nanostructures were characterized by SEM (NOVA dual-beam SEM/FIB, FEI Nova 200 Nano-Lab), HRTEM (FEI Tecnai G2 F20 S-Twin 200 keV), and selected-area electron microscopy (SAED) pattern analysis. Composition and structural analysis were further done by EDX and x-ray diffraction (XRD) (Rigaku Ultima III). The hexagonal ZnO nanotubes had wall thickness of 20 nm and pore diameter in the range of 80 nm to 120 nm, as shown in Figure 7(a) and (b). EDX analysis of the nanotubes showed a Fe peak, confirming a dilute concentration of Fe (< 4*wt*.%) in the nanotubes. Electrolyte concentration, growth time, and growth temperature play a vital role in determining the morphology of the nanostructures. The first stage of the synthesis process was the growth of nanorods, which was followed by etching of the nanorods along a specific plane. The etching process was facilitated by the high concentration of *KCl* in the

Figure 8 shows results of XRD measurements on undoped and Fe doped ZnO nanotubes. The diffraction peaks have been indexed to the hexagonal ZnO wurtzite structure. The strong peak at 2*<sup>θ</sup>* <sup>=</sup> 34.36*<sup>o</sup>* corresponds to the (0002) c-axis of ZnO, which indicates that the nanotubes preferentially grow along (0001) direction. No additional peak attributed to Fe or its oxide could be detected in the XRD spectrum of Fe-doped ZnO nanotubes, though a small shift of the (0002) peak was observed. This shift of the (0002) peak position for The nanostructures were optically characterized by photoluminescence (PL) spectroscopy and Raman spectroscopy. PL spectra were obtained using a TRIAX 320 spectrometer with excitation provided by a Kimmon continuous-wave HeCd laser with spot size of about 1 *mm*2. The Raman spectrum was obtained using an Almega XR 532 nm (green) laser.

Figure 9 shows the room-temperature PL spectra for the undoped and Fe-doped ZnO nanotubes. The curves with solid lines and symbols represent the experimentally obtained data for the undoped and Fe-doped ZnO nanotubes.

**Figure 8.** XRD spectrum of undoped (solid line) and Fe-doped (dotted line) ZnO nanotubes. Inset shows the broadening of (0002) peak for Fe-doped nanotubes.

The broad defect peak in the Fe-doped sample was de-convoluted using a Gaussian distribution (shown by dashed curves). The PL spectra for undoped ZnO tubes showed strong band-edge emission at 370.5 nm (3.35 eV) and no other peaks, attesting to the high crystalline quality of the undoped nanotubes. However, when doped with Fe, the PL spectrum showed an ultraviolet (UV) emission peak at ≈ 381*nm* (red-shifted) and a

**Figure 9.** Room-temperature PL spectra of undoped and Fe-doped ZnO nanotubes. The normalized curves with solid lines and symbols represent the experimentally obtained data. The curves shown by dotted lines represent the Gaussian fit of PL spectra for Fe-doped nanotubes.

broad emission peak in the visible region. The red-shift of the band edge in Fe-doped II-VI semiconductors has been attributed to the *sp* − *d* spin exchange interactions between the band electrons of the host and localized d electrons of the Fe ion substituting the group II ion [40]. A *Fe*2<sup>+</sup> cation is in 3*d*<sup>6</sup> configuration. The sUp and p ˝ Ud exchange interactions between ZnO ˝ as host and Fe as dopant give rise to a negative and positive correction to the conductionand valence-band edges, leading to narrowing of the band gap. Thus, the red-shift and quenching of the UV emission peak in the doped sample provide further evidence of the fact that Fe substitutes for Zn in the ZnO crystal [41]. The broad defect-related emission peak was deconvoluted into Gaussian peaks at 474 nm, 489 nm, 499 nm, 551 nm, and 563 nm. The peak observed at 474 nm is attributed to interstitial Zn and oxygen. The peaks at 489 nm and 499 nm are attributed to singly ionized oxygen vacancies, whereas the peaks at 551 nm and 563 nm are attributed to oxygen interstitials.

12 Nanowires - Recent Advances

(0002) peak for Fe-doped nanotubes.

for Fe-doped nanotubes.

**Figure 8.** XRD spectrum of undoped (solid line) and Fe-doped (dotted line) ZnO nanotubes. Inset shows the broadening of

The broad defect peak in the Fe-doped sample was de-convoluted using a Gaussian distribution (shown by dashed curves). The PL spectra for undoped ZnO tubes showed strong band-edge emission at 370.5 nm (3.35 eV) and no other peaks, attesting to the high crystalline quality of the undoped nanotubes. However, when doped with Fe, the PL spectrum showed an ultraviolet (UV) emission peak at ≈ 381*nm* (red-shifted) and a

**Figure 9.** Room-temperature PL spectra of undoped and Fe-doped ZnO nanotubes. The normalized curves with solid lines and symbols represent the experimentally obtained data. The curves shown by dotted lines represent the Gaussian fit of PL spectra Figure 10(a) and (b) compare the Raman spectrum for the undoped and Fe-doped ZnO nanotubes. Both samples show a Raman shift at 521 *cm*−<sup>1</sup> corresponding to the Si substrate used for the growth of ZnO nanotubes. The undoped sample shows a characteristic Raman peak at 438 *cm*<sup>−</sup>1, related to the hexagonal wurtzite phase of ZnO. The Fe-doped sample shows additional peaks at 336 *cm*<sup>−</sup>1, 384 *cm*<sup>−</sup>1, and 644 *cm*<sup>−</sup>1.

**Figure 10.** Raman scattering spectrum for: (a) undoped ZnO nanotubes and (b) Fe-doped ZnO nanotubes

The peaks at 336 *cm*−<sup>1</sup> and 384 *cm*−<sup>1</sup> are attributed to intrinsic defects in ZnO such as oxygen vacancies and zinc interstitials. The additional vibrational mode at 644 *cm*−<sup>1</sup> in the Fe-doped ZnO nanotubes is also related to intrinsic ZnO lattice defects, which either become activated as vibrating complexes upon addition of Fe, or their concentration increases upon Fe incorporation into the ZnO lattice.

### *3.2.3. Determination of effective magnetic moment in Zn*1−*xFexO nanotubes*

Further evidence of the successful incorporation of Fe into ZnO was obtained through determination of the effective magnetic moment (*µeff*) for the undoped and Fe-doped ZnO nanotubes. This was done using a magnetic susceptibility balance (Evans balance), which measures the force experienced by a substance when it is placed in a magnetic field. The nanotubes were placed in a cylindrical tube, which was placed in a uniform magnetic field. The effective magnetic moment was measured in units of Bohr magnetron (*µB*)using the equation:

$$
\mu\_{eff} = 2.828(\chi\_A T)^{1/2} \tag{2}
$$

where *χA* is the corrected molar susceptibility of the sample at temperature T (room temperature). *χ<sup>A</sup>* for *Zn*1<sup>−</sup>*xFexO* and undoped ZnO nanotubes was determined to be ≈ 1.31 × <sup>10</sup>−<sup>2</sup> cgs units and ≈ 1.68 × <sup>10</sup><sup>−</sup>4cgs units respectively. The effective magnetic moment was then calculated for both samples and found to be ≈ 5.41*µ<sup>B</sup>* for Fe-doped and ≈ 0.62*µ<sup>B</sup>* for undoped ZnO nanotubes. These values of *µeff* correspond to four unpaired electrons in the *Zn*1<sup>−</sup>*xFexO* nanotubes and zero unpaired electrons in the undoped ZnO nanotubes. The four unpaired electrons found in the Fe-doped material is consistent with the electron configuration of *Fe*2<sup>+</sup> in a high spin tetrahedral geometry, and is expected from the incorporation of *Fe*2<sup>+</sup> as a substitutional dopant in ZnO. The undoped ZnO nanotubes showed a near zero effective magnetic moment.

#### **4. Conclusion**

There are several works in favor of and against ferromagnetism in TM doped ZnO. Experimentalists have frequently observed ferromagnetism in this material though the theory to support this claim does not exist. The promise of obtaining materials with high Curie temperature and thus develop spin-based semiconductor devices keeps the field of II-VI DMS very active. The electronic properties of nanoscale DMS are strongly affected by quantum confinement effect and large surface to volume ratios; hence the magnetic coupling between TM ions in low-dimensional systems may be distinctively different from that of bulk materials. Ferromagnetism in Mn-doped ZnO nanowires is yet to be understood, as Mn in its +2 valence state is an isovalent dopant and does not provide any carrier to ZnO. Hence, by first-principles studies, it must lead to an antiferromagnetic ground state. However, experimental observation of room temperature ferromagnetism in Mn-doped ZnO thin films and nanostructures have posed a challenge to the theoretical understanding of ferromagnetism for such systems. To complicate matters further, recent experiments reveal inconsistent ferromagnetic properties for TM doped ZnO, leaving the real mechanism for ferromagnetism in ZnO unclear. Experimental studies have revealed the importance of defects such as zinc and oxygen interstitials and vacancies on the magnetic ordering in such systems. It is evident that the observed ferromagnetic ordering as well as the measured magnetic moment and *Tc* depends largely on sample preparation techniques. Even in low dimensional systems, there could be several ferromagnetic sources, such as (a) contaminated substrate, (b) clusters of intrinsic defects with ferromagnetically aligned spins, (c) clusters of TM ions, and (d) uncompensated surface spins.

The first step in the synthesis of DMS is the successful incorporation of the TM ion into the host (ZnO) lattice. In this regard, we present our recent work on the synthesis of *Zn*1<sup>−</sup>*xFexO* nanotubes with clear evidence of Fe ions substituting for Zn as 2+ ions in the ZnO crystal lattice. Nanotubes are interesting since the pores in them provide an additional dimension for band gap engineering and tuning of properties.

10.5772/52825

There is no clear understanding on the origin of ferromagnetism in low dimensional II-VI systems such as ZnO when doped with TM ions. There are several important issues that need to be resolved. It needs to be ascertained if the existence of room temperature ferromagnetism is related to some intrinsic origin and not to the presence of defects. If defect induced ferromagnetism is implicated, then the application of this DMS becomes very limited.

### **Author details**

14 Nanowires - Recent Advances

showed a near zero effective magnetic moment.

TM ions, and (d) uncompensated surface spins.

for band gap engineering and tuning of properties.

equation:

**4. Conclusion**

The effective magnetic moment was measured in units of Bohr magnetron (*µB*)using the

where *χA* is the corrected molar susceptibility of the sample at temperature T (room temperature). *χ<sup>A</sup>* for *Zn*1<sup>−</sup>*xFexO* and undoped ZnO nanotubes was determined to be ≈ 1.31 × <sup>10</sup>−<sup>2</sup> cgs units and ≈ 1.68 × <sup>10</sup><sup>−</sup>4cgs units respectively. The effective magnetic moment was then calculated for both samples and found to be ≈ 5.41*µ<sup>B</sup>* for Fe-doped and ≈ 0.62*µ<sup>B</sup>* for undoped ZnO nanotubes. These values of *µeff* correspond to four unpaired electrons in the *Zn*1<sup>−</sup>*xFexO* nanotubes and zero unpaired electrons in the undoped ZnO nanotubes. The four unpaired electrons found in the Fe-doped material is consistent with the electron configuration of *Fe*2<sup>+</sup> in a high spin tetrahedral geometry, and is expected from the incorporation of *Fe*2<sup>+</sup> as a substitutional dopant in ZnO. The undoped ZnO nanotubes

There are several works in favor of and against ferromagnetism in TM doped ZnO. Experimentalists have frequently observed ferromagnetism in this material though the theory to support this claim does not exist. The promise of obtaining materials with high Curie temperature and thus develop spin-based semiconductor devices keeps the field of II-VI DMS very active. The electronic properties of nanoscale DMS are strongly affected by quantum confinement effect and large surface to volume ratios; hence the magnetic coupling between TM ions in low-dimensional systems may be distinctively different from that of bulk materials. Ferromagnetism in Mn-doped ZnO nanowires is yet to be understood, as Mn in its +2 valence state is an isovalent dopant and does not provide any carrier to ZnO. Hence, by first-principles studies, it must lead to an antiferromagnetic ground state. However, experimental observation of room temperature ferromagnetism in Mn-doped ZnO thin films and nanostructures have posed a challenge to the theoretical understanding of ferromagnetism for such systems. To complicate matters further, recent experiments reveal inconsistent ferromagnetic properties for TM doped ZnO, leaving the real mechanism for ferromagnetism in ZnO unclear. Experimental studies have revealed the importance of defects such as zinc and oxygen interstitials and vacancies on the magnetic ordering in such systems. It is evident that the observed ferromagnetic ordering as well as the measured magnetic moment and *Tc* depends largely on sample preparation techniques. Even in low dimensional systems, there could be several ferromagnetic sources, such as (a) contaminated substrate, (b) clusters of intrinsic defects with ferromagnetically aligned spins, (c) clusters of

The first step in the synthesis of DMS is the successful incorporation of the TM ion into the host (ZnO) lattice. In this regard, we present our recent work on the synthesis of *Zn*1<sup>−</sup>*xFexO* nanotubes with clear evidence of Fe ions substituting for Zn as 2+ ions in the ZnO crystal lattice. Nanotubes are interesting since the pores in them provide an additional dimension

*µeff* = 2.828(*χAT*)1/2 (2)

Usha Philipose and Gopal Sapkota

Department of Physics, University of North Texas, USA

### **References**


[26] L. V. Korsunska et al., (2003). The influence of defect drift in external electric field on green luminescence of ZnO single crystals. *Journal of Luminescence*, Vol.102-103: 733-736

16 Nanowires - Recent Advances

Vol.83[5]: 920-922

*Engineering: R: Reports*, Vol.62[1]: 1-35

*Electronic Materials*, Vol.36[4]: 462-471

*Zn*. *Physical Review Letters*, Vol.97[3]: 037203:1-4

*Review B*, Vol.81[3]: 440-444

*Letters*, Vol.78[7]: 958-960

*State Communications*, Vol.121[6]: 371-373

semiconductors. *Nature Materials*, Vol.5: 291-297

[11] F. Pan, C. Song, X. J. Liu, Y. C. Yang and F. Zeng (2008). Ferromagnetism and possible application in spintronics of transition-metal-doped ZnO films. *Materials Science and*

[12] S.J. Pearton et al., (2007). Ferromagnetism in transition-metal doped ZnO. *Journal of*

[13] S. J. Han, T. H. Jang, Y. B. Kim, B. G. Park, J. H. Park and Y. H. Jeong (2003). Magnetism in Mn-doped ZnO bulk samples prepared by solid state reaction. *Applied Physics Letters*,

[14] S. Kolesnik and B. Dabrowski (2004). Absence of room temperature ferromagnetism in

[15] H. S. Hsu et al.,(2006). Evidence of oxygen vacancy enhanced room-temperature ferromagnetism in Co-doped ZnO. *Applied Physics Letters*, Vol.88[24]: 242507:1-3

[16] K. R. Kittilstved, D. A. Schwartz, A. C. Tuan, S. M. Heald, S. A. Chambers and D. R. Gamelin (2001). Direct Kinetic Correlation of Carriers and Ferromagnetism in *Co*2<sup>+</sup> :

[17] H. S. Hsu, et al.,(2008). Magnetic coupling properties of Mn-doped ZnO nanowires: First-principles calculations. *Journal of Applied Physics*, Vol.103[7]: 073903:1-5

[18] C. Zener (1951). Interaction Between the d Shells in the Transition Metals. *Physical*

[19] H. Akai (1998). Ferromagnetism and Its Stability in the Diluted Magnetic

[20] J.M.D Coey, M. Venkatesan and C.B Fitzgerald (2005). Donor impurity band exchange

[21] M. Herbich, A. Twardowski, D. Scalbert and A. Petrou (1998). Bound magnetic polaron in Cr-based diluted magnetic semiconductors. *Physical Review B*, Vol.58[11]: 7024-7034

[22] T. Fukumura et al., (2001). Magnetic properties of Mn-doped ZnO. *Applied Physics*

[23] A. Tiwari, C. Jin, A. Kvit, D. Kumar, J.F Muth and J. Narayan (2002). Structural, optical and magnetic properties of diluted magnetic semiconducting *Zn*1−*xMnxO* films. *Solid*

[24] S.W Jung, S.J An, G. Yi, C.U Jung, S. Lee and S. Cho (2002). Ferromagnetic properties of *Zn*1−*xMnxO* epitaxial thin films. *Applied Physics Letters*, Vol.80[24]: 4561-4563

[25] K.R Kittilstved, W.K Liu and D.R Gameln (2006). Electronic structure origins of polarity-dependent high *TC* ferromagnetism in oxide-diluted magnetic

Semiconductor (In, Mn)As. *Physical Review Letters*, Vol.81[14]: 3002-3005

in dilute ferromagnetic oxides. *Nature Materials*, Vol.4: 173-179

bulk Mn-doped ZnO. *Journal of Applied Physics*, Vol.96[9]: 5379-5381


## **Self-Assembly of Copper Oxide Core-Shell Nanowires Through Ethyl Alcohol**

Sheng Yun Wu

18 Nanowires - Recent Advances

62 Nanowires - Recent Advances

4150-4153

*Physics*, Vol.39: 4762-4765

[40] K.J. Kim and Y.R (2004). Optical investigation of *Zn*1−*xFexO* films grown on *Al*2*O*3(0001) by radio-frequency sputtering. *Journal of Applied Physics*, Vol.96[8]:

[41] A.J. Chen, X.M. Wu, Z.D. Sha, L.T. Zhuge and Y.D. Meng (2006). Structure and photoluminescence properties of Fe-doped ZnO thin films. *Journal of Physics D: Applied*

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52584

## **1. Introduction**

The control of the core-shell structure at the nanometer is one of the most fundamental chal‐ lenges in condensed matter science [Deville et al., 2009; Woodley et al., 2008; Bannin, 2007; Ji et al., 2007; Lauhon et al., 2002]. One of the fundamental concepts regarding the synthesis of nanowires via the spontaneous self-organization of individual nanoparticles, called spontane‐ ous organization nanocrystals, refers to the formation of two-dimensional (2-D) and 3-D ar‐ rays of nanoparticles or nanowires [Tang et al., 2002; Lu et al., 2002; Balazs et al., 2006; Tang et al., 2006]. The copper-oxide based system, such as Cu2O [Huang et al., 2012; Hong et al., 2011; Yao et al., 2010], CuO [Wu, 2012; Cheng et al., 2007; Chou et al., 2008], and Cu2O@CuO [Yec et al., 2012], has been known to facilitate oxidation reactions in the bulk material, which may al‐ low it to be a cost effective substitute for noble metals in various catalytic systems. With recent developments in nanostructures synthesis leading to the ability to control size, reproducibility and structural complexity [Lauhon et al., 2002; Fan et al., 2006; Lu et al., 2005], it becomes worthwhile and possibly paramount to define specific target structures for the nanoparticles based on an understanding of the mechanism of the self-organization mechanism. Upto now, many complex procedures, such as vapor-liquid-solid, chemical vapor deposition, thermal evaporation, and chemical reactions, have been developed for the synthesis of one-dimension‐ al materials of Copper-based nano compounds [Huang et al., 2012; Hong et al., 2011; Yao et al., 2010]. In addition to these methods, solution chemical route including solvothermal, hydro‐ thermal, self-assembly, and template-assisted chemical vapor deposition has become a prom‐ ising option for large –scale production of nanomaterials, due to the simple, fast, and less expensive virtues [Xin et al., 2002; Li et al., 1999; Lu et al., 2000; Hung et al., 2004; Roy et al., 2003; Baxter et al., 2003]. However, a new type of immersing nanoparticles into liquid alcohol

© 2012 Wu; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Wu; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

to form core-shell nanowires at room temperature has been difficult, primarily because control of nucleation and growth is still a challenge.

In the present work, we report on a simple method for bottom-up fabrication by means of template-free growth of high density Cu2O/CuO core/shell nanowires. These can be spontaneously self-assembled by the oxidation of ultra-small copper nanoparticles simply immersed in ethyl alcohol. This leads to the formation of Cu2O/CuO core/shell nano‐ wires. Then nanosized growth effects of these nanowires are observed experimentally. We find that the formation of nanowires may be far more due to chemical reactions than previously realized.

## **2. Synthesis method**

The initial Cu nanoparticles used as a precursor in the synthesis of Cu2O/CuO core/shell nano‐ wires, are fabricated by employing the gas condensation method, as shown in Fig. 1. High puri‐ ty Cu ingots (~ 0.5 g, 99.999% pure and 3 mm in diameter) are heated by a current source of 90 A. They are evaporated at a rate of 0.01 Å/s in an Ar pressure of 0.1 Torr. The evaporated Cu nano‐ particles are collected on a non-magnetic quartz plates (1 cm x 1 cm), maintained at liquid nitro‐ gen temperature. After restoration to room temperature, the nanoparticles, which are only loosely attracted to the collector, are stripped off. The resultant samples were in the form of dried powder, consisting of a macroscopic amount of individual Cu nanoparticles. The com‐ bined effect of high gas flow rate, low deposition pressure and relatively low substrate temper‐ ature resulted in the deposition of copper nanoparticles. The moving Cu nanoparticles were washed and stripped off from quartz plates using ethyl alcohol(99.5%) in argon gas environ‐ ment (chemical hood), as shown in Fig. 2(a). They were then kept in a glass container and in a sealed quartz capillary (Fig. 2(b) and 2(c) for few days) for *in situ* x-ray scattering examination. The sample is apparently not sensitive to being exposed to air for a short period of time, which is supported by the fact that the x-ray diffraction patterns of the as-grown sample and sample exposed to air for few hours are indistinguishable. However, the sample used in the study is kept in glass container with liquid alcohol. The main key method for sample syntheses in this study was that the ethyl alcohol solution-phase was important tofostering spontaneous growth of Cu nanoparticles which then led to the formation of Cu2O/CuO core/shell nanowires at room temperature. In general, metal-oxide nanowires can be synthesized using a solid-state method involving high temperature oxidation of the metal [Cheng et al., 2007; Devan et al., 2008] and a solution-phase under hydrothermal conditions [Tan et al., 2007]. However, a typi‐ cal reaction requires very high temperature or pressure and therefore would seem to be incom‐ patible with our system [Yang et al., 1998]. Copper oxide nanoparticles in a limited size range have been prepared with other techniques using chemical routes, and some of these techniques have resulted in alimited growth of rodlikenanocrystals reported by Fang group [Fang et al., 2004]. Ethyl alcohol (CH3-CH2OH)is chosen as the reactive solution for the spontaneous organ‐ ization method, because of its dipole moment (1.684 Debye), which is comparable to that of wa‐ ter (1.69 Debye), and its polar character, which allows the formation of Cu2O/CuO core/shell nanowires with an alcohol bridge. In the presence of a rich OH solution, spontaneous oxida‐ tion can be drastically accelerated. This procedure has been reported by Tello and coauthors [Tello et al., 2003] for the oxidation of *p*-silicon in water and ethyl alcohol. They demonstrated that local oxidation in liquid ethyl alcohol environments is different from that in aqueous solu‐ tions, with a significant improvement in the growth rate. The Liu group [Liu et al., 2009] also re‐ ported that the dosage rate can influence the morphologies of Cu2O, so that the shape of the nanostructures can be controlled through adjusting the reduction rate of Cu2+ ions. In this study, Cu2+ ions are released continuously from the cooper nanoparticles into the ethyl alcohol under ambient conditions. These can be immediately captured through coordination with OH- , to form Cu(OH)2. After 4-days (D=4) of immersion, the Cu(OH)2 loses water and CuO or Cu2O is formed by the release of O2. The resultant copper-based phases can be either cuprous oxide or cuprite oxide depending on the temperature and surrounding environment. At room temperature, however, the cuprous phase is the dominant product of oxidation. The cuprous surface can be terminated by a few mono layers of cuprite oxide product (crystalline or amor‐ phous phase) at the solution interface, forming Cu2O/CuO core-shell nanowires. The prepared samples were mounted in a capillary rotor sample holder and the process of the formation of nanowires was systematically studied using an *in situ* x-ray diffractometer (Philips X'Pert PRO)with CuKα radiation (λ=1.541 Å) at times ranging from one to seven days. The morpholo‐ gy of sample was characterized by a field emission scanning electron microscope (FE-SEM, JEOL JSM-6500F) equipped with anenergy dispersive x-ray spectroscope (EDS, Oxford Instru‐ ment INCAx-sight 7557). Atomic-resolution transmission electron microscopic (TEM) analysi‐ sand high-resolution transmission electron microscopy (HRTEM) images were taken with the CCD-camera of an electron microscope (JEOL JEM-2100) at 200 kV. Analysis software (Digital Micrograph) was employed to digitalize and analyze the obtained images.

to form core-shell nanowires at room temperature has been difficult, primarily because control

In the present work, we report on a simple method for bottom-up fabrication by means of template-free growth of high density Cu2O/CuO core/shell nanowires. These can be spontaneously self-assembled by the oxidation of ultra-small copper nanoparticles simply immersed in ethyl alcohol. This leads to the formation of Cu2O/CuO core/shell nano‐ wires. Then nanosized growth effects of these nanowires are observed experimentally. We find that the formation of nanowires may be far more due to chemical reactions than

The initial Cu nanoparticles used as a precursor in the synthesis of Cu2O/CuO core/shell nano‐ wires, are fabricated by employing the gas condensation method, as shown in Fig. 1. High puri‐ ty Cu ingots (~ 0.5 g, 99.999% pure and 3 mm in diameter) are heated by a current source of 90 A. They are evaporated at a rate of 0.01 Å/s in an Ar pressure of 0.1 Torr. The evaporated Cu nano‐ particles are collected on a non-magnetic quartz plates (1 cm x 1 cm), maintained at liquid nitro‐ gen temperature. After restoration to room temperature, the nanoparticles, which are only loosely attracted to the collector, are stripped off. The resultant samples were in the form of dried powder, consisting of a macroscopic amount of individual Cu nanoparticles. The com‐ bined effect of high gas flow rate, low deposition pressure and relatively low substrate temper‐ ature resulted in the deposition of copper nanoparticles. The moving Cu nanoparticles were washed and stripped off from quartz plates using ethyl alcohol(99.5%) in argon gas environ‐ ment (chemical hood), as shown in Fig. 2(a). They were then kept in a glass container and in a sealed quartz capillary (Fig. 2(b) and 2(c) for few days) for *in situ* x-ray scattering examination. The sample is apparently not sensitive to being exposed to air for a short period of time, which is supported by the fact that the x-ray diffraction patterns of the as-grown sample and sample exposed to air for few hours are indistinguishable. However, the sample used in the study is kept in glass container with liquid alcohol. The main key method for sample syntheses in this study was that the ethyl alcohol solution-phase was important tofostering spontaneous growth of Cu nanoparticles which then led to the formation of Cu2O/CuO core/shell nanowires at room temperature. In general, metal-oxide nanowires can be synthesized using a solid-state method involving high temperature oxidation of the metal [Cheng et al., 2007; Devan et al., 2008] and a solution-phase under hydrothermal conditions [Tan et al., 2007]. However, a typi‐ cal reaction requires very high temperature or pressure and therefore would seem to be incom‐ patible with our system [Yang et al., 1998]. Copper oxide nanoparticles in a limited size range have been prepared with other techniques using chemical routes, and some of these techniques have resulted in alimited growth of rodlikenanocrystals reported by Fang group [Fang et al., 2004]. Ethyl alcohol (CH3-CH2OH)is chosen as the reactive solution for the spontaneous organ‐ ization method, because of its dipole moment (1.684 Debye), which is comparable to that of wa‐ ter (1.69 Debye), and its polar character, which allows the formation of Cu2O/CuO core/shell

nanowires with an alcohol bridge. In the presence of a rich OH-

solution, spontaneous oxida‐

of nucleation and growth is still a challenge.

previously realized.

64 Nanowires - Recent Advances

**2. Synthesis method**

**Figure 1.** Schematic representation of the gas condensation method. Reproduced with permission [4]; copyrights 2010, Institute of Physics.

**Figure 2.** The processing steps for Cu nanoparticles include: (a) washing of the quartz plate with liquid ethyl alcohol and then collecting the liquid in a glass container; (b)-(c) quartz capillary for the experiments of time dependency of the *in situ* x-ray scattering at D=4 and 7, and (d) at D=7-240. Reproduced with permission [4]; copyrights 2010, Insti‐ tute of Physics.

#### **3. Results and discussion**

The *in situ* x-ray diffraction patterns at various time periods were shown in Fig. 3. As indi‐ cated on the bottom of Fig. 3, there are three nuclear peaks at the {1 1 1}, {2 0 0} and {2 2 0} positions, as indexed based on a cubic Cu -*Fm-3m* structure, with a lattice expansion of ~1.5 % compare with Cu bulk. According to the Scherrer diffraction formula, the average size of initial Cu nanoparticles can be described as:

$$\frac{\left(\Lambda 2\theta\_{\text{B}}\right)^{2}}{\tan^{2}\theta\_{\text{B}}} = \frac{\lambda}{\angle d \, ^{\circ}} \times \frac{\Lambda 2\theta\_{\text{B}}}{\tan \theta\_{\text{B}} \sin \theta\_{\text{B}}} + 16e^{2} \, ^{\circ} \tag{1}$$

where <d> is the mean size, *λ* (=1.541 Å) is the incident x-ray and *Δ*2*θB* is FWHM at the scatter‐ ing angle of 2θB.The obtained mean size of Cu nanoparticles is <d>=5.4 nm, estimating from these peaks in the diffraction pattern. After immersion in an ethyl alcohol solution four days, we observe a significant broader peak around 2*θ*=36°, which is associated with the Cu2O struc‐ ture of the Miller index {1 1 1}, and exhibits the coexistence of core/shell nanocrystals. This may be understood by assuming the existence of the Cu2O phase near the surface, with the oxida‐ tion contribution coming from the Cu atoms in the core. The pattern in the upper part of Fig. 3 should contain, in principle, contributions from the Cu2O phase, after 7-days (D=7) of immer‐ sion. X-ray diffraction patterns are known as the fingerprints of crystalline materials. They re‐ veal details of the crystalline structure and their formation during synthesis, and even the crystalline phase transitions or separation at various temperatures. An example of x-ray and Rietveld refined diffraction patterns of the sample, taken at D=7, is shown in Fig. 4. Diffraction patterns were utilized to characterize the crystalline structure in the prepared samples. The diffraction peaks appeared to be much broader than the instrumental resolution, reflecting the nano-size effects. The analysis was performed using the program package of the General Struc‐ ture Analysis System (GSAS) [Larson et al., 1990] following the Rietveld method [Rietveld, 1969]. Several models with different symmetries were assumed during the preliminary analy‐ sis. In our structural analysis we then pay special attention to searching for the possible sym‐ metries that candescribe the observed diffraction pattern well. All the structural and lattice parameters were allowed to vary simultaneously, and refining processes were carried out un‐ til *R*p, the weighted *Rwp* factor, differed by less than one part in a thousand within two succes‐ sive cycles. Figure 4 shows the diffraction pattern (red crosses) taken at room temperature, where the solid curve (green curve) indicates the fitted pattern and the difference (pink curve) between the observed and the fitted patterns is plotted at the bottom of Fig. 4. The refined lat‐ tice parameters are *a=b=c=*4.2821(5) Å with a space group of Cu2O-*Pn-3m* structure. The forma‐ tion of Cu2O requires an expansion of only 0.026 % of the metallic Cu nanoparticles to make room for the O atoms, comparing with Cu2O bulk (*a*=4.281 Å). The decrease in surface energy of the nanowires will tend to cause an increase in their size by elastic distortion of their crystal lattice. Of course, this kind of distortion is very small compared to the whole particle size. It is worth notingthat the contribution of the intensity of the CuO shell (as *can not* be seen in the xray patterns) is undetectable or amorphous in state. These characteristics (i.e., the amorphous CuO shell) agree with previously reported results for Cu2O/CuO nanoparticles obtained by Yin et al. [Yin et al., 2005]. In their comprehensive analysis, they found that Cu nanocrystals formed initially, but were subsequently oxidized to form a highly crystalline Cu2O coated with an amorphous epitaxial CuO layer.

**Figure 2.** The processing steps for Cu nanoparticles include: (a) washing of the quartz plate with liquid ethyl alcohol and then collecting the liquid in a glass container; (b)-(c) quartz capillary for the experiments of time dependency of the *in situ* x-ray scattering at D=4 and 7, and (d) at D=7-240. Reproduced with permission [4]; copyrights 2010, Insti‐

The *in situ* x-ray diffraction patterns at various time periods were shown in Fig. 3. As indi‐ cated on the bottom of Fig. 3, there are three nuclear peaks at the {1 1 1}, {2 0 0} and {2 2 0} positions, as indexed based on a cubic Cu -*Fm-3m* structure, with a lattice expansion of ~1.5 % compare with Cu bulk. According to the Scherrer diffraction formula, the average size of

<sup>2</sup> <sup>2</sup> 16 ,

*<sup>e</sup> <sup>d</sup>*

q q

where <d> is the mean size, *λ* (=1.541 Å) is the incident x-ray and *Δ*2*θB* is FWHM at the scatter‐ ing angle of 2θB.The obtained mean size of Cu nanoparticles is <d>=5.4 nm, estimating from these peaks in the diffraction pattern. After immersion in an ethyl alcohol solution four days, we observe a significant broader peak around 2*θ*=36°, which is associated with the Cu2O struc‐ ture of the Miller index {1 1 1}, and exhibits the coexistence of core/shell nanocrystals. This may be understood by assuming the existence of the Cu2O phase near the surface, with the oxida‐ tion contribution coming from the Cu atoms in the core. The pattern in the upper part of Fig. 3 should contain, in principle, contributions from the Cu2O phase, after 7-days (D=7) of immer‐ sion. X-ray diffraction patterns are known as the fingerprints of crystalline materials. They re‐ veal details of the crystalline structure and their formation during synthesis, and even the

q

tan tan sin *B B B B B*

l

2

<sup>D</sup> <sup>D</sup> =´ + < > (1)

tute of Physics.

66 Nanowires - Recent Advances

**3. Results and discussion**

initial Cu nanoparticles can be described as:

( ) 2

2

q

q

**Figure 3.** In situtime dependence of x-ray diffraction patterns taken at room temperature.

**Figure 4.** The observed (crosses) and Rietveld refined (solid lines) x-ray diffraction patterns of Cu2O/CuO core-shell nanowires. The contribution of the intensity of the CuO shell *can not* be seen in the x-ray patterns due to amorphous state.

The morphology of selected samples at D=1, 4 and 7, was characterized by FE-SEM, as shown in Figs. 5(a) to (c). As can be seen in Fig. 5(a), it is clearly evident that the nanoparti‐ cles are spherical and stick together due to an electrostatic effect as well as an artifact of the drying of aqueous suspensions. It can be seen that the Cu2O/CuO nanowires grew homoge‐ neously in a large area to form straight nanowires, as shown in Fig. 5(b). The diameters ranged from 100-500 nm and the lengths were up to several tens of microns. As can be seen in Fig. 5(c), the Cu2O/CuO core-shell nanowires are long and straight, ranging from tens of micrometers to more than a hundred micrometers. The growth of the nanowires is clear. An evaluation of the diameter distribution can be obtained and calculated from a portion of the SEM image, as shown in the Figs. 5(d) to 5(f). They are quite asymmetric and can be descri‐ bed using a log-normal distribution function defined as follows:

$$f(d) = \frac{1}{d\sigma\sqrt{2\pi}} \exp\left(-\frac{\left(\ln d - \ln c < d >\right)^2}{2\sigma^2}\right),\tag{2}$$

where <d> is the mean value and *σ* is the standard deviation of the function. The mean di‐ ameters and standard deviations for the nanowires obtained from the fits are <d>=201(5) to 225(5) nm, σ=0.42 and 0.31 nm at D=4 and 7, indicating increasing diameter as the reaction time increases. The small standard deviation (σ< 0.5) of the function indicates that the distri‐ bution is confined to a limited range. It can be seen that the width of the distribution profile broadens due to the crystalline structure. The stability of growth during the formation of core-shell nanowires is confirmed by examination of the SEM images even at D=240 (not shown). This result indicates that increasing the reaction time by immersion in ethyl alcohol *cannot* make the growth of diameter of the Cu2O core more intensive, due to the termination of amorphous CuO at D=7.

**Figure 5.** SEM images of Cu2O/CuO core/shell after various reaction time with (a) D=1, (b) d=4 and (c) D=7.(d)-(f) The distribution of mean diameter <d> at D=1, 4, and 7,respectively.

**Figure 4.** The observed (crosses) and Rietveld refined (solid lines) x-ray diffraction patterns of Cu2O/CuO core-shell nanowires. The contribution of the intensity of the CuO shell *can not* be seen in the x-ray patterns due to amorphous

The morphology of selected samples at D=1, 4 and 7, was characterized by FE-SEM, as shown in Figs. 5(a) to (c). As can be seen in Fig. 5(a), it is clearly evident that the nanoparti‐ cles are spherical and stick together due to an electrostatic effect as well as an artifact of the drying of aqueous suspensions. It can be seen that the Cu2O/CuO nanowires grew homoge‐ neously in a large area to form straight nanowires, as shown in Fig. 5(b). The diameters ranged from 100-500 nm and the lengths were up to several tens of microns. As can be seen in Fig. 5(c), the Cu2O/CuO core-shell nanowires are long and straight, ranging from tens of micrometers to more than a hundred micrometers. The growth of the nanowires is clear. An evaluation of the diameter distribution can be obtained and calculated from a portion of the SEM image, as shown in the Figs. 5(d) to 5(f). They are quite asymmetric and can be descri‐

2

è ø (2)

2

s

1 (ln ln ) ( ) exp , <sup>2</sup> <sup>2</sup> *d d f d d*

æ ö - <> <sup>=</sup> ç ÷ -

where <d> is the mean value and *σ* is the standard deviation of the function. The mean di‐ ameters and standard deviations for the nanowires obtained from the fits are <d>=201(5) to 225(5) nm, σ=0.42 and 0.31 nm at D=4 and 7, indicating increasing diameter as the reaction time increases. The small standard deviation (σ< 0.5) of the function indicates that the distri‐ bution is confined to a limited range. It can be seen that the width of the distribution profile broadens due to the crystalline structure. The stability of growth during the formation of core-shell nanowires is confirmed by examination of the SEM images even at D=240 (not shown). This result indicates that increasing the reaction time by immersion in ethyl alcohol *cannot* make the growth of diameter of the Cu2O core more intensive, due to the termination

bed using a log-normal distribution function defined as follows:

of amorphous CuO at D=7.

s p

state.

68 Nanowires - Recent Advances

The results of an early study (x-ray patterns) show that the prepared core-shell nanowires are crystalline. We further characterized the structure of sample using TEM microscopy. Fig‐ ure 6 shows an example of the Cu2O/CuO core-shell nanowires taken at D=4, which has a uniform coating of amorphous CuO (light contrast) surrounding a single-crystal Cu2O core (dark contrast), forming core-shell Cu2O-CuO nanowires. The chain-like configuration of the Cu2O nanoparticles intertwined with the nanowires can also be observed. The results are in good agreement with the NVT Monte Carlo (MC) method [Phillies, 1974] results for the in‐ termediate state of pearl necklace-like aggregates. In the MC simulation gives the formation of nanowires involves a process where Au nanoparticles could self-assemble into chain-like intermediate states then recrystallize into one-dimensional nanowire. The dipole-dipole-in‐ teraction in the Cu2O nanoparticles is strong and long-range, leading to the formation of the pearl necklace-like nanostructures. A significant surface CuO layer (a few nms in thickness) was observed, indicating the existence of an amorphous CuO shell. In order to determine the amorphous CuO and the surface characteristics of the core-shell nanowires, we have uti‐ lized XANES technique to examine the existence of surface CuO thickness, simulating through an intensity ratio of Cu2O-core and CuO-shell by using a proposed core-shell nano‐ wire model.

For a cylindrically multilayered structure without consider the effect of diameter distribu‐ tion. An electron being photoionized into the surface CuO has a probability of escaping through the surface without suffering any inelastic processes. Therefore, it can be detected as a photoelectron. In generally, the probability is conveniently expressed in terms of the mean free path for surface layer *λ*CuO and core *λ*Cu2O, respectively. The intensity of ionized electron is proportional to *Io exp(-f/λ)*, where *f* is the distance between the surface and the point of origin of the photoelectron, Io is the normalization constant depends on the nano‐ crystals. Thus, integrating the portion of contribution intensity over the cylindrical volume, then we can easily obtain the total intensity from each of the two regions of the nanocrystals. The total intensity from each region in terms of cylindrical polar coordinates can be ex‐ pressed as [Nanda et al., 1999]:

$$I = I\_0 \int\_{-\mathcal{Z}}^{\mathcal{L}} \int\_0^{\mathcal{L}} \int\_{\mathcal{R}\_l}^{\mathcal{R}\_2} \exp\left[\frac{-f(r,\phi,z)}{\mathcal{Z}}\right] r dr d\phi dz,\tag{3}$$

where L is the length of the nanowire, *λ* is the mean free path of the photoionized elec‐ tron,*R1*is the radius of the core, *R2* is the radius including the surface layer. The distance be‐ tween the surface and the point of origin of the photoelectron then can be schematically plotted in Fig. 7 and defined as:

**Figure 6.** Representative TEM image Cu2O/CuO core/shell nanowires, revealing the core-shell structure and a number of interesting portions. (a) The formation of a single Cu2O/CuO core/shell nanowire reveals a straight polycrystalline Cu2O core; and (b) an amorphous CuO layer. (c) The chain-like configuration of Cu2O nanoparticles and (d) pearl-neck‐ lace-like aggregation intertwined with the nanowires can be observed.

$$f(r, \phi, z) = \left[ \left( \sqrt{R^2 - r^2 \cos^2 \phi} - r \sin \phi \right)^2 + z^2 \right]^{\frac{1}{2}} \tag{4}$$

The mean free paths of *λ*CuO and *λ*Cu2O used in this study are obtained from previous reports [36]. The mean free paths of *λ*CuO and *λ*Cu2O used in the simulation are defined as

$$
\lambda\_{\rm CuO} = \frac{0.0062 \,\mathrm{E\_p(Cu^{2+})}}{\log(0.0501 \,\mathrm{E\_p(Cu^{2+})})} \,\mathrm{and}\,\lambda\_{\rm Cu\_2O} = \frac{0.00767 \,\mathrm{E\_p(Cu^+)}}{\log(0.0501 \,\mathrm{E\_p(Cu^+)})} \,\mathrm{nm} \tag{5}
$$

In order to clarify the contribution from Cu2O- or CuO- intensity, we have defined a theoret‐ ical integrated intensity ratio *γ* taken as:

crystals. Thus, integrating the portion of contribution intensity over the cylindrical volume, then we can easily obtain the total intensity from each of the two regions of the nanocrystals. The total intensity from each region in terms of cylindrical polar coordinates can be ex‐

(, ,) exp , *<sup>L</sup> <sup>R</sup>*

where L is the length of the nanowire, *λ* is the mean free path of the photoionized elec‐ tron,*R1*is the radius of the core, *R2* is the radius including the surface layer. The distance be‐ tween the surface and the point of origin of the photoelectron then can be schematically

**Figure 6.** Representative TEM image Cu2O/CuO core/shell nanowires, revealing the core-shell structure and a number of interesting portions. (a) The formation of a single Cu2O/CuO core/shell nanowire reveals a straight polycrystalline Cu2O core; and (b) an amorphous CuO layer. (c) The chain-like configuration of Cu2O nanoparticles and (d) pearl-neck‐

f

The mean free paths of *λ*CuO and *λ*Cu2O used in this study are obtained from previous reports

2

*p p E Cu E Cu and nm E Cu E Cu*

+ +

0.0062 ( ) 0.00767 ( ) log(0.0501 ( )) log(0.0501 ( )) *p p*

 l  f

22 2 22 <sup>2</sup> *fr z R r r z* ( , , ) ( cos sin )

[36]. The mean free paths of *λ*CuO and *λ*Cu2O used in the simulation are defined as

2 2

*CuO Cu O*

l

1

+ + = = (5)

é ù =- - + ê ú ë û (4)

f

l

f

ë û ò òò (3)

2 1

é ù - <sup>=</sup> ê ú

*fr z I I rdrd dz*

<sup>2</sup> <sup>0</sup> <sup>0</sup> <sup>2</sup>


lace-like aggregation intertwined with the nanowires can be observed.

f *L R*

p

pressed as [Nanda et al., 1999]:

70 Nanowires - Recent Advances

plotted in Fig. 7 and defined as:

$$\gamma = \frac{I\_{\rm Cu\_2O}}{I\_{\rm CuO}} = \frac{I\_0 \int\_{-\infty}^{\infty} \int\_0^x \int\_0^{R\_1} \int\_0^{R\_1} \exp\left|\frac{-f(r,\phi,z)}{\lambda\_{\rm Cu\_2O}}\right| r dr d\phi dz}{I\_0 \int\_{-\infty}^{\infty} \int\_0^x \int\_{R\_1}^{R\_2} \exp\left|\frac{-f(r,\phi,z)}{\lambda\_{\rm CuO}}\right| r dr d\phi dz} \tag{6}$$

The above ratio cannot be evaluated analytically, and we can compute numerically by vary‐ ing these parameters of R1 and R2, respectively, to match the experimental integrated inten‐ sity ratio of γ=ICu2O/ICuO. The electronic geometric structure around Cu in the as-synthesized material was characterized with X-ray absorption near edges(XANES) measurements. XANES spectroscopy of Cu L2,3–edge was performed using the 6-m HSGM beam-line (BL20A1) of the National Synchrotron Radiation Research Center in Hsinchu, Taiwan. The recorded spectra were corrected for the energy dependent incident photon intensity as well as for self-absorption effects and normalized to the tabulated standard absorption cross sec‐ tions in the energy range of 900-1020 eV for the Cu L-edge. The radiation was monochrom‐ atized using a 6m SGM monochromator and the photon energy was calibrated by measuring X-ray absorption spectrum of Cu metal foil by using the Cu L3 white line. Figure 8 shows XANES spectra of as-synthesized material and copper based references such as Cu2O (red curve), CuO (green curve) bulk. The main peak of Cu 2P3/2 at 932.3 eV reveals the presence of Cu2O phase. It is accompanied by a second peak at 935.1 eV related to the CuO phase, and can be regarded as characteristic of Cu+ and Cu2+ (2*p*3/2→3*d*), respectively [Yin et al., 2005; Xu et al. 2007; Wen et al., 2005]. It is worth noting that the occurrence of a weak broad satellite feature on the higher energy (~ 943.5 eV) side of the Cu 2p main peak indi‐ cates the slight presence of CuO on the surface. The observed amount of CuO is very tiny due to the slightly oxidized surface of Cu2O nanowires. These results are well consistent with the XRD and TEM observations shown in Figs. 3 and 6. This result can be understood as due to the occurrence of amorphous CuO and crystalline Cu2O on the surface and core, respectively. Based on the proposed core-shell model, we use the proposed model of a sin‐ gle core-shell nanowire with an inner crystalline Cu2O core coated by a thin amorphous CuO surface layer to simulate the ratio of integrated intensity. These can be obtained from the data. The intensity ratio of the Cu2O core and CuO shell is thus determined by integrat‐ ing the peak intensity ICu2O /ICuO. Then we can tune the value of R1 and R2 in the model to fit the ratio of ICu2O/ICuO. All the parameters are allowed to vary simultaneously. Refining proc‐ esses are carried out until the results meet the ratio of ICu2O /ICuO =3.302 obtained from XANES. The refined parameters, as calculated from the model are listed in Table 1. From XANES analysis, we can conclude that the core-shell nanowires are mostly of Cu2O phase. However, CuO phase in the as-grown sample could be explained either by the surface oxi‐ dation in liquid ethyl alcohol or by the core-shell-type of structure of Cu2O nanowires with CuO as the capping layer [Sahoo et al., 2009; Yu et al., 2007].

**Figure 7.** Schematic models of core-shell nanowires based on the photoemission core-level analysis. (a) Top view of this model, where R1 is the radius of core and R2 is the radius including surface layer. (b) Side view of the proposed model, where the penetrated distance is f(r, φ, z) and the r integration is over suitable limits.

**Figure 8.** X-ray absorption spectra of Cu2O/CuO core-shell nanowires at D=4. The experimental spectra and fitted curve are indicated by the solid circles and the line, respectively.


**Table 1.** Summary of the simulation of XANES results for Cu2O/CuO core-shell nanowires at D=4. The XANES of finite size core-shell model is convoluted with the Gaussian instrumental resolution function *G*(*w*1, *E*¯ *<sup>p</sup>*),where *w* and I are the FWHM and amplitudes of peak *E*¯ *<sup>p</sup>*.

## **4. Conclusion**

A new type of spontaneous organization method was successfully utilized to grow Cu2O/CuO core/shell nanowires by immersing copper nanoparticles into liquid alcohol at room temperature. The microstructure of the nanowires during the nucleation and growth process from CuO nanoparticles into Cu2O/CuO core-shell nanowires can be examined us‐ ing time dependent *in situ* x-ray diffraction, SEM, and TEM. We show that the reactions of Cu nanoparticles oxide to form Cu2O/CuO core-shell nanowires can occur at room tempera‐ ture with unusually fast reaction rates. An intermediate state of pearl necklace-like aggre‐ gates formed by a chain-like configuration of Cu2O nanoparticles was observed which then intertwined to form the nanowires. The results agreed with the prediction of NVT Monte Carlo simulation[Phillies, 1974]. A significant thin surface layer was observed on Cu2O nanowires, indicating the existence of an amorphous CuO shell, as also confirmed by the xray absorption near-edge structure technique.

## **Acknowledgment**

**Figure 7.** Schematic models of core-shell nanowires based on the photoemission core-level analysis. (a) Top view of this model, where R1 is the radius of core and R2 is the radius including surface layer. (b) Side view of the proposed

Cu2

 CuO 935.1 eV  background Cu2 O bulk CuO bulk

O/CuO core-shell nanowires

 Cu 2p1/2 952.1 eV

> CuO 955.3 eV

*R1 (nm) R2 (nm) CuO thickness2 (nm)*

925 930 935 940 945 950 955 960

Photon energy (eV)

**Figure 8.** X-ray absorption spectra of Cu2O/CuO core-shell nanowires at D=4. The experimental spectra and fitted

**Simulated ICu2O/ICuO**

**Table 1.** Summary of the simulation of XANES results for Cu2O/CuO core-shell nanowires at D=4. The XANES of finite size core-shell model is convoluted with the Gaussian instrumental resolution function *G*(*w*1, *E*¯ *<sup>p</sup>*),where *w* and I are

**ICu2O/ICuO**

932.3 3.423 935.1 1.037 3.302 3.36 99 101 2

 CuO 943.5 eV

model, where the penetrated distance is f(r, φ, z) and the r integration is over suitable limits.

Cu 2p3/2 932.3 eV

0.0

curve are indicated by the solid circles and the line, respectively.

**Ep(Cu+)** *I1* **Ep(Cu2+)** *I2* **Observed**

the FWHM and amplitudes of peak *E*¯ *<sup>p</sup>*.

0.5

1.0

Relative intensity (a. u.)

72 Nanowires - Recent Advances

1.5

2.0

2.5

The authors would like to thank the National Science Council of the Republic of China for financially supporting this research under Contract No. NSC-100-2112-M-259-003-MY3. We would also like to thank Professor Yuan-Ron Ma of National Dong Hwa University and Dr. Ting Shan Chan of National Synchrotron Radiation Research Center (Taiwan) for their val‐ uable contributions and discussions in this work

## **Author details**

#### Sheng Yun Wu

Department of Physics, National Dong Hwa University, Hualien, Taiwan, Republic of China

## **References**


[18] Lu, C. H.; Qi, L. M.; Yang, J. H.; Wang, X. Y.; Zhang, D. Y.;Xie, J. L. & Ma, J. M. (2005) One-Pot Synthesis of Octahedral Cu2O Nanocages via a Catalytic Solution Route, *Adv. Mater.* Vol. 17: 2562-2567.

[4] Ji, J.-Y.; Shih, P.-H.; Yang, C. C.; Chan, T.-S.; Ma, Y.-R.& Wu, S. Y. (2010) Spontaneous self-organization of Cu2O/CuO core-shell nanowires from copper nanoparticles,

[5] Lauhon, L. J.; Gudlksen, M. S.; Wang, D. & Lieber, C. M. (2002) Epitaxial core–shell

[6] Tang, Z.; Kotov, N. A. & Giersig, M. (2002) Spontaneous Organization of Single CdTe

[7] Lu, Q.; Gao, F. & Zhao, D. (2002) One step synthesis and assembly of copper sulfide nanoparticles to nanowires, nanotubes and nanovesicles through an organic amine-

[8] Balazs, A. C.; Emrick, T.& Russell, T. P. (2006) Nanoparticle Polymer Composites:

[9] Tang, Z.; Zhang, Z.; Wang, Y.;Glotzer, S. C. &Kotov, N. A. (2006) Self-Assembly of

[10] Huang, W. C.;Lyu, L. M.; Yang, Y. C.& Huang, M. H. (2012) Synthesis of Cu2O Nano‐ crystals from Cubic to Rhombic Dodecahedral Structures and Their Comparative

[11] Hong, F.; Sun, S.; You, H.; Yang, S.; Fang, J.; Guo, S.; Yang, Z.; Ding, B.& Song, X. (2011) Cu2O Template Strategy for the Synthesis of Structure-Definable Noble Metal

[12] Yao, K. X.; Yin, X. M.; Wang, T. H. & Zeng, H. C. (2010) Synthesis, Self-Assembly, Disassembly, and Reassembly of Two Types of Cu2O Nanocrystals Unifaceted with

[13] Wu, S. Y. (2012) Design and Controlling of in-plane CuO Nanowires: An Investiga‐ tion of Growth Mechanism and Phonon Confinement Effect, *J. Nanosci. Lett.* Vol. 2: 5.

[14] Cheng, C. L.; Ma, Y.R.; Chou, M. H.; Huang, C. Y.;Yeh, V. & Wu, S. Y. (2007) Direct Observation of Short-Circuit Diffusion During the Formation of a Single Cupric Ox‐

[15] Chou, M. H.; Liu,S. B.; Huang, C. Y.;Wu,S. Y. & Cheng,C.L. (2008) Confocal Raman Spectroscopic Mapping Studies on a Single CuO Nanowire, *Applied Surface Science*

[16] Yec, C. C. & Zeng, Hua, C. (2012) Synthetic Architecture of Multiple Core–Shell and Yolk–Shell Structures of (Cu2O@)nCu2O (n = 1–4) with Centricity and Eccentricity,

[17] Fan, H. J.; Knez, M.; Scholz, R.; Nielsch, K.; Pippel, E.; Hesse, D.; Zacharias, M. & Gö‐ sele, U. (2006) Monocrystalline spinel nanotube fabrication based on the Kirkendall

CdTe Nanocrystals into Free-Floating Sheets, *Science* Vol. 314, 274-278.

and core–multishell nanowire heterostructures, *Nature* Vol. 420: 57-61.

Nanoparticles into Luminescent Nanowires, *Science* Vol. 297: 237.

assisted hydrothermal process,*Nano Lett.* Vol. 2, 725-728.

Where Two Small Worlds Meet, *Science* Vol. 314, 1107-1110.

Photocatalytic Activity, *J. Am. Chem. Soc.* Vol. 134: 1261-1267.

Alloy Mesocages, *Crystal Growth & Design* Vol. 11: 3694-3697.

{001} or {110} Planes, *J. Am. Chem. Soc.* Vol. 132: 6131–6144.

ide Nanowire, *Nanotechnology* Vol. 24: 245604-245608.

*Chemistry of Materials* Vol.24: 1917-1929.

effect, *Nature Materials* Vol. 5: 627-631.

Vol. 254: 7539-7543.

*Nanotechnology* Vol. 21: 045603.

74 Nanowires - Recent Advances


## **Synthesis, Superhydrophobicity, Enhanced Photoluminescence and Gas Sensing Properties of ZnO Nanowires**

M.G. Gong, Y.Z. Long, X.L. Xu, H.D. Zhang and B. Sun

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52586

## **1. Introduction**

[34] Phillies, G. D. J. (1974) Excess chemical potential of dilute solutions of spherical poly‐

[35] Nanda, J.; Kuruvilla, B. A.& Sarma, D. D. (1999) Photoelectron spectroscopic study of

[36] Xu, J. S. & Xue, D. F. (2007) Five branching growth patterns in the cubic crystal sys‐ tem: A direct observation of cuprous oxide microcrystals, *Acta Mater.* Vol. 55:

[37] Wen, X.; Xie, Y.; Choi, C. L.; Wan, K. C.; Li, X. Y.& Yang, S. (2005) Copper-based nanowire materials-Templated syntheses, characterizations, and applications, *Lang‐*

[38] Sahoo, S.; Husale, S.; Colwill, B.; Lu, T. M.; Nayak, S. & Ajayan, P. M. (2009) Electric field directed self-assembly of cuprous oxide nanostructures for photon sensing. *ACS*

[39] Yu, Z. Q.; Wang, C. M.; Engelhard, M. H.; Nachimuthu, P.; McCready, D. E.; Lyubi‐ netsky, I. V. & Thevuthasan, S. (2007) Epitaxial growth and microstructure of Cu2O

nanoparticle/thinfilms on SrTiO3(100), *Nanotechnology* Vol. 18: 115601-5.

electrolytes, *J. Chem. Phys.*Vol.60: 2721-2731.

2397-2406.

76 Nanowires - Recent Advances

*muir* Vol. 21: 4729-4737.

*nano* Vol. 3: 3965-3944.

CdSe nanocrystallites, *Phys. Rev. B* Vol. 59: 7473-7479.

The past decades have witnessed major worldwide interested in the fabrication of nanodevi‐ ces based on quasi one-dimensional (1D) semiconducting nanostructures (Xu and Wang, 2011; Long *et al.*, 2011; Long *et al.*, 2012). ZnO, an important II-VI group semiconductor with many excellent properties (Xu and Wang, 2011), has been suggested for a diverse range of applications, including optoelectronics (Ko *et al.*, 2011; Wang *et al.*, 2004), ultraviolet laser devices (Liu *et al.*, 2011), gas sensors (Wang *et al.*, 2011; Ra *et al.*, 2008; Wang *et al.*, 2006) and transparent electrodes (Kusinski *et al.*, 2010; Goris *et al.*, 2009), etc. 1D ZnO nanorods, nano‐ wires and nanofibers have been synthesized by a range of techniques, such as top-down ap‐ proaches by etching (Wu *et al.*, 2004), and wet chemical methods (Vayssieres *et al.*, 2001; Gong *et al.*, 2009), metal-organic chemical vapor deposition (MOCVD) (Yang *et al.*, 2004; Pfüller *et al.*, 2011; Lee *et al.*, 2004), physical vapor deposition (Huang *et al.*, 2001; Jeong and Lee, 2010; Wang *et al.*, 2005; Hsu *et al.*, 2005), molecular beam epitaxy (MBE) (Heo *et al.*, 2002), pulsed laser deposition (Shen *et al.*, 2010; Cao *et al.*, 2007), sputtering (Chiou *et al.*, 2003), flux (Kong and Li, 2003), electrospinning (Sen *et al.*, 2011; Wu *et al.*, 2008; Lin *et al.*, 2007; Sui *et al.*, 2005) methods, etc.

Among those techniques, MOCVD and MBE can yield high quality ZnO nanowires arrays, but usually suffer from low product yield, poor deposition uniformity, and limited choices of substrate. Especially, the fabrication cost is usually very high, so they have been less widely used. The physical vapor deposition and flux methods usually require high tempera‐ ture, which are less likely to be able to integrate with flexible organic substrates for future

© 2012 Gong et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Gong et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

foldable and portable electronics, and can easily incorporate catalysts or other impurities in‐ to the ZnO nanowires. Top-down, pulsed deposition and sputtering approaches have less reproducibility and controllability compared with other techniques. Electrospinning usually yields polycrystalline nanowires or fibers, which may lower down the mobility of charge carriers in devices. Compared with the above methods, wet chemical methods are lower cost, higher reproducibility, less hazardous, easier controllability, and thus can be able to large area of preparation. Meanwhile, the nanowires growth temperature is relatively low and compatible with most flexible organic substrates. There are a variety of parameters that can be tuned to effectively control the nanowires length, diameter, distribution and proper‐ ties of the final product (Gong *et al.*, 2009; Govender *et al.*, 2004). In summary, wet chemical methods have been demonstrated as one of very powerful and versatile techniques for growing 1D ZnO nanowires.

In this chapter, we focus on the 1D ZnO nanostructures that have been prepared by simple hydrothermal self-assembly method and electrospinning. The morphology of ZnO nano‐ structures were characterized by field emission scanning electron microscopy (SEM). The special reticulate structure assembled by ZnO nanowires exhibits superhydrophobicity with a contact angle of 170° and a sliding angle of 2°. Significant photoluminescence enhance‐ ment from ZnO nanowires coated with Ag nanoparticles is observed at 387.6 nm, the ratio of the enhancement reaches 6.6 compared with pure ZnO nanowires. Electrospun ZnO nanowires have also been fabricated into field-effect transistors and sensor devices. The sen‐ sors exhibit fast and large response to CO even at room temperature due to highly porous, continuous and compact-grains structures of the electrospun ZnO nanowires.

## **2. Experimental details**

#### **2.1. Hydrothermal self-assembly method**

Well-aligned single crystalline ZnO nanowires were prepared from zinc acetate dehydrate in a neutral aqueous solution under hydrothermal conditions. In detail, the Si substrates were first immersed in boiling 98% H2SO4 for 10 min and then ultrasonically washed with acetone, ethanol and deionized water, respectively. In order to fabricate well-aligned dis‐ persed ZnO nanowires, a two-step method was used (Wang *et al.*, 2008; Gong *et al.*, 2010): (1) A 10 nm ZnO seed layer was first deposited on a Si substrate by a radio-frequency magnet‐ ron sputtering method. A commercially supplied ZnO ceramic plate with a purity of 99.99% was used as a target. The background pressure of the vacuum chamber was 1×10-4 Pa. A mixed gas of oxygen and argon with a volume ratio of 1:1 was used as the sputtering gas with a total pressure of 3 Pa. The ZnO seeds were annealed in air for 1hr at 550°C. (2) The hydrothermal growth was carried out at 100°C by immersing the substrates vertically in a mixed aqueous solution containing zinc acetate dehydrate (Zn-AD) and hexamethylenetetr‐ amine (He-T). The mixed volume ratio of the two solutions is 1:1, in which the concentration of the two solutions is always the same (i.e. 0.01M/L zinc acetate dehydrate mixed with 0.01M/L hexamethylenetetramine, etc.). The growth time is 12hr. After growth, the sub‐ strates were dried in the Far Infrared Drying Oven at 80°C for 5 min (samples 1, 3-6) just after they were immersed in deionized water for 10 seconds. As a comparison, sample 1\* is the same as sample 1 except it was not immersed in deionized water. Another substrate was immersed in ethanol for 30min, then in n-Hexane for 90min, drying under natural condi‐ tions (sample 2). The reason we chose the different temperature is that the evaporation rate of water at 80°C is almost the same as that of ethanol and n-Hexane at room temperature. Hydrothermal growth and treatment process of ZnO samples are shown in Table 1.The re‐ sults and the formation mechanism will be discussed in section 3. The morphology of the ZnO film was revealed by SEM images.


\*The advancing and the receding contact angles (AA/RA), and the contact and the sliding angles (CA/SA) of water droplets on the samples surface (error: ±1°).

**Table 1.** Hydrothermal solution composition (Zn-AD: zinc acetate dehydrate,andHe-T:hexamethylenetetramine) process and superhydrophobicity of the ZnO nanowires structure

#### **2.2. Electrospinning**

foldable and portable electronics, and can easily incorporate catalysts or other impurities in‐ to the ZnO nanowires. Top-down, pulsed deposition and sputtering approaches have less reproducibility and controllability compared with other techniques. Electrospinning usually yields polycrystalline nanowires or fibers, which may lower down the mobility of charge carriers in devices. Compared with the above methods, wet chemical methods are lower cost, higher reproducibility, less hazardous, easier controllability, and thus can be able to large area of preparation. Meanwhile, the nanowires growth temperature is relatively low and compatible with most flexible organic substrates. There are a variety of parameters that can be tuned to effectively control the nanowires length, diameter, distribution and proper‐ ties of the final product (Gong *et al.*, 2009; Govender *et al.*, 2004). In summary, wet chemical methods have been demonstrated as one of very powerful and versatile techniques for

In this chapter, we focus on the 1D ZnO nanostructures that have been prepared by simple hydrothermal self-assembly method and electrospinning. The morphology of ZnO nano‐ structures were characterized by field emission scanning electron microscopy (SEM). The special reticulate structure assembled by ZnO nanowires exhibits superhydrophobicity with a contact angle of 170° and a sliding angle of 2°. Significant photoluminescence enhance‐ ment from ZnO nanowires coated with Ag nanoparticles is observed at 387.6 nm, the ratio of the enhancement reaches 6.6 compared with pure ZnO nanowires. Electrospun ZnO nanowires have also been fabricated into field-effect transistors and sensor devices. The sen‐ sors exhibit fast and large response to CO even at room temperature due to highly porous,

Well-aligned single crystalline ZnO nanowires were prepared from zinc acetate dehydrate in a neutral aqueous solution under hydrothermal conditions. In detail, the Si substrates were first immersed in boiling 98% H2SO4 for 10 min and then ultrasonically washed with acetone, ethanol and deionized water, respectively. In order to fabricate well-aligned dis‐ persed ZnO nanowires, a two-step method was used (Wang *et al.*, 2008; Gong *et al.*, 2010): (1) A 10 nm ZnO seed layer was first deposited on a Si substrate by a radio-frequency magnet‐ ron sputtering method. A commercially supplied ZnO ceramic plate with a purity of 99.99% was used as a target. The background pressure of the vacuum chamber was 1×10-4 Pa. A mixed gas of oxygen and argon with a volume ratio of 1:1 was used as the sputtering gas with a total pressure of 3 Pa. The ZnO seeds were annealed in air for 1hr at 550°C. (2) The hydrothermal growth was carried out at 100°C by immersing the substrates vertically in a mixed aqueous solution containing zinc acetate dehydrate (Zn-AD) and hexamethylenetetr‐ amine (He-T). The mixed volume ratio of the two solutions is 1:1, in which the concentration of the two solutions is always the same (i.e. 0.01M/L zinc acetate dehydrate mixed with 0.01M/L hexamethylenetetramine, etc.). The growth time is 12hr. After growth, the sub‐

continuous and compact-grains structures of the electrospun ZnO nanowires.

growing 1D ZnO nanowires.

78 Nanowires - Recent Advances

**2. Experimental details**

**2.1. Hydrothermal self-assembly method**

Electrospinning has been recognized as an efficient and highly versatile method which can be further developed for mass production of uniform, ultrafine and continuous fibers with nanometer-to-micrometer sized diameter. Up to date, a variety of materials such as poly‐ mers, metal oxides, ceramics, metals and carbon have been successfully electrospun into ul‐ trafine fibers mostly from solvent solution(Huang *et al.*, 2003; Tan *et al.*, 2008; Li *et al.*, 2010; Li *et al.*, 2011; Sun *et al.*, 2012; Zhang *et al.*, 2012; Zheng et al., 2012). In this work, ZnO nano‐ wires are also fabricated by calcination of the electrospun fibers of polyvinyl acetate (PVA)/ zinc acetate composite. The processing involved three steps: (1) Preparation of a sol with suitable inorganic precursor and polymer content, and achieving an appropriate viscosity for electrospinning. (2) Spinning of the solution to obtain composite fibers. (3) Calcination of the composite fibers to remove the polymer coumpounds and obtain the pure oxide nano‐ fibers. In a typical procedure, aqueous PVA solution (2.0 g PVA and 18.0 g deionized water) is dropped slowly into the solution of zinc acetate (2.0 g zinc acetate and 2.0 g deionized wa‐ ter), and the reaction proceeds in a water bath at 90 °C for 2 h.The PVA/zinc acetate compo‐ site is cooled down to room temperature, and then the precursor solution is introduced into a syringe with metal spinneret, which is connected to positive electrode of a high-voltage power supply. The spinneret tip-to-collector distance is usually in the range of 8-12 cm, and a dc high voltage in the range of 10-15 kV is necessary to generate the electrospinning, as shown in Fig. 1a. A dense web of PVA/zinc acetate composite wires is then collected on the aluminium foil. Finally the as-electrospun composite wires are calcinated at elevated tem‐ peratures (500-600°C) to decompose the PVA organic components, meanwhile the inorganic precursors oxidize and crystallize to form ZnO nanowires.

Fig.1b-e show the morphology evolution of electrospun ZnOwires following hydrolyzation, electrospinning, and calcination processes by our group. After calcination, the average di‐ ameter of nanowires appears to be decreased (80-200 nm) attributed to the loss of polymer compounds from the composite fibers and the crystallization of metal oxide. Moreover, the wires often exhibited continuous and compact-grains structures. It reveals that the electro‐ spun wires have a high surface-to-volume ratio. This peculiar morphology enlarges contact area between the fiber and surrounding gas. It may play a pivotal role in the ultra-high sen‐ sitivity gas sensor of metal oxide by electrospinning (Kim *et al.*, 2006; Landau *et al.*, 2009).

**Figure 1.** a) Schematic illustration to show polymer nanofibers by electrospinning.SEM images of (b) as-spun PVA/zinc acetate composite fibers and (c) ZnO fibers calcined at 600 °C for 5 h. (d-e) High-magnified SEM images of individual ZnO fibers.

## **3. Superhydrophobicity of ZnO Nanowires by Hydrothermal Method**

#### **3.1. Amalgamation of ZnO Nanowires**

the composite fibers to remove the polymer coumpounds and obtain the pure oxide nano‐ fibers. In a typical procedure, aqueous PVA solution (2.0 g PVA and 18.0 g deionized water) is dropped slowly into the solution of zinc acetate (2.0 g zinc acetate and 2.0 g deionized wa‐ ter), and the reaction proceeds in a water bath at 90 °C for 2 h.The PVA/zinc acetate compo‐ site is cooled down to room temperature, and then the precursor solution is introduced into a syringe with metal spinneret, which is connected to positive electrode of a high-voltage power supply. The spinneret tip-to-collector distance is usually in the range of 8-12 cm, and a dc high voltage in the range of 10-15 kV is necessary to generate the electrospinning, as shown in Fig. 1a. A dense web of PVA/zinc acetate composite wires is then collected on the aluminium foil. Finally the as-electrospun composite wires are calcinated at elevated tem‐ peratures (500-600°C) to decompose the PVA organic components, meanwhile the inorganic

Fig.1b-e show the morphology evolution of electrospun ZnOwires following hydrolyzation, electrospinning, and calcination processes by our group. After calcination, the average di‐ ameter of nanowires appears to be decreased (80-200 nm) attributed to the loss of polymer compounds from the composite fibers and the crystallization of metal oxide. Moreover, the wires often exhibited continuous and compact-grains structures. It reveals that the electro‐ spun wires have a high surface-to-volume ratio. This peculiar morphology enlarges contact area between the fiber and surrounding gas. It may play a pivotal role in the ultra-high sen‐ sitivity gas sensor of metal oxide by electrospinning (Kim *et al.*, 2006; Landau *et al.*, 2009).

**Figure 1.** a) Schematic illustration to show polymer nanofibers by electrospinning.SEM images of (b) as-spun PVA/zinc acetate composite fibers and (c) ZnO fibers calcined at 600 °C for 5 h. (d-e) High-magnified SEM images of individual

ZnO fibers.

precursors oxidize and crystallize to form ZnO nanowires.

80 Nanowires - Recent Advances

Fig. 2 shows the long and slender ZnO nanowires fabricated on the seed layer in the mixed aqueous solution of Zn-AD plus He-T. The ratio of length to diameter is about 104 to 1. Interestingly, the ZnO nanowires in sample 1 self-assembled into reticulate papillary node film (ZnO-RPNF). The growth process of the ZnO-RPNF is illustrated in Fig.3.As shown in Fig.3, during the evaporation and coagulation process of water film on the ZnO nanowires, the ZnO nanowires were aggregated to be a reticulate papillary nodes structure(see Fig.2a). As for the comparison, the ZnO nanowires treated by etha‐ nol and n-Hexane only have a small bending and also remain in their separate condi‐ tion (see Fig.2c). On the other hand, the nano-wires of the sample 1\* soaked in residual water of the reacted solution can also be bended and assembled to a papillary like structure (as shown in Fig.2e), after the water evaporated. Since the solution on the sur‐ face of sample 1\* was not pure water and the surface tension of the solution is lower, its bending and assembly grade is lower than that of sample 1. The bending and bun‐ dling mechanism of vertically aligned ZnO nanowires synthesized by vapor phase dep‐ osition method was discussed by other group (Liu *et al.*, 2008). In their report the bending mechanism was attributed to electrostatic interactions between different termi‐ nation facets with opposite charges. However, the mechanism is different in a liquid phase. Although there are a lot of positive and negative particles in an aqueous solu‐ tion, a large number of homologous charges would not gather in a given region in a stoichiometric solution. So the EI effect among ZnO nanowires grown by the liquid phase deposition method can be ignored. At 80 °C, the water film on the top of ZnO nanowires evaporates and contracts rapidly to form tiny drops. Therefore, the water has a centripetal surface tension, making the nanowires to be bent and aggregated. On the contrast, the low liquid surface tension (n-Hexane) makes a little bending but not aggre‐ gation; the specific reason is explained by the tension junction model.

It is known that the wettability of an ideal surface, expressed with contact angle *θ* of water droplets, is given by Young's equation:

$$\cos \theta = \frac{\mathcal{Y}\_{SV} - \mathcal{Y}\_{SL}}{\mathcal{Y}\_{LV}} \tag{1}$$

where*γSV*,*γSL*and*γLV*refer to the interfacial surface tensions with solid-vapor, solid-liquid and liquid-vapour, respectively. Young's angle is a result of the thermodynamic equilibrium of the free energy at the solid-liquid-vapor interphase.

**Figure 2.** FESEM images of (a) and (b) sample 1, (c) and (d) sample 2, and (e) and (f) sample 1\*, where (b), (d) and (f) are the high-magnification for (a), (c) and (e), respectively. The insets show images of droplets that formed on the sur‐ face of the different samples during contact angle measurement.

**Figure 3.** Formation mechanism of the reticulate ZnO film sample 1.

When the ZnO nanowires are free standing in water (or in n-Hexane), it is in a balance state, and there is no force to overcome the elastic deformation force for bending. The model pro‐ posed to calculate the magnitude of the forces responsible for the bending is shown in Fig.4. During the drying process, the balance will be broken. The liquid surface tension can be div‐ ided parallel to an X-axis factor *γ<sup>1</sup>* and a Y-axis parallel to the nanowire edge factor *γ2*. Note that the *γ2* and the pressure difference *∆P=PL-PG (PL* and *PG* are the pressure of the liquid and the atmospheric pressure, respectively) on the droplets arising from the difference in the hollow of the drop cannot let the nanowires be bent. The self-assembling force *F1*, parallel to *γ1*.*F1* can be obtained from:

Synthesis, Superhydrophobicity, Enhanced Photoluminescence and Gas Sensing Properties of ZnO Nanowires http://dx.doi.org/10.5772/52586 83

$$F\_1 = F\sin\theta = \eta b \sin\theta \tag{2}$$

where*γ*= *γ1*+*γ2*is the liquid tension, *b*= 3d/3 is the hexagonal side length of the ZnO nano‐ wires (*d* is the diameter of the ZnO nanowires).

**Figure 4.** Schematic model of liquid evaporation induced change of the bending angles between two ZnO nanowires, where (a) indicates the initial state of water with two nanowires system; (b) is any intermediate state during the bend‐ ing process.The bending angle vs. the nanowires diameter is as shown in (c).

To simplify the case, we just consider the liquid surface tension between two nanowires. The initial state for two nanowires with lengths *T* is shown by Fig.4a, where the two nanowires lengths are defaulted as the same. The self-assembling parallel tension pressed on ZnO nanowires with water *F*w and with n-Hexane *F*h can be expressed as:

**Figure 2.** FESEM images of (a) and (b) sample 1, (c) and (d) sample 2, and (e) and (f) sample 1\*, where (b), (d) and (f) are the high-magnification for (a), (c) and (e), respectively. The insets show images of droplets that formed on the sur‐

When the ZnO nanowires are free standing in water (or in n-Hexane), it is in a balance state, and there is no force to overcome the elastic deformation force for bending. The model pro‐ posed to calculate the magnitude of the forces responsible for the bending is shown in Fig.4. During the drying process, the balance will be broken. The liquid surface tension can be div‐ ided parallel to an X-axis factor *γ<sup>1</sup>* and a Y-axis parallel to the nanowire edge factor *γ2*. Note that the *γ2* and the pressure difference *∆P=PL-PG (PL* and *PG* are the pressure of the liquid and the atmospheric pressure, respectively) on the droplets arising from the difference in the hollow of the drop cannot let the nanowires be bent. The self-assembling force *F1*, parallel to

**ZnO dispersed nanowires**

face of the different samples during contact angle measurement.

**Figure 3.** Formation mechanism of the reticulate ZnO film sample 1.

**ZnO seed layer**

82 Nanowires - Recent Advances

*γ1*.*F1* can be obtained from:

$$\frac{F\_w}{b} = \mathcal{V}\_w \sin \theta\_w \tag{3}$$

$$\frac{F\_h}{b} = \mathcal{V}\_h \sin \theta\_h \tag{4}$$

where*γw* and *γ<sup>h</sup>* are the surface tension of water and n-Hexane, *θw* and *θ<sup>h</sup>* are the contact an‐ gle of water and n-Hexane on a ZnO nanowire, respectively. *γw*=62.61mN/m at 80°C, and *γh*=18.4 mN/m, *θw* =33°, *θh*= 0° at 20 °C, respectively. From the calculation, we observe *Fw*=34.1*b*mN/m, and *Fh*=0. For sample 1, the average diameter of the nanowires *d* is 42.3 nm. Therefore, the parallel surface tension of water is *Fw*= 8.33×10-10 N.

A tension junction model as shown in Fig.4 is used to estimate the minimum parallel surface tension which is necessary to make two nanowires contact each other. For simplicity, we as‐ sumed two symmetric flexible nanowires separated by a distance *n*. Furthermore, we ap‐ proximated the parallel surface tension applied on the top of the nanowire. The dependence of the lateral displacement x perpendicular to the nanowire on the elastic bending force ap‐ plied at the top is given as (Song *et al.*, 2005):

$$F\_{\text{elastic}} = \frac{\text{3EI}}{T^3} \,\text{x} \tag{5}$$

where *T*, *E* and *I* are the length, Young's modulus and moment of inertia of the ZnO nano‐ wire, respectively. The moment of inertia of a hexagonal-cross-section nano-wire can be ex‐ pressed as *I* =5 <sup>3</sup> *d* <sup>4</sup> / 144. To make these two symmetric wires a contact at *x*=*n*/2, the parallel surface tension has to be larger than the bending force for *n*/2 displacement.

Reported values for elastic modulus of ZnO nanostructures vary greatly depending on ex‐ perimental methods and nanostructure dimensions. For example, Wang *et al.* measured the elastic modulus of vertical ZnO nanowires on sapphire to be 29±8 GPa based on atomic force microscopy (Song *et al.*, 2005). Their nanowires had an average diameter of 45nm and length ranging from 0.277 to 0.638 *μ*m. By manipulating the vertical ZnO nanowires with an atomic force microscopy tip, Hoffmann *et al.* measured an average elastic modulus of about 100 GPa for 250 nm diameter ZnO nanowires (Hoffmann *et al.*,2007). Huang *et al.* used trans‐ mission electron microscopy to measure the elastic bending modulus of ZnO nano-wires via electric-field-induced resonant excitation and obtained a mean elastic modulus of ~58 Gpa (Huang *et al.*, 2006). In particular, they measured the elastic modulus of 48.9GPa from a ZnO nanowire with diameter of 43 nm and length of 4.77 *μ*m, which is similar to that (*d*=42.3 nm, T=4.41 *μ*m) of sample 1. If choosing *E*=50 Gpa while *n* = 224 nm based on the nanowires den‐ sity in Table 1 to calculate the elastic bending force, we therefore obtain

$$F\_{\text{elastic}} = 3.77 \times 10^{-11} \text{(N)}\tag{6}$$

For sample 1 the water's parallel surface tension *Fw*= 8.33×10-10 (N) is about 22 times more than *Felastic*. Therefore, the surface tension of deionized water is big enough to let two or more nanowires bundling together to form the reticulate papillary nodes structure. Note that al‐ though the parallel surface tension of n-Hexane is *F*h = 0, after the substrate was immersed in the n-Hexane, a little bending of the nanowires of sample 2 still happened (see Fig.2d), due to that the residual water in solution cannot be thoroughly removed.

As shown in Fig.4b, the angle *α* (the angle between the Y-axis and the tangent at the nano‐ wire tip) is used for defining any intermediate state during the bending process. The rela‐ tionship between the *F1* (required to overcome the bending-induced elastic energy, for water: *F1*= *Fw*) and the bending angle is *α*=*F1T2* /2*EI* (Hibbeler, 1997). Since there is no other external force in X-axis, the bending forces experienced by the two nanowires must be equal. To simply obtain the relation between the bending angle and the nanowires diameter, we assumed that the parallel surface tension of water, Young's modulus and the length of ZnO nanowires are constant as *Fw*=8.33×10-10N, *E*=50 GPa and *T*=4.41 *μ*m, respectively. The curve in Fig.4c is the bending angle as a function of the nanowires diameter of samples 1, 3-6. The bending angle decreases sharply as the diameter increases. The bending will be faded out as the diameter increases enough.

All obtained ZnO samples are wurtzite structure, as shown by SEM images in Figs.2 and 5. Fig.2a is SEM top image of the reticulate ZnO film, exhibiting the homogeneous micro/nanostructures on a large scale, Fig. 2b is the high magnification. The top inset of Fig.5a shows a selected area electron diffraction pattern of the ZnO-RPNF that confirms the single crystal‐ line property of the ZnO nanowire. The average reticulate width and density, and the fac‐ tors for the surface roughness of the reticulate ZnO film, are also determined from top-view SEM image. The reticulate ZnO nanowires (sample 1) with the diameters from 20 to 60 nm (the average is 42.3 nm) are even distributed on the whole substrates. The density of surface nodes (top of the nanowire or the papillary) was decreased via assembly of about ten nano‐ wires to one papillary node. In our data, the density was changed from (2.0±0.5)×109 /cm2 to (2.0±0.5)×108 /cm2 (see Table1).

3 3 *elastic* <sup>=</sup> *EI F x*

where *T*, *E* and *I* are the length, Young's modulus and moment of inertia of the ZnO nano‐ wire, respectively. The moment of inertia of a hexagonal-cross-section nano-wire can be ex‐

Reported values for elastic modulus of ZnO nanostructures vary greatly depending on ex‐ perimental methods and nanostructure dimensions. For example, Wang *et al.* measured the elastic modulus of vertical ZnO nanowires on sapphire to be 29±8 GPa based on atomic force microscopy (Song *et al.*, 2005). Their nanowires had an average diameter of 45nm and length ranging from 0.277 to 0.638 *μ*m. By manipulating the vertical ZnO nanowires with an atomic force microscopy tip, Hoffmann *et al.* measured an average elastic modulus of about 100 GPa for 250 nm diameter ZnO nanowires (Hoffmann *et al.*,2007). Huang *et al.* used trans‐ mission electron microscopy to measure the elastic bending modulus of ZnO nano-wires via electric-field-induced resonant excitation and obtained a mean elastic modulus of ~58 Gpa (Huang *et al.*, 2006). In particular, they measured the elastic modulus of 48.9GPa from a ZnO nanowire with diameter of 43 nm and length of 4.77 *μ*m, which is similar to that (*d*=42.3 nm, T=4.41 *μ*m) of sample 1. If choosing *E*=50 Gpa while *n* = 224 nm based on the nanowires den‐

For sample 1 the water's parallel surface tension *Fw*= 8.33×10-10 (N) is about 22 times more than *Felastic*. Therefore, the surface tension of deionized water is big enough to let two or more nanowires bundling together to form the reticulate papillary nodes structure. Note that al‐ though the parallel surface tension of n-Hexane is *F*h = 0, after the substrate was immersed in the n-Hexane, a little bending of the nanowires of sample 2 still happened (see Fig.2d), due

As shown in Fig.4b, the angle *α* (the angle between the Y-axis and the tangent at the nano‐ wire tip) is used for defining any intermediate state during the bending process. The rela‐ tionship between the *F1* (required to overcome the bending-induced elastic energy, for

external force in X-axis, the bending forces experienced by the two nanowires must be equal. To simply obtain the relation between the bending angle and the nanowires diameter, we assumed that the parallel surface tension of water, Young's modulus and the length of ZnO nanowires are constant as *Fw*=8.33×10-10N, *E*=50 GPa and *T*=4.41 *μ*m, respectively. The curve in Fig.4c is the bending angle as a function of the nanowires diameter of samples 1, 3-6. The bending angle decreases sharply as the diameter increases. The bending will be faded out as

surface tension has to be larger than the bending force for *n*/2 displacement.

sity in Table 1 to calculate the elastic bending force, we therefore obtain

to that the residual water in solution cannot be thoroughly removed.

water: *F1*= *Fw*) and the bending angle is *α*=*F1T2*

the diameter increases enough.

*d* <sup>4</sup> / 144. To make these two symmetric wires a contact at *x*=*n*/2, the parallel

( ) <sup>11</sup> 3.77 10 *elastic F N* - = ´ (6)

/2*EI* (Hibbeler, 1997). Since there is no other

pressed as *I* =5 <sup>3</sup>

84 Nanowires - Recent Advances

*<sup>T</sup>* (5)

It is also found in Table 1 that the nanowires' density decreased as the zinc acetate dehy‐ drate concentration increased, due to the nanowires fusing easily together in their growth process. However, as the concentration is very low, the density of nanowires is also lower which is due to the fact that some seed particles did not grow into the longer nanowires, just like sample 1.

**Figure 5.** The surface FESEM images of ZnO nanowires of (a) sample 1, (b) sample 3, (c) sample 4, (d) sample 5, and (e) sample 6. The insets show images of the droplets that formed on the surface of the different samples during the con‐ tact angle measurement. The top inset of (a) shows the selected area electron diffraction pattern of the ZnO-RPNF that confirms the single crystalline ZnO nanowire. Variation of the ZnO nanowire's diameter and water contact angle measured on the surface of different samples (sample 1, samples 3 to 6) as a function of the concentration of hexame‐ thylenetetramine and zinc acetate dehydrate solution is shown in (f), with the variation range of the nanowires diam‐ eter indicated by the error bars.

The general trend was that as ZnO nanowires had a larger aspect ratio and smaller diame‐ ter, they were more susceptible to assemble reticulate film with papilla structure. For exam‐ ple, for any given diameter and density, nanowires would not overcome their own elastic modulus and assemble, but assembling by liquid surface tension would become prominent as nanowires grew longer and thinner. The assembling degree of nanowires in Fig.5 was consistent with this general trend. The aspect ratio and diameter in Fig.5a-e were 104:1 and 42.3nm, 57:1 and 88.8nm, 47:1 and 110.3nm, 17:1 and 246.3nm, and 6:1 and 698.5nm, respec‐ tively. As expected the degree of assembling was largest in Fig.5a and smallest in Fig.5e.The diameter *d* variation as functions of the concentration *c* is shown in Fig.5f. The function is expressed as *d*=4440*c* (nm) when the concentration is lower than 0.04M/L, indicating a growth mode for single nano-wire. However, as the concentration increases to 0.04M/L and 0.05M/L, the variation function changes to d=45220c*-*1562.5(nm). The slope of which is 10 times more than the former.This could be attributed to the merging between the nanowires. As the solution concentration increases, the diameters of the nanowires increase while the distance between them decreases. When the diameters reached a certain value, the spaces between the nanowires became zero, resulting in the amalgamation of nanowires.

#### **3.2. Superhydrophobicity**

Wettability of a solid surface has attracted much attention during the past decades, due to its broad technological applications and fundamental researches. A direct expression of the wettability is the contact angle of a water droplet on the surface. Surfaces with very high water contact angles particularly greater than 150° are usually called superhydrophobic sur‐ faces, which have been extended to more applications, such as self-cleaning materials, antifog, anti snow, fluid microchips and microreactors (Richard *et al.*, 2002; Erbil *et al.*, 2003; Jiang *et al.*, 2004; Chen *et al.*, 2006).

The obtained ZnO-RPNF was superhydrophobic (see the insert of Fig.2a). The water contact angle is measured to be 170±1o , and the water volume is 10*μ*L. The contact angle hysteresis, which is the difference between the advancing and receding angles, is measured to be less than 2o . The water droplets roll off of the substrate at a sliding angle of less than 2o . The wa‐ ter contact angle on the plane ZnO thin film modified via heptadecafluorodecyltrimethoxysilane (HTMS) was about 114±1o , which is near to the maximum 119o (Nishino *et al.*, 1999). The largest variance in the contact angle is due to the surface roughness. These data suggest that the predominant contribution to the large contact angle is the reticulate papillary nodes structure. Furthermore, the obtained ZnO-RPNF maintains superhydrophobic properties even after ambient environment for three months, showing their favorable stability. This is very important for extending the practical applications of ZnO materials.

The basic effect of the surface structure and chemistry on superhydrophobicity has been elu‐ cidated by Wenzel (Wenzel, 1936) and Cassie (Cassie and Baxter, 1944) models. Wenzel's model hypothesizes that the liquid completely penetrates into the troughs of the surface tex‐ ture. Cassie's model is based on the assumption that the liquid drop doesn't fill the troughs of the rough surface but sits on a composite surface composed of solid material and air. Both the Wenzel and Cassie models have put emphasis on the geometrical structure of a solid surface as an important factor in determining the hydrophobicity. The trapping of air in the surface troughs causes an increase in the water contact angle according to Cassie-Baxter'seq‐ uation (Jiang *et al.*, 2004).

$$\cos\theta^\* = f\_1\cos\theta\_1 + f\_2\cos\theta\_2\tag{7}$$

Synthesis, Superhydrophobicity, Enhanced Photoluminescence and Gas Sensing Properties of ZnO Nanowires http://dx.doi.org/10.5772/52586 87

$$f\_1 + f\_2 = 1 \tag{8}$$

where *θ*\* is the contact angle on HTMS coating the rough surface of reticulate ZnO film with papillary nodes, *θ*1 and *θ*2are the contact angles on HTMS coating the plane ZnO film and air, respectively; *f*1 and *f*2are the fractions of the papillary nodes' end surface and of the trap‐ ped air on reticulate ZnO film. Here, *θ*2 is 180°. Substituting *θ*2 and (8) into equation (7) will lead to:

42.3nm, 57:1 and 88.8nm, 47:1 and 110.3nm, 17:1 and 246.3nm, and 6:1 and 698.5nm, respec‐ tively. As expected the degree of assembling was largest in Fig.5a and smallest in Fig.5e.The diameter *d* variation as functions of the concentration *c* is shown in Fig.5f. The function is expressed as *d*=4440*c* (nm) when the concentration is lower than 0.04M/L, indicating a growth mode for single nano-wire. However, as the concentration increases to 0.04M/L and 0.05M/L, the variation function changes to d=45220c*-*1562.5(nm). The slope of which is 10 times more than the former.This could be attributed to the merging between the nanowires. As the solution concentration increases, the diameters of the nanowires increase while the distance between them decreases. When the diameters reached a certain value, the spaces

between the nanowires became zero, resulting in the amalgamation of nanowires.

Wettability of a solid surface has attracted much attention during the past decades, due to its broad technological applications and fundamental researches. A direct expression of the wettability is the contact angle of a water droplet on the surface. Surfaces with very high water contact angles particularly greater than 150° are usually called superhydrophobic sur‐ faces, which have been extended to more applications, such as self-cleaning materials, antifog, anti snow, fluid microchips and microreactors (Richard *et al.*, 2002; Erbil *et al.*, 2003;

The obtained ZnO-RPNF was superhydrophobic (see the insert of Fig.2a). The water contact

which is the difference between the advancing and receding angles, is measured to be less

ter contact angle on the plane ZnO thin film modified via heptadecafluorodecyltrimethoxy-

The largest variance in the contact angle is due to the surface roughness. These data suggest that the predominant contribution to the large contact angle is the reticulate papillary nodes structure. Furthermore, the obtained ZnO-RPNF maintains superhydrophobic properties even after ambient environment for three months, showing their favorable stability. This is

The basic effect of the surface structure and chemistry on superhydrophobicity has been elu‐ cidated by Wenzel (Wenzel, 1936) and Cassie (Cassie and Baxter, 1944) models. Wenzel's model hypothesizes that the liquid completely penetrates into the troughs of the surface tex‐ ture. Cassie's model is based on the assumption that the liquid drop doesn't fill the troughs of the rough surface but sits on a composite surface composed of solid material and air. Both the Wenzel and Cassie models have put emphasis on the geometrical structure of a solid surface as an important factor in determining the hydrophobicity. The trapping of air in the surface troughs causes an increase in the water contact angle according to Cassie-Baxter'seq‐

very important for extending the practical applications of ZnO materials.

\*

q

1 12 2 cos cos cos

qq

= + *f f* (7)

, which is near to the maximum 119o

. The water droplets roll off of the substrate at a sliding angle of less than 2o

, and the water volume is 10*μ*L. The contact angle hysteresis,

. The wa‐

(Nishino *et al.*, 1999).

**3.2. Superhydrophobicity**

86 Nanowires - Recent Advances

Jiang *et al.*, 2004; Chen *et al.*, 2006).

angle is measured to be 170±1o

silane (HTMS) was about 114±1o

uation (Jiang *et al.*, 2004).

than 2o

$$\cos \theta^\* = f\_s \cos \theta\_\ell + f\_s - 1 \tag{9}$$

here *f*s is the fractions of the papillary nodes end surface, and *θ*eis the contact angle on HTMS coating ZnO plane film (*θ*e=114o ).

We notice from eq. (8) that if the reticulate ZnO film with papillary nodes has a very low *f*s, a superhydrophobic thin film with a high water contact angle could be fabricated. The value of *f*<sup>s</sup> is estimated by the SEM experiments. As mentioned above, the average diameter of one nanowire is 42.3 nm, and the growth density *D* is about (2.0±0.5)×109 per square centimeter by a mass of statistic.Due to the top of ZnO nanowire being hexagonal, the nanowire diame‐ ter *d* and the growth density *D* is 42.3nm and 2.0×109 per square centimeter, respectively. The *f*s is obtained as:

$$f\_s = (2 \times \frac{\sqrt{3}}{12} \, d^2 + \frac{d^2}{2 \cos 30^\circ}) D = \frac{\sqrt{3}}{2} \, d^2 D = 0.031 \tag{10}$$

Substituting (10) into eq. (9) will lead to:*θ*\* =169o . The calculated value is in good agreement with the experimental datum 170o ±1o of sample 1. However, the experimental data of sam‐ ple 1\* and sample 2 are 167o and 165o (see Table 1 and Fig.2), respectively. We know from the Cassie model that the air trapped in the surface structure plays an important role for the surface with a larger contact angle. The reticulate structure may serve as a means of trap‐ ping sufficient air for larger contact angles to be exhibited. A reticulate papillary node on sample 1 is the assembly of about ten nanowires on average, and the wires bending with a spiral line structure before bundling together at the wires' top. This makes the nano-papil‐ lary nodes distribute with circularity and forms a larger column-like micro space (see Fig. 2b). This micro/nano complex structure is closed to the bionic lotus-leaf-structure and much in line with the Cassie model. Nevertheless, the papillary nodes of sample 1\* are bended simply by seven nanowires; there is no larger mirco space as that in sample 1. Therefore its superhydrophobicity is lower than that of sample 1. As a comparison, the contact angle of sample 2 is the lowest, because it has fewer or no reticulate papillary nodes.

The contact angle on the surface of different samples formed by nanowires tends to decrease as the surface density of nanowires and the width between them decrease (see Table 1 and Fig.5). As represented in Fig.5f, the contact angle decreases from 170o (sample 1) to 154o (sample 5). This result is consistent with that of the superhydrophobicity behavior deter‐ mined by the density of the nanowires and the space between them (Borras *et al.*, 2008). Note that although the nanowires fused together, the contact angle on the surface of sample 6 (Fig.5e-f) is not the minimum, due to the protuberance on the surface.

**Figure 6.** Room-temperature photoluminescence spectra (a), and XRD patterns (b) of ZnO as-grown and annealed at 550 °C for 1h.

## **4. Photoluminescence of ZnO nanowires via hydrothermal method**

ZnO exhibited a wide band gap at room temperature with a large exciton binding energy, which is suitable for effective ultraviolet emission. However, due to the poor crystalline quality of the nano materials, *i.e.* high density of structure defects, the ultraviolet emission of nano-scaled ZnO is liable to be quenched and only defect emission in visible region is de‐ tected (Kong *et al.*, 2001). This deficiency hinders process for the application of ZnO in opto‐ electronic and lasing devices. Therefore, improving the crystal quality of ZnO by synthetic processing and realizing ultraviolet emission and lasing are still major challenges.

The study of photoluminescence is a favorable way to evaluate both ZnO optical properties and its structural defects. Normally, ZnO nanostructures can exhibit ultraviolet emission re‐ sulting from the recombination of free excitons and visible emission due to some structural defects(Djurisic and Leung, 2006). Fig.6a gives the room-temperature photoluminescence spectra of ZnO nanowires growing at low temperature of 50 °C by hydrothermal method with as-grown and annealed, in which only ultraviolet emission peaks centered at about 378.5 nm dominate, while the well-known broader emission situated in the blue-red part of the visible spectrum could not be observed (Wang *et al.*, 2007). This result can be attributed to the fine crystallization of the ZnO nanowires. The ultraviolet emission intensity of the asgrown nanowires is better than the annealed one, which indicates the annealing process de‐ creased the crystal quality of ZnO just as the XRD showed Fig.6b. Therefore, ZnO nanowires growing at lower temperature by hydrothermal method have huge potential to fabricate high crystalline quality and good ultraviolet luminescence devices.

(sample 5). This result is consistent with that of the superhydrophobicity behavior deter‐ mined by the density of the nanowires and the space between them (Borras *et al.*, 2008). Note that although the nanowires fused together, the contact angle on the surface of sample

**Figure 6.** Room-temperature photoluminescence spectra (a), and XRD patterns (b) of ZnO as-grown and annealed at

ZnO exhibited a wide band gap at room temperature with a large exciton binding energy, which is suitable for effective ultraviolet emission. However, due to the poor crystalline quality of the nano materials, *i.e.* high density of structure defects, the ultraviolet emission of nano-scaled ZnO is liable to be quenched and only defect emission in visible region is de‐ tected (Kong *et al.*, 2001). This deficiency hinders process for the application of ZnO in opto‐ electronic and lasing devices. Therefore, improving the crystal quality of ZnO by synthetic

The study of photoluminescence is a favorable way to evaluate both ZnO optical properties and its structural defects. Normally, ZnO nanostructures can exhibit ultraviolet emission re‐ sulting from the recombination of free excitons and visible emission due to some structural defects(Djurisic and Leung, 2006). Fig.6a gives the room-temperature photoluminescence spectra of ZnO nanowires growing at low temperature of 50 °C by hydrothermal method with as-grown and annealed, in which only ultraviolet emission peaks centered at about 378.5 nm dominate, while the well-known broader emission situated in the blue-red part of the visible spectrum could not be observed (Wang *et al.*, 2007). This result can be attributed to the fine crystallization of the ZnO nanowires. The ultraviolet emission intensity of the asgrown nanowires is better than the annealed one, which indicates the annealing process de‐

**4. Photoluminescence of ZnO nanowires via hydrothermal method**

processing and realizing ultraviolet emission and lasing are still major challenges.

6 (Fig.5e-f) is not the minimum, due to the protuberance on the surface.

550 °C for 1h.

88 Nanowires - Recent Advances

The ultraviolet emission intensity of ZnO nanowires is very important in lasing devices. As mentioned above, the intensity can be enhanced by improving crystalline quality of ZnO nanowires. In addition to amending their own quality, surface plasmons or localized surface plasmons of various noble metals also can be used to realize ZnO nanowires pho‐ toluminescence enhancement. You's group sputtered ZnO film on Si (001) substrate which had already been coated with 100 nm Ag film previously, and the ultraviolet emission of the composite is found to be greatly enhanced (You *et al.*, 2007). Xiao *et al.* have also found enhancement of near-band-emission of ZnO by coating a layer of Ag implanted sili‐ ca film(Xiao *et al.*, 2010).

**Figure 7.** SEM images of (a) ZnO nanowires non-coating and (b) ZnO nanowires coated with separated Ag nanoparti‐ cles.

ZnO nanowires grown by the two-step method as mentioned above and annealed in air at 400°C for 1 hr in order to eliminate the defects in ZnO. Then, 40 nm Ag deposited on the ZnO nanowire arrays surface by the radio-frequency magnetron sputtering method at room temperature. Fig.7 shows SEM images of ZnO nanowires non-coating and coated with Ag nanoparticles on the surface. The nanowires, dispersed, lay on the substrate, 4-6 *μ*m in length and 400-600 nm in diameter, and formed a perfect hexagonal structure. It is clear that the nanowires in Fig.7 show almost no difference-either in size or density. There is apparent‐ ly a layer of 40-50 nm Ag nanoparticles on the nanowires in the inset image of Fig.7b, and the roughness of the surface of the metal is quite high.

The photoluminescence spectra of the ZnO nanowires non-coating and coated Ag nanoparti‐ cles are shown in Fig.8. The spectrum of the ZnO nanowires is shown as curve *a*. The peak posi‐ tion at 381.6 nm with 87 meV full width at half maximum, this was attributed to FX-1LO (Shan *et al.*, 2005; Hamby *et al.*, 2003). The emission for the visible light mostly caused by the defect in ZnO is very week, indicating good quality of the ZnO nanowires. Accordingly, the ZnO nano‐ wires are one of the good optional materials for ultraviolet lasing devices. Curve *b* in Fig.8 is the spectrum of ZnO nanowires coated with 40 nm Ag nanoparticles, showing a greatly enhanced photoemission. The ratio enhancement reaches 6.6 compared with pure ZnO nanowires. Therefore, the photoluminescence of the Ag/ZnO nanowires is apparently stronger than that of ZnO nanowires non-coating because of the intensive coupling between ZnO nanowires and lo‐ calized surface plasmons of Ag nanoparticles (Lü *et al.*, 2008), which informs us of a good orien‐ tation for the composite of metal and semiconductor as a laser source material.

**Figure 8.** Photoluminescence spectra of (a) ZnO nanowires non-coating and (b) ZnO nanowires coated with 40 nm Ag nanoparticles.

## **5. Electrical and sensing properties of ZnO nanofibers via electrospinning**

Besides hydrothermal self-assembly method, ZnO nanowires were also fabricated by calci‐ nation of electrospun poly(vinyl alcohol)/zinc acetate precursor wires at 600 °C for 5 h, as shown in Fig. 1. Raman spectroscopy and XRD were used to confirm the formation of a pure ZnO phase in the nanowires after calcination treatment. Fig.9a shows the clear phonon modes of the wires calcined at 600 °C: the two phonon modes at 439 and 582 cm-1 could be assigned to E2 (high) and A1 (LO), separately; the phonon mode at 325 cm-1 is a second-order phonon, which generally originates from the zone-boundary phonons of 2E2 (low). This Raman spectrum is typical for pure, polycrystalline ZnO. Fig.9b gives the XRD pattern for the as-spun fibers after calcinations at 600 °C for 5 h. A clear diffraction pattern is shown, and nine reflection peaks appear at 2θ=31.9° (100), 34.6° (002), 36.5° (101), 47.7° (102), 56.8° (110), 63.1° (103), 66.5° (200), 68.1° (112), and 69.2 (201). It agrees well with the JCPDS card, No. 89-1397 for pure ZnO, confirming the as-prepared product is pure ZnO nanowires after calcination at that temperature.

Synthesis, Superhydrophobicity, Enhanced Photoluminescence and Gas Sensing Properties of ZnO Nanowires http://dx.doi.org/10.5772/52586 91

ZnO is very week, indicating good quality of the ZnO nanowires. Accordingly, the ZnO nano‐ wires are one of the good optional materials for ultraviolet lasing devices. Curve *b* in Fig.8 is the spectrum of ZnO nanowires coated with 40 nm Ag nanoparticles, showing a greatly enhanced photoemission. The ratio enhancement reaches 6.6 compared with pure ZnO nanowires. Therefore, the photoluminescence of the Ag/ZnO nanowires is apparently stronger than that of ZnO nanowires non-coating because of the intensive coupling between ZnO nanowires and lo‐ calized surface plasmons of Ag nanoparticles (Lü *et al.*, 2008), which informs us of a good orien‐

**Figure 8.** Photoluminescence spectra of (a) ZnO nanowires non-coating and (b) ZnO nanowires coated with 40 nm Ag

Besides hydrothermal self-assembly method, ZnO nanowires were also fabricated by calci‐ nation of electrospun poly(vinyl alcohol)/zinc acetate precursor wires at 600 °C for 5 h, as shown in Fig. 1. Raman spectroscopy and XRD were used to confirm the formation of a pure ZnO phase in the nanowires after calcination treatment. Fig.9a shows the clear phonon modes of the wires calcined at 600 °C: the two phonon modes at 439 and 582 cm-1 could be assigned to E2 (high) and A1 (LO), separately; the phonon mode at 325 cm-1 is a second-order phonon, which generally originates from the zone-boundary phonons of 2E2 (low). This Raman spectrum is typical for pure, polycrystalline ZnO. Fig.9b gives the XRD pattern for the as-spun fibers after calcinations at 600 °C for 5 h. A clear diffraction pattern is shown, and nine reflection peaks appear at 2θ=31.9° (100), 34.6° (002), 36.5° (101), 47.7° (102), 56.8° (110), 63.1° (103), 66.5° (200), 68.1° (112), and 69.2 (201). It agrees well with the JCPDS card, No. 89-1397 for pure ZnO, confirming the as-prepared product is pure ZnO nanowires after

**5. Electrical and sensing properties of ZnO nanofibers via**

nanoparticles.

**electrospinning**

90 Nanowires - Recent Advances

calcination at that temperature.

tation for the composite of metal and semiconductor as a laser source material.

**Figure 9.** a) Raman spectra of as-spun PVA/zinc acetate nanowires before and after calcinations at 600 °C for 5 h in atmosphere. (b) XRD pattern of the ZnO nanowires.

To fabricate ZnO nanowire-based field-effect transistor or gas sensor device, SiO2/Si chip with pre-patterned Au electrode arrays via traditional e-beam lithography technology was used as collector to collect PVA/zinc acetate nanowires. The collecting time was about 5 s. The chip was then calcined at 600 °C for 4 h to obtain ZnO nanowires device. As shown in Fig. 10a, the square Au electrodes have a side length of 50 *μ*m, and a long ZnO wire is clear‐ ly deposited across the electrodes. In another device, PVA/zinc acetate nanowire yarns were deposited on SiO2/Si chip by a two-parallel-electrode method or gapping method (Tan *et al.*, 2008). After calcination of the chip at 600 °C for 4 h, two electrodes made of silver paste were attached onto the ZnO nanowire yarns, as shown in Fig. 10b.

**Figure 10.** a) Optical image of an individual electrospun ZnO nanowire measured by two-probe method; (b) Optical image of a bundle of electrospun ZnO nanowires and attached Ag paste electrodes.

Current-voltage (I-V) curves were performed on the single ZnO nanowire device by using two-metal-microprobe testing platform. Fig.10a shows the optical picture that two metal mi‐ croprobes were pressed directly on the square Au electrodes. The scanning voltage was ap‐ plied from -20 to 20 V with a step of 0.1 V. At the same time the current corresponding to each applied voltage was measured. Fig.11a shows the Ids-Vds characteristic curves of the sin‐ gle ZnO nanowire at dark and under illustration. It is evident that the electrospun ZnO nanowire has a very large resistance (~1.4x1010Ω) at dark: under bias voltage of 20 V, the current is only about 1.4 nA. But the current increases sharply under illustration. The cur‐ rent can reach 19.8 nA (20 V bias) when the color temperature of illustration is 2900 K. In addition, the Si substrate was used as a back gate to measure gate effect of this ZnO nano‐ wire device. As shown in Fig. 11b, higher source-drain currents were measured when posi‐ tive gate voltages Vg were applied. It indicates that the electrospun ZnO nanowire is an intrinsic *n*-type semiconductor. The above results are in accordance with that obtained from other ZnO nanowire field-effect transistor devices fabricated by the chemical vapour deposi‐ tion or hydrothermal self-assembly method.

**Figure 11.** (a) Current-voltage (Ids-Vds) characteristic curves of an individual ZnO nanowire (as shown Fig.10a) at dark and under illustration. The illustration intensity is represented by color temperature of 2500, 2700and 2900 K. (b) Source-drain current (Ids) versus gate voltage (Vg) at Vds=8 V for the same ZnO nanowire device.

**Figure 12.** Response of the gas sensor based on (a) an individual ZnO nanowire, and (b) a bundle of ZnO nanowires towards 500 or 250 ppm CO as a function of time at room temperature. The applied voltage is 20 V.

In addition to transistors and photodetectors, ZnO has been proved to be a kind of highly sensitive material for the flammable or toxic gas detection, such as H2, CO, NO, or ethanol (Zeng *et al.*, 2009; Krishnakumar *et al.*, 2009). To improve the sensing performance of the gas sensors, usually the ZnO nanowires were doped with metal or metal-oxide to form surface depletion layer, and thus enhance the sensing properties (Li *et al.*, 2012; Liu *et al.*, 2009; Pan *et al.*, 2001). Moreover, a heater was necessary for most of the gas sensors to work at high temperatures of about 200-400°C, which leads to the sensor complex and high-cost. In this work, since the electrospun ZnO nanowires exhibit continuous and compact-grains struc‐ tures and thus have a high surface-to-volume ratio, gas sensing measurements of nanowire devices based on one single ZnO nanowire and one bundle of ZnO nanowires were carried out. As shown in Fig. 12, the ZnO nanowires exhibit rapid, apparent response when ex‐ posed to 500ppm or 250 ppm CO even at room temperature. Since the gas response can be defined as the ratio: S≈|Rair-Rgas|/Rair, where Rair and Rgas are the resistance of the sensor in dry air and in the test gas, separately, Fig. 12 indicates that the gas response can reach 500 % in 60 min towards 500 ppm CO or 26% in 5 min towards 250 ppm CO.

The ZnO carbon monoxide gas sensing can be explained by the chemical reaction between the active gas and oxygen ion adsorption on the surface of ZnO. In air atmosphere, oxygen molecules are adsorbed onto the surface of the ZnO sensor to form O or O2- ions by attract‐ ing electrons from the conduction band of the ZnO. Under carbon monoxide atmosphere, the CO gas reacts with oxygen ion molecule on the surface and gives back electrons into the conduction band, thereby lowering the resistance of ZnO sensors. ZnO nanostructures ex‐ hibit sensor response higher than the bulk; this can be simply explained by the effect of the surface-to-volume ratio. In the present case, the fast and large response to CO at room tem‐ perature can be ascribed to small nanocrystal size, high surface area and high porosity of the resultant ZnO nanowires.

### **6. Summary**

Current-voltage (I-V) curves were performed on the single ZnO nanowire device by using two-metal-microprobe testing platform. Fig.10a shows the optical picture that two metal mi‐ croprobes were pressed directly on the square Au electrodes. The scanning voltage was ap‐ plied from -20 to 20 V with a step of 0.1 V. At the same time the current corresponding to each applied voltage was measured. Fig.11a shows the Ids-Vds characteristic curves of the sin‐ gle ZnO nanowire at dark and under illustration. It is evident that the electrospun ZnO nanowire has a very large resistance (~1.4x1010Ω) at dark: under bias voltage of 20 V, the current is only about 1.4 nA. But the current increases sharply under illustration. The cur‐ rent can reach 19.8 nA (20 V bias) when the color temperature of illustration is 2900 K. In addition, the Si substrate was used as a back gate to measure gate effect of this ZnO nano‐ wire device. As shown in Fig. 11b, higher source-drain currents were measured when posi‐ tive gate voltages Vg were applied. It indicates that the electrospun ZnO nanowire is an intrinsic *n*-type semiconductor. The above results are in accordance with that obtained from other ZnO nanowire field-effect transistor devices fabricated by the chemical vapour deposi‐

**Figure 11.** (a) Current-voltage (Ids-Vds) characteristic curves of an individual ZnO nanowire (as shown Fig.10a) at dark and under illustration. The illustration intensity is represented by color temperature of 2500, 2700and 2900 K. (b)

**Figure 12.** Response of the gas sensor based on (a) an individual ZnO nanowire, and (b) a bundle of ZnO nanowires

towards 500 or 250 ppm CO as a function of time at room temperature. The applied voltage is 20 V.

Source-drain current (Ids) versus gate voltage (Vg) at Vds=8 V for the same ZnO nanowire device.

tion or hydrothermal self-assembly method.

92 Nanowires - Recent Advances

ZnO nanostructures have drawn great attention for their wide applications in electronic, op‐ tical, optoelectronic, sensing, catalysis, hydrophobic surface modification, and energy har‐ vesting devices (Xu and Wang, 2011)*.* In this chapter, our recent results in synthesis, physical properties and sensor applications of ZnO nanowires are presented. Crystalline ZnO nanowires and polycrystalline ZnO nanofibers are prepared by simple hydrothermal self-assembly method and electrospinning, respectively. The surfaces of ZnO nanowires thin film after surface modification are superhydrophobic with a contact angle of 170° and a slid‐ ing angle of 2°. As one of the best optional materials for ultra-violet laser devices, ZnO nano‐ wires grown by hydrothermal method have good photoluminescence properties and photoluminescence enhancement at 387.6 nm by coating a layer of Ag nanoparticles. ZnO nanowires with porous fibrous morphology fabricated via electrospinning have specific sur‐ face approximately one to two orders of the magnitude larger than flat films, making them excellent candidates for potential applications in gas sensors. For example, the granular ZnO nanowires exhibit fast and large response to CO even at room temperature. Further studies are in progress to improve the performance of these nanodevices.

## **Acknowledgments**

This work was supported by the National Natural Science Foundation of China (Grant Nos.: 11074138, 50872129 and 11004114), the Taishan Scholars Program of Shandong Province, the Natural Science Foundation of Shandong Province, China for Distinguished Young Scholars (Grant No.: JQ201103), and the National Key Basic Research Development Program of China (973 special preliminary study plan) (Grant No.: 2012CB722705).

## **Author details**

M.G. Gong1,2, Y.Z. Long1,3,4, X.L. Xu2 , H.D. Zhang1 and B. Sun1

1 College of Physics, Qingdao University, People's Republic of China

2 Department of Physics, University of Science and Technology of China, People's Republic of China

3 Key Laboratory of Photonics Materials and Technology in Universities of Shandong, Qingdao University, People's Republic of China

4 State Key Laboratory Cultivation Base of New Fiber Materials and Modern Textile, Qingdao University, People's Republic of China

### **References**


[4] Chen, Y. B.; Ambade, A. V.; Vutukuri, D. R.; Thayumanavan, S. (2006). Self-assembly of facially amphiphilic dendrimers on surfaces. *J. Am. chem. Soc.*, Vol. 128(46), Nov. 2006, pp. 14760-14761, ISSN: 0002-7863.

nanowires exhibit fast and large response to CO even at room temperature. Further studies

This work was supported by the National Natural Science Foundation of China (Grant Nos.: 11074138, 50872129 and 11004114), the Taishan Scholars Program of Shandong Province, the Natural Science Foundation of Shandong Province, China for Distinguished Young Scholars (Grant No.: JQ201103), and the National Key Basic Research Development Program of China

, H.D. Zhang1

2 Department of Physics, University of Science and Technology of China, People's Republic

3 Key Laboratory of Photonics Materials and Technology in Universities of Shandong,

4 State Key Laboratory Cultivation Base of New Fiber Materials and Modern Textile,

[1] Borras, A.; Barranco, A.; Gonzalez-Elipe, A. R. (2008). Reversible superhydrophobic to superhydrophilic conversion of Ag@TiO2 composite nanofiber surfaces. *Langmuir*,

[2] Cao, B. Q.; Lorenz, M.; Rahm, A.; von Wenckstern, H.; Czekalla, C.; Lenzner, J.; Benn‐ dorf, G.; Grundmann, M. (2007). Phosphorus acceptor doped ZnO nanowires pre‐ pared by pulsed-laser deposition. *Nanotechnology*, Vol. 18(45), Oct. 2007, pp.

[3] Cassie, A. B. D.; Baxter, S. (1944). Wettability of porous surfaces. *Trans. Faraday Soc.*,

and B. Sun1

are in progress to improve the performance of these nanodevices.

(973 special preliminary study plan) (Grant No.: 2012CB722705).

1 College of Physics, Qingdao University, People's Republic of China

Vol. 24, June 2008, pp. 8021-8026, ISSN: 1520-5827.

Vol. 40, June 1944, pp. 546-551, ISSN: 0014-7672.

**Acknowledgments**

94 Nanowires - Recent Advances

**Author details**

of China

**References**

M.G. Gong1,2, Y.Z. Long1,3,4, X.L. Xu2

Qingdao University, People's Republic of China

Qingdao University, People's Republic of China

455707-455711, ISSN: 1361-6528.


Glass templates. *Journal of the Electrochemical Society*, Vol. 152(5), 2005, pp. G378-G381, ISSN: 0013-4651.


[28] Landau, O.; Rothschild, A.; Zussman, E. (2009). Processing-microstructure-properties correlation of ultrasensitive gas sensors produced by electrospinning. *Chem. Mater.*, Vol. 21(1), Jan. 2009, pp. 9-11, ISSN: 0897-4756.

Glass templates. *Journal of the Electrochemical Society*, Vol. 152(5), 2005, pp. G378-G381,

[17] Huang, M. H.; Wu, Y. Y.; Feick, H.; Tran, N.; Weber, E.; Yang, P. D. (2001). Catalytic growth of znic oxide nanowires by vapor transport. *Advanced Materials*, Vol. 13(2),

[18] Huang, Z. M.; Zhang, Y. Z.; Kotaki, M.; Ramakrishna, S. (2003). A review on polymer nanofibers by electrospinning and their applications in nanocomposites. *Compos. Sci.*

[19] Huang, Y. H.; Bai, X. D.; Zhang, Y. (2006). In situ mechanical properties of individual ZnO nanowires and the mass measurement of nanoparticles. *J. Phys.: Condens. Mat‐*

[20] Jeong, J. S.; Lee, J. Y. (2010). Substrate-facilitated nanoparticles sintering and compo‐ nent interconnection procedure. *Nanotechnology*, Vol. 21(47), Oct. 2010, p. 475204,

[21] Jiang, L.; Zhao, Y.; Zhai, J. (2004). A lotus-leaf-like superhydrophobic surface: a po‐ rous microsphere/nanofiber composite film prepared by electrohydrodynamics. *An‐*

[22] Kim, I. D.; Rothschild, A.; Lee, B. H.; Kim, D. Y.; Jo, S. M.; Tuller, H. L. (2006). Ultra‐ sensitive chemiresistors based on electrospun TiO2 nanofibers. *Nano. Lett.*, Vol. 6(9),

[23] Ko, S. H.; Lee, D.; Kang, H. W.; Nam, K. H.; Yeo, J. Y.; Hong, S. J.; Grigoropoulos, C. P.; Sung, H. J. (2011). Nanoforest of hydrothermally grown hierarchical ZnO nano‐ wires for a high efficiency dye-sensitized solar cells. *Nano Letters*, Vol. 11(2), Jan.

[24] Kong, Y. C.; Yu, D. P.; Zhang, B.; Fang, W.; Feng, S. Q. (2001). Ultraviolet-emitting ZnO nanowires synthesized by a physical vapor deposition approach. *Appl. Phys.*

[25] Kong, X. H.; Li, Y. D. (2003). Temperature-dependent regulation of antisense activity using a DNA/poly(N-isopropylacrylamide) conjugate. *Chemistry Letters*, Vol. 32(11),

[26] Kusinski, G. J.; Jokisaari, J. R.; Noriega, R.; Goris, L.; Donovan, M.; Salleo, A. (2010). Transmission electron microscopy of solution-processe, intrinsic and Al-doped ZnO nanowires for transparent electrode fabrication. *Journal of microscopy*, Vol. 237(3),

[27] Krishnakumar, T.; Jayaprakash, R.; Pinna, N.; Donato, N.; Bonavita, A.; Micali, G.; Neri, G. (2009). CO gas sensing of ZnO nanostructures synthesized by an assisted mi‐ crowave wet chemical route. *Sensor and Actuators B: Chemical*, Vol. 143(1), Dec. 2009,

*gew. Chem. Int. Ed.*, Vol. 43, 2004, pp. 4338-4341, ISSN: 1521-3773.

ISSN: 0013-4651.

96 Nanowires - Recent Advances

ISSN: 1361-6528.

Jan. 2001, pp. 113-116, ISSN: 1521-4095.

2006, pp. 2009-2013, ISSN: 1530-6984.

2011, pp. 666-671, ISSN: 1530-6984.

2003, pp. 1062-1063, ISSN: 1348-0715.

Mar. 2010, pp. 443-449, ISSN: 1365-2818.

pp. 198–204, ISSN: 092-4005.

*Lett.*, Vol. 78, Nov. 2001, p. 407, ISSN: 1616-3028.

*Technol.* Vol. 63, 2003; pp. 2223-2253, ISSN: 0266-3538.

*ter*, Vol. 18, Mar. 2006, p. L179, ISSN: 1361-648X.


[53] Wang, X. D.; Summers, C. J.; Wang, Z. L. (2004). Large-scale hexagonal-patterned growth of aligned ZnO nanorods for nano-optoelectronics and nanosensor arrays. *Nano Letters*, Vol. 4(3), Jan. 2004, pp. 423-426, ISSN: 1530-6984.

[40] Nishino, T.; Meguro, M.; Nakamae, K.; Matsushita, M.; Ueda, Y. (1999). The lowest surface free energy based on –CF3 alignment. *Langmuir*, Vol. 15(13), May 1999, p.

[41] Pan, Z. W.; Dai, Z. R.; Wang, Z. L. (2001). Nanobelts of semiconducting oxides. *Sci‐*

[42] Pfüller, C.; Brandt, O.; Flissikowski, T.; Grahn, H. T.; Ive, T.; Speck, J. S.; DenBaars, S. P. (2011). Comparison of the spectral and temporal emission characteristics of homoepitaxial and hetero-epitaxial ZnO nanowires. *Applied Physics Letters*, Vol. 98(11),

[43] Ra, Y.-W.; Choi, K.-S.; Kim, J.-H.; Hahn, Y.-B.; Im, Y.-H. (2008). Fabrication of ZnO nanowires using nanoscale spacer lithography for gas sensor. *Small*, Vol. 4(8), 2008,

[44] Richard, D.; Clanet, C.; Quere, D. (2002). Surface phenomena: contact time of a

[45] Sen, B.; Stroscio, M.; Dutta, M. (2011). Photoluminescence and Raman spectroscopy of polycrystalline ZnO nanofibers deposited by electrospinning. *Journal of Electronic*

[46] Shan, W.; Walukiewicz, W.; Ager, J. W.; Yu, K. M.; Yuan, H. B.; Xin, H. P.; Cantwell, G.; Song, J. J. (2005). Nature of room-temperature photoluminescene in ZnO. *Appl.*

[47] Shen, Y.; Hong, J.-I.; Xu, S.; Lin, S. S.; Fang, H.; Zhang, S.; Ding, Y.; Snyder, R. L.; Wang, Z. L. (2010). A general approach for fabricating arc-shaped composite nano‐ wire arrays by pulsed laser deposition. *Advanced Functional Materials*, Vol. 20(5), Mar.

[48] Song, J. H.; Wang, X. D.; Riedo, E.; Wang, Z. L. (2005). Elastic property of vertically aligned nanowires. *Nano Lett.*, Vol. 5(10), Sep. 2005, pp. 1954-1958, ISSN: 1616-3028.

[49] Sui, X. M.; Shao, C. L.; Liu, Y. C. (2005). White-light emission of polyvinyl alcohol/ZnO hybrid nanofibers prepared by electrospinning. *Applied Physics Letters*,

[50] Sun, B.; Long, Y. Z.; Yu, F.; Li, M. M.; Zhang, H. D.; Li, W. J.; Xu, T. X. (2012). Selfassembly of a three-dimensional fibrous polymer sponge by electrospinning. *Nano‐*

[51] Tan, J. S.; Long, Y. Z.; Li, M. M. (2008). Preparation of aligned polymer micro/nano‐ fibers by electrospinning. Chin. Phys. Lett., Vol. 25(8), Aug. 2008, pp. 3067-3070,

[52] Vayssieres, L.; Keis, K.; Lindquist, S. E.; Hagfeldt, A. (2001). Purpose-Built anisotrop‐ ic metal oxide material: 3D highly oriented microrod array of ZnO. *Journal of Physical*

*Chemistry B*, Vol. 105(17), Mar. 2001, pp. 3350-3352, ISSN: 1520-6106.

bouncing drop. *Nature*, Vol. 417, June 2002, p. 811, ISSN: 1061-4036.

*Materials*, Vol. 40(9), 2011, pp. 2015-2019, ISSN: 0361-5235.

*Phys. Lett.*, Vol. 86(19), May 2005, p. 191911, ISSN: 1616-3028.

*ence*, Vol. 291, Feb. 2001; pp. 1947–1949, ISSN: 0036-8075.

Mar. 2011, p. 113113, ISSN: 1616-3028.

pp. 1105-1109, ISSN: 1613-6829.

2010, pp. 703-707, ISSN: 1616-3028.

ISSN: 0256-307X.

Vol. 87(11), 2005, pp. 113-115, ISSN: 1616-3028.

*scale*, Vol. 4(6), 2012, pp. 2134-2137, ISSN: 2040-3364.

4321, ISSN: 1520-5827.

98 Nanowires - Recent Advances


## **Magnetoresistance of Nanowires Electrodeposited into Anodized Aluminum Oxide Nanochannels**

Takeshi Ohgai

[65] You, J. B.; Zhang, X.W.; Fan, Y. M.; Qu, S.; Chen, N. F. (2007). Surface plasmon en‐ hanced ultraviolet emission from ZnO films deposited on Ag/Si(001) by magnetron

[66] Zeng, Y.; Zhang, T.; Wang, L. J.; Kang, M. H.; Fan, H.; Wang, R.; He, Y. (2009). En‐ hanced toluene sensing characteristics of TiO2-doped flowerlike ZnO nanostructures.

[67] Zhang, H. D.; Long, Y. Z.; Li, Z. J.; Sun, B.; Sheng, C. H. (2012). Synthesis, electrical and humidity sensing properties of BaTiO3 nanofibers via electrospinning. *Advanced*

[68] Zheng, J.; Long, Y. Z.; Sun, B.; Zhang, Z. H.; Shao, F.; Zhang, H. D.; Zhang, Z. M.; Huang, J. Y. (2012). Polymer nanofibers prepared by low-voltage near-field electro‐

spinning. *Chin. Phys. B*, Vol. 21(4), April 2012, p. 048102, ISSN: 1674-1056.

sputtering. *Appl. Phys. Lett.*, Vol. 91, Dec. 2007, p. 231907, ISSN: 1616-3028.

*Sensor and Actuators B*, Vol. 140, 2009, pp. 73–78, ISSN: 092-4005.

100 Nanowires - Recent Advances

*Materials Research*, Vols. 418-420, 2012, pp. 684-687, ISSN: 1022-6680.

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/3367

## **1. Introduction**

One-dimensional nanowires with a large aspect ratio have received much attention due to their unique shape anisotropy and extremely large surface area. Metallic nanowires such as Au, Ag and Cu nanowires [1-3] and semiconductor nanowires [4-9] would show unique electron transport properties which are not observed in bulk metals. Ferromagnetic nano‐ wires such as Ni, Co and Fe alloy nanowires as well as Co/Cu multilayered nanowires would be best candidate materials for magnetic field sensors with anisotropic magnetore‐ sistance (AMR) and giant magnetoresistance (GMR) effect. Nanowires can be fabricated by manipulating metallic atoms one by one using scanning tunneling microscopy (STM) or cat‐ alyst-assisted wet chemical etching technique [10], while they can be also prepared by elec‐ trochemically depositing metallic atoms into a nanochannel template with numerous nanochannels [11]. Nanochannel templates such as polycarbonate membrane films or ano‐ dized aluminum oxide films with high density of nanochannels (about 108 ~1010 cm–2) can be used in the template synthesis technique [12].

Using polymer membrane filters, ferromagnetic metal nanowires have been synthe sized so far. Whitney *et al.* reported that the arrays of Ni and Co nanowires were electrodeposited in polymer templates with the nanometer-sized pores prepared by nuclear track etching tech‐ nique [13]. They found that the preferred magnetization direction is perpendicular to the film plane and enhanced coercive force as high as 680 Oe. Piraux *et al.* reported that the ar‐ ray of Co/Cu multilayered nanowires with GMR response was electrodeposited in nanopo‐ rous polymer template [14]. Blondel *et al*. also reported that Co/Cu and Ni-Fe/Cu multilayered nanowires with GMR response was demonstrated [15]. They synthesized the multilayered nanowires with length of 6 μm, diameter of 80 nm and each layer thickness of 5~10 nm into the nanochannels of ion track-etched polycarbonate membrane filters. In the

© 2012 Ohgai; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Ohgai; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

report, GMR of 14% for Co/Cu and of 10% for Ni-Fe/Cu was demonstrated at ambient tem‐ perature in the current perpendicular to the layers. Ohgai *et al.* reported that magneto-sensi‐ tive nickel nanowires with AMR response could be fabricated into multi- and single-ion track templates with nanochannels using electrode position technique [16,17].

On the contrary, using anodized aluminum oxide pores on the surface of metallic alumi‐ num substrates, Ni, Co and Fe homogeneous ferromagnetic nanowires have been also electro‐ deposited and characterized in terms of their magnetization properties. Kawai *et al*. reported that magnetic properties of Ni, Co, Fe, Co-Ni, Fe-Ni and Fe-Co alloy nanowires with the co‐ ercive force ranging from 0.5 to 3.2kOe were electrodeposited into the nanopores of an ano‐ dic oxide coating films on aluminum [18-20]. Tsuya *et al.* also reported that the alumite films containing Fe nanowires were fabricated as an application to perpendicular magnetic record‐ ing medium [21,22]. Huysmans *et al*. reported that the magnetization curling process in per‐ pendicular direction was investigated using Fe nanowire arrays in alumite media [23]. Cheng *et al.* also reported that the magnetic anisotropy of electrodeposited Co nanowires on alu‐ mite substrate [24]. Zhang *et al* . reported that the influence of the packing density on the mag‐ netic behaviour was investigated using alumite media containg magnetic nanowires [25]. AlMawlawi *et al*. reported that the coercive force of Fe nanowires electrodeposited into ano‐ dic aluminum oxide pores increased up to around 2.1 kOe with increasing the aspect ratio of the nanowires [26]. Nielsch *et al*. reported that the uniform Ni nanowires array was synthe‐ sized into the ordered alumina pores by pulsed electrodeposition technique [27]. Evans *et al*. reported that the current perpendicular to plane giant magnetoresistance (CPP-GMR) was found using the multilayered nanowires electrodeposited in a commercially available ano‐ dic aluminum oxide membrane filter [28]. Ohgai *et al*. reported that the Co/Cu multilay‐ ered nanowires with spin-valve effect and GMR response were synthesized in selforganized anodized aluminum oxide nanopores grown at the surface of bulk aluminum. They also reported that the pore bottom oxide layer was removed by a chemical etching techni‐ que using aluminum oxide template with pore-length as short as 2000 nm. In their report, 20% of GMR was demonstrated in Co/Cu multilayered nanowires at room temperature, while the typical resistance switching of spin-valves was also demonstrated in Co/Cu/Co tri-lay‐ ered nanowires [29-32]. Wu *et al*. reported that the Ni nanodot and nanowire arrays could be prepared using a porous alumina layer on a bulk silicon as a template without a conduc‐ tive interlayer [33]. Chu *et al*. reported that the integrated ultra high-density Fe-Pt alloy nano‐ wires array was synthesized in a porous alumina layer on an ITO-glass template [34]. Wang *et al*. reported that the Ni nanowires with different diameters were prepared by direct-cur‐ rent electrodeposition into the pores of porous anodic alumina membrane [35]. Friedman *et al*. reported that the optimal parameters were found for synthesis of magnetic nanowires in porous alumina templates [36].

Electrodeposition of metallic nanowires into the as anodized aluminum template can be car‐ ried out using an alternating current or a pulsed current in order to reduce the charging up effect of the barrier layer at the interface between the anodized aluminum oxide layer and the metallic aluminum substrate. However, this resistive barrier layer makes it difficult to achieve good electric contacts at the pore bottom and well-controlled layered structure of electrodeposited nanowires. In this chapter, fabrication process of anodized aluminum ox‐ ide template without the barrier layer and the magnetoresistance properties of electrodepos‐ ited Ni, Co alloy nanowires and Co/Cu multilayered nanowires were investigated to synthesize novel functional ferromagnetic devices with anisotropic magnetoresistance (AMR) effect and giant magnetoresistance (GMR) effect.

## **2. Fabrication of Anodized Aluminum Oxide Nanochannels Templates**

#### **2.1. Barrier Layer Thinning by Chemical Etching Technique**

report, GMR of 14% for Co/Cu and of 10% for Ni-Fe/Cu was demonstrated at ambient tem‐ perature in the current perpendicular to the layers. Ohgai *et al.* reported that magneto-sensi‐ tive nickel nanowires with AMR response could be fabricated into multi- and single-ion

On the contrary, using anodized aluminum oxide pores on the surface of metallic alumi‐ num substrates, Ni, Co and Fe homogeneous ferromagnetic nanowires have been also electro‐ deposited and characterized in terms of their magnetization properties. Kawai *et al*. reported that magnetic properties of Ni, Co, Fe, Co-Ni, Fe-Ni and Fe-Co alloy nanowires with the co‐ ercive force ranging from 0.5 to 3.2kOe were electrodeposited into the nanopores of an ano‐ dic oxide coating films on aluminum [18-20]. Tsuya *et al.* also reported that the alumite films containing Fe nanowires were fabricated as an application to perpendicular magnetic record‐ ing medium [21,22]. Huysmans *et al*. reported that the magnetization curling process in per‐ pendicular direction was investigated using Fe nanowire arrays in alumite media [23]. Cheng *et al.* also reported that the magnetic anisotropy of electrodeposited Co nanowires on alu‐ mite substrate [24]. Zhang *et al* . reported that the influence of the packing density on the mag‐ netic behaviour was investigated using alumite media containg magnetic nanowires [25]. AlMawlawi *et al*. reported that the coercive force of Fe nanowires electrodeposited into ano‐ dic aluminum oxide pores increased up to around 2.1 kOe with increasing the aspect ratio of the nanowires [26]. Nielsch *et al*. reported that the uniform Ni nanowires array was synthe‐ sized into the ordered alumina pores by pulsed electrodeposition technique [27]. Evans *et al*. reported that the current perpendicular to plane giant magnetoresistance (CPP-GMR) was found using the multilayered nanowires electrodeposited in a commercially available ano‐ dic aluminum oxide membrane filter [28]. Ohgai *et al*. reported that the Co/Cu multilay‐ ered nanowires with spin-valve effect and GMR response were synthesized in selforganized anodized aluminum oxide nanopores grown at the surface of bulk aluminum. They also reported that the pore bottom oxide layer was removed by a chemical etching techni‐ que using aluminum oxide template with pore-length as short as 2000 nm. In their report, 20% of GMR was demonstrated in Co/Cu multilayered nanowires at room temperature, while the typical resistance switching of spin-valves was also demonstrated in Co/Cu/Co tri-lay‐ ered nanowires [29-32]. Wu *et al*. reported that the Ni nanodot and nanowire arrays could be prepared using a porous alumina layer on a bulk silicon as a template without a conduc‐ tive interlayer [33]. Chu *et al*. reported that the integrated ultra high-density Fe-Pt alloy nano‐ wires array was synthesized in a porous alumina layer on an ITO-glass template [34]. Wang *et al*. reported that the Ni nanowires with different diameters were prepared by direct-cur‐ rent electrodeposition into the pores of porous anodic alumina membrane [35]. Friedman *et al*. reported that the optimal parameters were found for synthesis of magnetic nanowires in

Electrodeposition of metallic nanowires into the as anodized aluminum template can be car‐ ried out using an alternating current or a pulsed current in order to reduce the charging up effect of the barrier layer at the interface between the anodized aluminum oxide layer and the metallic aluminum substrate. However, this resistive barrier layer makes it difficult to achieve good electric contacts at the pore bottom and well-controlled layered structure of electrodeposited nanowires. In this chapter, fabrication process of anodized aluminum ox‐

track templates with nanochannels using electrode position technique [16,17].

porous alumina templates [36].

102 Nanowires - Recent Advances

Aluminum sheets with thickness of 500 μm were used as a starting material to prepare ano‐ dized aluminum templates. First, the aluminum sheets were electrochemically polished in a ethanol solution containing 25 vol% of perchloric acid to achieve a mirror like surface. Dur‐ ing the electrochemical polishing, the cell voltage was kept at 8 V for 10 min at room tem‐ perature. Then, the polished aluminum sheets were anodized in an aqueous solution containing 0.3 mol/L oxalic acid to obtain a nanoporous aluminum oxide layer on the sur‐ face. During the anodization, the cell voltage was kept at 50 V for 10 min at room tempera‐ ture. Subsequently, the anodized aluminum sheets were immersed in an aqueous solution containing 5 vol% of phosphoric acid for 50 min to widen the pores, and to thin the oxide layer at the pore bottom. These processing parameters give pores 2 μm long with a diameter of 60 nm.

**Figure 1.** Time dependence of current density on fabrication of self-organized nanoporous anodized aluminum tem‐ plates. Anodization was conducted under 50 V in aqueous solution containing 0*.*3 mol/L oxalic acid.

Figure 1 shows the time dependence of current density on the fabrication of self-organized nanoporous anodized aluminum templates. Anodization was conducted under 50 V in aqueous solution containing 0*.*3 mol/L oxalic acid. At the beginning of anodization, the cur‐ rent density is very high due to oxygen evolution at the surface of aluminum as shown in Figure 1(a). Then, the current density rapidly decreases due to the formation of a highly re‐ sistive oxide layer as shown in Figure 1(b). Next, the current density gradually increases due to self-organized formation of nanopores in the oxide layer as shown in Figures 1(c), (d). Fi‐ nally, the current density reaches a constant value due to a stable growth rate of the porous oxide layer as shown in Figure 1(e).

**Figure 2.** SEM image of a cross section view of an anodized aluminum template. Anodization was conducted under 50 V for 10 min. The anodized aluminum oxide layer has a typical porous columnar structure and the pore length is approximately 2μm. The pore diameter is approximately 60 nm.

Figure 2 shows SEM images of a cross section view of an anodized aluminum template. The oxide layer has a typical porous columnar structure and the pore length is approximately 2μm as shown in Figure 2. The pore diameter is approximately 60 nm and the order of pore density is 1010 pores/cm2 as shown in Figure 2. It is well known that an anodized aluminum template has a resistive oxide layer (barrier layer) at the interface between the anodized alu‐ minum oxide film and the metallic aluminum substrate. Figure 3 shows SEM images of a pore top planar view (a) and a pore bottom back side view (b) of an anodized aluminum template. For observation of pore bottom back side (barrier layer), the massive metallic alu‐ minum substrate was dissolved in hydrochloric acid containing traces of cupric ions. The pore diameter is approximately 60 nm and the order of pore density is 1010 pores/cm2 as shown in Figure 3(a). The existence of a barrier layer is made obvious by looking at the pore bottom of anodized aluminum oxide layer (Figure3(b)).This resistive barrier layer makes well-controlled electrodeposition of layered structures difficult and lowers their AMR and GMR performance. Electrodeposition of metals into the pores of an anodized aluminum sur‐ face is usually carried out using alternating current or a pulsed current technique [37–40] in order to reduce the charging up effect of the barrier layer. To remove this barrier layer, the massive metallic aluminum backing is usually dissolved in an aqueous solution containing HgCl2 prior to thinning the barrier layer [41–45]. However, this aluminum substrate dissolv‐ ing technique can be applied to thick aluminum oxide layers with the thickness of several tens micro meters. On the contrary, when this barrier layer is removed or thinned without removing the aluminum backing, anodized aluminum templates with short pores length can be obtained. Since the thickness of the barrier layer is about several tens of nanometres, this layer can be removed or thinned using a chemical etching technique from the pore side direction [46]. Therefore, after the anodization process, to remove or thin this barrier layer without dissolving the metallic aluminum backing, the anodized aluminum templates were subsequently immersed in an aqueous solution containing phosphoric acid prior to the elec‐ trode position process.

Figure 1 shows the time dependence of current density on the fabrication of self-organized nanoporous anodized aluminum templates. Anodization was conducted under 50 V in aqueous solution containing 0*.*3 mol/L oxalic acid. At the beginning of anodization, the cur‐ rent density is very high due to oxygen evolution at the surface of aluminum as shown in Figure 1(a). Then, the current density rapidly decreases due to the formation of a highly re‐ sistive oxide layer as shown in Figure 1(b). Next, the current density gradually increases due to self-organized formation of nanopores in the oxide layer as shown in Figures 1(c), (d). Fi‐ nally, the current density reaches a constant value due to a stable growth rate of the porous

**Figure 2.** SEM image of a cross section view of an anodized aluminum template. Anodization was conducted under 50 V for 10 min. The anodized aluminum oxide layer has a typical porous columnar structure and the pore length is

Figure 2 shows SEM images of a cross section view of an anodized aluminum template. The oxide layer has a typical porous columnar structure and the pore length is approximately 2μm as shown in Figure 2. The pore diameter is approximately 60 nm and the order of pore

template has a resistive oxide layer (barrier layer) at the interface between the anodized alu‐ minum oxide film and the metallic aluminum substrate. Figure 3 shows SEM images of a pore top planar view (a) and a pore bottom back side view (b) of an anodized aluminum template. For observation of pore bottom back side (barrier layer), the massive metallic alu‐ minum substrate was dissolved in hydrochloric acid containing traces of cupric ions. The pore diameter is approximately 60 nm and the order of pore density is 1010 pores/cm2 as shown in Figure 3(a). The existence of a barrier layer is made obvious by looking at the pore bottom of anodized aluminum oxide layer (Figure3(b)).This resistive barrier layer makes

as shown in Figure 2. It is well known that an anodized aluminum

oxide layer as shown in Figure 1(e).

104 Nanowires - Recent Advances

approximately 2μm. The pore diameter is approximately 60 nm.

density is 1010 pores/cm2

**Figure 3.** SEM images of pore top plan view (a) and pore bottom plan view (b) of an anodized aluminum template. Anodization was conducted under 50 V for 10 min. The metallic aluminum substrate was dissolved in the hydrochloric acid aqueous solution containing small amount of cupric ions.

#### **2.2. Barrier Layer Thinning by Anodization Voltage Controlling**

Figure 4 shows experimental apparatus and time-dependence of applied voltage to synthe‐ size an anodized aluminum oxide membrane filter. Anodized aluminum oxide thick films with numerous nanochannels were exfoliated mechanically by the pressure of hydrogen gas generated at the interface between an oxide layer and a metallic aluminum during the subse‐ quent cathodic reduction process after the growing of anodic aluminum oxide layer. At first, 70 V was applied to growing aluminum oxide long nanochannel. Then, the anodization volt‐ age was decreased gradually down to 0 V for thinning the pore bottom oxide layer (barrier layer). Finally, cathodic voltage was applied to exfoliate an anodized aluminum oxide mem‐ brane film from the metallic aluminum rod due to the hydrogen evolution. Figure 5 shows sur‐ face appearance of anodized aluminum oxide membrane filters exfoliated from a metallic aluminum rod. These membrane filters were obtained by anodizing at 50 V (Figure 5(a)) and 70 V (Figure 5(b)). The membrane filter anodized at 70 V had round and disc shape with the diameter of 10 mm (Figure 5(b)) while that anodized at 50 V had several cracks (Figure 5(a)).

**Figure 4.** Experimental apparatus and time-dependence of applied voltage to synthesize an anodized aluminum ox‐ ide membrane filter.

**Figure 5.** Surface appearance of anodized aluminum oxide membrane filters exfoliated from a metallic aluminum rod. These membrane filters were anodized at 50 V (a) and 70 V (b).

Figure 6 shows SEM images of barrier layer side (plan view (a), cross section (b)) and sur‐ face side (plan view (c), cross section (d)) of an anodized aluminum oxide membrane filter exfoliated from a metallic aluminum rod. The anodization was conducted at 70 V. The aver‐ age channel diameter at the surface side was around 60 nm while that at the barrier layer side was ca. 20 nm. The oxide layer has a typical porous columnar structure and the channel length was ca. 60 μm while the channel density was around 108 channels / cm2 .

70 V was applied to growing aluminum oxide long nanochannel. Then, the anodization volt‐ age was decreased gradually down to 0 V for thinning the pore bottom oxide layer (barrier layer). Finally, cathodic voltage was applied to exfoliate an anodized aluminum oxide mem‐ brane film from the metallic aluminum rod due to the hydrogen evolution. Figure 5 shows sur‐ face appearance of anodized aluminum oxide membrane filters exfoliated from a metallic aluminum rod. These membrane filters were obtained by anodizing at 50 V (Figure 5(a)) and 70 V (Figure 5(b)). The membrane filter anodized at 70 V had round and disc shape with the diameter of 10 mm (Figure 5(b)) while that anodized at 50 V had several cracks (Figure 5(a)).

**Figure 4.** Experimental apparatus and time-dependence of applied voltage to synthesize an anodized aluminum ox‐

**Figure 5.** Surface appearance of anodized aluminum oxide membrane filters exfoliated from a metallic aluminum rod.

Figure 6 shows SEM images of barrier layer side (plan view (a), cross section (b)) and sur‐ face side (plan view (c), cross section (d)) of an anodized aluminum oxide membrane filter exfoliated from a metallic aluminum rod. The anodization was conducted at 70 V. The aver‐ age channel diameter at the surface side was around 60 nm while that at the barrier layer

These membrane filters were anodized at 50 V (a) and 70 V (b).

ide membrane filter.

106 Nanowires - Recent Advances

**Figure 6.** SEM images of barrier layer side (plan view (a), cross section (b)) and surface side (plan view (c), cross section (d)) of an anodized aluminum oxide membrane filter exfoliated from a metallic aluminum rod. The anodization was conducted at 70 V.

Figure 7 shows the effect of anodization voltage on the growth rate of anodized aluminum oxide layer. The oxide layer thickness was determined by observing the cross sectional SEM images. With increasing the anodization voltage *V <sup>A</sup>*(V), the oxide layer thickness was in‐ creased exponentially. The average growth rate *R <sup>G</sup> Ave*(nm/sec) of anodized aluminum oxide layer can be expressed by the following equation (1).

$$
\log R\_G^{\text{ave}} = 0.017V\_A - 0.34 \tag{1}
$$

According to the equation (1), if *V <sup>A</sup>* equal to 70 V, *R <sup>G</sup> Ave* is estimated to be around 7.1 nm/ s. Ohgai *et al*. [31] have reported that the average growth rate of the anodized aluminum ox‐ ide layer depends on the anodization voltage and the growth rate *R\* <sup>G</sup> Ave* (nm/sec) increas‐ es exponentially with increasing the anodization voltage *V <sup>A</sup>* (V) as shown in the following equation (2).

$$
\log R \, ^\ast \, \_G^{\text{rev}} = 0.03 V\_A - 0.78 \,\tag{2}
$$

According to the equation (2), if *V <sup>A</sup>* equal to 70 V, *R\* <sup>G</sup> Ave* is calculated to be around 21 nm/s which is ca. 3 times larger than that obtained from the equation (1). It is well known that the real time growth rate of the anodized aluminum oxide layer *R <sup>G</sup> Real* decreases with increase in the oxide layer thickness. In the equation (1), range of the oxide layer thickness is from 50 to 200 μm while the range is from 1 to 60 μm in the equation (2). Therefore, *R <sup>G</sup> Ave*calculated from the equation (1) would be smaller than that obtained from the equation (2). [1]

**Figure 7.** Effect of anodization voltage on the growth rate of anodized aluminum oxide layer.

Figure 8 shows the relationship between film thickness of anodized aluminum oxide layer and anodization time. The anodization was conducted at 70 V. The film thickness *y* (μm) in‐ creased with increasing the anodization time *x* (hour) according to a parabolic low as shown in the following equation (3).

$$\mathbf{y} = \mathbf{44.7x}^{0.47} \tag{3}$$

Therefore, the real time growth rate of the anodized aluminum oxide layer *R <sup>G</sup> Real* can be ex‐ pressed by the following equation(4).

$$R\_{\odot} \stackrel{\text{Real}}{=} \text{dy} / \text{dx} = \text{ 21.2x}^{-0.83} \text{(mm} / \text{hour)} = \text{ 5.9x}^{-0.83} \text{(mm} / \text{sec)} \tag{4}$$

Consequently, *R <sup>G</sup> Real* decreases with increase in the anodization time *x* and the oxide layer thickness *y*. Ohgai *et al*. [30] have reported that the film thickness *y\**(μm) of anodized alumi‐ num oxide layer increases linearly with increasing the anodization time *x* (hour) as shown in the following equation (5).

$$y = \mathbf{50x} \tag{5}$$

Therefore, the real time growth rate of the anodized aluminum oxide layer *R\* <sup>G</sup> Real* would be constant as shown in the following equation (6).

Magnetoresistance of Nanowires Electrodeposited into Anodized Aluminum Oxide Nanochannels http://dx.doi.org/10.5772/3367 109

$$R\text{ \*}\_{G}^{\text{\*}\_{\text{Roll}}} = d\text{y} / d\text{x} = \\$0(\text{mm} / \text{hour})\text{ =} 14 \text{(nm} / \text{sec})\tag{6}$$

As mentioned previously, the real time growth rate of the anodized aluminum oxide layer *R <sup>G</sup> Real* decreases with increase in the oxide layer thickness. In the equation (4), range of the oxide layer thickness is from 50 to 200 μm while the range is from 0.2 to 50 μm in the equa‐ tion (6). Therefore, *R <sup>G</sup> Real*calculated from the equation (4) would be smaller than that ob‐ tained from the equation (6).

**Figure 8.** Relationship between film thickness of anodized aluminum oxide layer and anodization time. The anodiza‐ tion was conducted at 70 V.

## **3. Electrodeposition and Magnetoresistance of Ni and Co Nanowires Array**

#### **3.1. Electrodeposition of Ni and Co Nanowires Array**

**Figure 7.** Effect of anodization voltage on the growth rate of anodized aluminum oxide layer.

in the following equation (3).

108 Nanowires - Recent Advances

the following equation (5).

constant as shown in the following equation (6).

pressed by the following equation(4).

Figure 8 shows the relationship between film thickness of anodized aluminum oxide layer and anodization time. The anodization was conducted at 70 V. The film thickness *y* (μm) in‐ creased with increasing the anodization time *x* (hour) according to a parabolic low as shown

Therefore, the real time growth rate of the anodized aluminum oxide layer *R <sup>G</sup> Real* can be ex‐

Consequently, *R <sup>G</sup> Real* decreases with increase in the anodization time *x* and the oxide layer thickness *y*. Ohgai *et al*. [30] have reported that the film thickness *y\**(μm) of anodized alumi‐ num oxide layer increases linearly with increasing the anodization time *x* (hour) as shown in

Therefore, the real time growth rate of the anodized aluminum oxide layer *R\* <sup>G</sup> Real* would be

( ) 0.53 0.53 / 21.2 (mm / hour) 5.9 nm / sec *Real R dy dx x <sup>G</sup> x* - - = = = (4)

0.47 *y x* = 44.7 (3)

*y x* = 50 (5)

Figure 9 shows experimental apparatus and a nanochannel template for electrodeposition of nanowires array. The exfoliated anodized aluminum oxide nanochannels were used as tem‐ plates for growing nanowires. On surface of the membrane filter, a gold layer was sputter-de‐ posited to cover the pores and make a cathode. A gold wire and Ag/AgCl electrode were used as an anode and a reference electrode. For example, aqueous solution containing NiSO4 (120 g/L) and H3BO3 (45 g/L) was used as electrolyte for electrodeposition of Ni nanowires.

**Figure 9.** Experimental apparatus and a nanochannel template for electrodeposition of nanowires array.

**Figure 10.** Cathode polarization curves for Ni and Co electrodeposition from aqueous solution containing Ni2+ or Co2+ ions.

Figure 10 shows cathode polarization curves for Ni and Co electrodeposition from aqueous solution containing Ni2+ or Co2+ions. According to Nernst equation, the equilibrium poten‐ tials of Ni/Ni2+ and Co/Co2+ are estimated to be around −0.46 and −0.48V (vs. Ag/AgCl). As shown in Figure 10, the cathodic currents occur at around -0.2V which is more nobler poten‐ tial region than the equilibrium potential of Ni and Co. Therefore, this cathodic current would be mainly caused by the reduction current of hydrogen ions. With increasing the cathodic current density, at around 2A/m2 , the potential polarizes significantly to the less noble region. This polarization would be caused by the diffusion limit of hydrogen ions. At around -0.8V which is less nobler potential region than the equilibrium potential of Ni and Co, the cathodic current increases again. It is well-known that the electrodeposition of irongroup metals such as Ni, Co and Fe is accompanied by the over potential due to the rate determination of multi-step reduction process even in the form of their aqua ions. Therefore, in the present work, Ni2+ and Co2+ ions would electrodeposit with accompanying the over‐ potential even in simple aqueous solutions containing sulfuric and boric acid. Consequently, this cathodic current would be mainly caused by the reduction current of Ni2+ and Co2+. These characteristics are identical to those obtained with the pores in nanochannel polycar‐ bonate templates with metallic gold cathode. This result supports the observation that the barrier layer at the pore bottom of an anodized aluminum template is well removed by the exfoliation process from a metallic aluminum rod and the nanowires growon metallic gold cathode. Using the polarization curves, the optimum electrodeposition potential range for growing Ni and Co nanowires are determined to be from −1.0 to -1.2V (vs. Ag/AgCl) which is the potential region more nobler than the diffusion limit of each metal ions.

**Figure 9.** Experimental apparatus and a nanochannel template for electrodeposition of nanowires array.

**Figure 10.** Cathode polarization curves for Ni and Co electrodeposition from aqueous solution containing Ni2+ or Co2+

Figure 10 shows cathode polarization curves for Ni and Co electrodeposition from aqueous solution containing Ni2+ or Co2+ions. According to Nernst equation, the equilibrium poten‐ tials of Ni/Ni2+ and Co/Co2+ are estimated to be around −0.46 and −0.48V (vs. Ag/AgCl). As shown in Figure 10, the cathodic currents occur at around -0.2V which is more nobler poten‐ tial region than the equilibrium potential of Ni and Co. Therefore, this cathodic current would be mainly caused by the reduction current of hydrogen ions. With increasing the

noble region. This polarization would be caused by the diffusion limit of hydrogen ions. At around -0.8V which is less nobler potential region than the equilibrium potential of Ni and Co, the cathodic current increases again. It is well-known that the electrodeposition of iron-

, the potential polarizes significantly to the less

ions.

110 Nanowires - Recent Advances

cathodic current density, at around 2A/m2

**Figure 11.** Effect of cathode potential on the time-dependence of cathode current during the electrodeposition of Ni nanowires and Co nanowires.

Time dependence of cathode current was monitored during the electrodeposition to investi‐ gate the growing process of nanowires. Ni nanowires and Co nanowires were potentio-stati‐ cally electrodeposited. Figure 11 shows the effect of cathode potential on the timedependence of cathode current during the electrodeposition of Ni nanowires and Co nanowires. During the electrodepostion, cathode potentials were fixed to -0.8, –0.9, –1.0, –1.1 and –1.2 V. If the potential was kept to –1.0 V for Ni deposition, the cathode current reached up to ca. 1 mA at the beginning of electrolysis within several tens of minutes. Then, the cur‐ rent rapidly decreased to be ca. 0.4 mA and kept the constant current until around 1000 sec. During this process, electrodeposition of Ni proceeds in the nanopores. At the initial stage of the electrodepostion, large cathode current was observed in each cathode potential. The con‐ centration of metal ions in the nanopores will decrease with increasing the electrodeposition time due to the reduction of metal ions, while the metal ions will be provided from the bulk solution to the nanopores, where the metal ions are consumed on the cathode due to the electrodeposition process. Finally, as shown in Figure 11, the cathode current rapidly in‐ creases at the deposition time more than 1000 sec. At this stage, electrodeposited nanowires reach the surface of the membranes and large hemispheric Ni deposits are formed. Growth rate of Ni nanowires can be estimated from dividing channel length by the filling time. For example, growth rate of Ni nanowires can be estimated as ca. 6 nm/sec at the cathode poten‐ tial of –1.0 V. Time-dependence of cathode current for Co deposition also showed similar behavior as well as Ni deposition. Growth rate of the nanowires increases up to ca. 30nm/sec with increasing the cathode potential up to -1.2 V.

#### **3.2. Structure of Ni and Co Nanowires Array**

After the growing nanowires, anodized aluminum oxide membrane filters were dissolved in aqueous solution containing sodium hydroxide and the remains consisted of nanowires and a gold layer was served as a sample for scanning electron microscope (SEM) and transmis‐ sion electron microscope (TEM) observation. Figure 12 shows SEM images of Ni nanowires separated from the anodized aluminum oxide templates. Diameter (60 and 300 nm) and length (6 and 30μm) of the nanowires corresponds well to that of nanopores and the cylin‐ drical shape was precisely transferred from the nanopores to the nanowires. Aspect ratio of the nanowires reaches up to around 100.

**Figure 12.** SEM images of electrodeposited Ni nanowires separated from anodized aluminum oxide membrane filters with the pore diameters of ca. 300nm (a),(b) and ca. 60nm (c),(d).

TEM bright field images and electron diffraction patterns of electrodeposited Ni and Co nanowires were also investigated as shown in Figure 13. According to TEM bright field im‐ ages, shape of the nanowires was almost cylindrical and the electron diffraction patterns are composed of spots, which suggests a nanowire consists of a crystalline phase with preferen‐ tial orientation.

During this process, electrodeposition of Ni proceeds in the nanopores. At the initial stage of the electrodepostion, large cathode current was observed in each cathode potential. The con‐ centration of metal ions in the nanopores will decrease with increasing the electrodeposition time due to the reduction of metal ions, while the metal ions will be provided from the bulk solution to the nanopores, where the metal ions are consumed on the cathode due to the electrodeposition process. Finally, as shown in Figure 11, the cathode current rapidly in‐ creases at the deposition time more than 1000 sec. At this stage, electrodeposited nanowires reach the surface of the membranes and large hemispheric Ni deposits are formed. Growth rate of Ni nanowires can be estimated from dividing channel length by the filling time. For example, growth rate of Ni nanowires can be estimated as ca. 6 nm/sec at the cathode poten‐ tial of –1.0 V. Time-dependence of cathode current for Co deposition also showed similar behavior as well as Ni deposition. Growth rate of the nanowires increases up to ca. 30nm/sec

After the growing nanowires, anodized aluminum oxide membrane filters were dissolved in aqueous solution containing sodium hydroxide and the remains consisted of nanowires and a gold layer was served as a sample for scanning electron microscope (SEM) and transmis‐ sion electron microscope (TEM) observation. Figure 12 shows SEM images of Ni nanowires separated from the anodized aluminum oxide templates. Diameter (60 and 300 nm) and length (6 and 30μm) of the nanowires corresponds well to that of nanopores and the cylin‐ drical shape was precisely transferred from the nanopores to the nanowires. Aspect ratio of

**Figure 12.** SEM images of electrodeposited Ni nanowires separated from anodized aluminum oxide membrane filters

with increasing the cathode potential up to -1.2 V.

**3.2. Structure of Ni and Co Nanowires Array**

112 Nanowires - Recent Advances

the nanowires reaches up to around 100.

with the pore diameters of ca. 300nm (a),(b) and ca. 60nm (c),(d).

**Figure 13.** TEM bright field images and electron diffraction patterns of Ni and Co nanowires electrodeposited in nanochannel membrane filters.

#### **3.3. Magnetic Properties of Ni and Co Nanowires Array**

Magnetization and anisotropic magnetoresistance (AMR) of Ni alloy and Co alloy nano‐ wires were measured using a vibrating sample magnetometer (VSM) and LCR meter with increasing the magnetic field up to 10 kOe. Figure 14 shows magnetic hysteresis loops of pure Ni, Ni-1.5%Co, Ni-0.8%Fe, pure Co, Co-0.9%Ni and Co-0.1%Fe alloy nanowires elec‐ trodeposited into anodized aluminum oxide templates with channel-diameter of 60 nm. Magnetic field was applied to in-plan direction and perpendicular direction to the mem‐ brane film plan. The perpendicular direction to the membrane film plan corresponds to the parallel direction to the long axis of nanowires. These nanowires were hardly magnetized in in-plan direction and the magnetization reached to saturation at more than 5 kOe as shown in Figure 14. On the contrary, these nanowires were easily magnetized in perpendicular di‐ rection and the coercive force reached up to around 1 kOe. These magnetization curves re‐ vealed that the electrodeposited nanowires have a typical perpendicular magnetization behaviour due to the uni-axial shape anisotropy.

**Figure 14.** Magnetic hysteresis loops of pure Ni, Ni-1.5%Co, Ni-0.8%Fe, pure Co, Co-0.9%Ni and Co-0.1%Fe alloy nanowires electrodeposited into anodized aluminum oxide templates with channel-diameter of 60 nm. Magnetic field was applied to in-plan direction and perpendicular direction to the membrane film plan. The perpendicular direc‐ tion to the membrane film plan corresponds to the parallel direction to the long axis of nanowires.

Figure 15 shows magnetoresistance curves of pure Ni, Ni-1.5%Co, Ni-0.8%Fe, pure Co, Co-0.9%Ni and Co-0.1%Fe alloy nanowires electrodeposited into anodized aluminum oxide templates with channel-diameter of 60 nm. Magnetic field was applied to in-plan direction and perpendicular direction to the membrane film plan. The perpendicular direction to the membrane film plan corresponds to the parallel direction to the long axis of nanowires. Here, the anisotropic magnetoresistance (AMR) ratio is defined by the following equation:

$$\text{AMR} \left( \% \right) \;= \; 100 \times (\text{R}\_{\text{max}} - \text{R}\_{\text{min}}) \;/ \text{R}\_{\text{min}} \tag{7}$$

The magnetoresistive hysteresis of the Ni alloy and Co alloy nanowires depended strongly on the direction of the magnetic field as shown in Figure 15. In the magnetic field direction of 0° (the long axis of nanowires is parallel to the magnetic field), the AMR ratio was almost zero, while a maximum AMR ratio was observed in the magnetic field direction of 90° (the long axis of nanowiresis perpendicular to the magnetic field). Resistance of the nanowires decreased with increase in the magnetic field and the AMR ratio reached 1.0–3.2% with the Ni alloy nanowires. The saturation field of the Ni nanowires was estimated to be about 7 kOe. Ferre *et al.* [47] reported on the magnetoresistance of Ni and Co nanowires electrode‐ posited in ion-track etched polycarbonate templates. In their report, the magnetoresistance curves obtained from Ni nanowires showed typical AMR behaviour (1.4% of AMR ratio) over a wide range of the diameter in the nanowires, which corresponds well with the results obtained in this study. On the contrary, AMR ratio of the Co alloy nanowires was only 0.6– 0.8%. The range of applied magnetic fields was not enough to measure the saturation mag‐ netic field of Co nanowires. Since the magnetic field was insufficient to saturate the magneti‐ zation of Co nanowires, the AMR value was not actual full value.

brane film plan. The perpendicular direction to the membrane film plan corresponds to the parallel direction to the long axis of nanowires. These nanowires were hardly magnetized in in-plan direction and the magnetization reached to saturation at more than 5 kOe as shown in Figure 14. On the contrary, these nanowires were easily magnetized in perpendicular di‐ rection and the coercive force reached up to around 1 kOe. These magnetization curves re‐ vealed that the electrodeposited nanowires have a typical perpendicular magnetization

**Figure 14.** Magnetic hysteresis loops of pure Ni, Ni-1.5%Co, Ni-0.8%Fe, pure Co, Co-0.9%Ni and Co-0.1%Fe alloy nanowires electrodeposited into anodized aluminum oxide templates with channel-diameter of 60 nm. Magnetic field was applied to in-plan direction and perpendicular direction to the membrane film plan. The perpendicular direc‐

Figure 15 shows magnetoresistance curves of pure Ni, Ni-1.5%Co, Ni-0.8%Fe, pure Co, Co-0.9%Ni and Co-0.1%Fe alloy nanowires electrodeposited into anodized aluminum oxide templates with channel-diameter of 60 nm. Magnetic field was applied to in-plan direction and perpendicular direction to the membrane film plan. The perpendicular direction to the membrane film plan corresponds to the parallel direction to the long axis of nanowires. Here,

The magnetoresistive hysteresis of the Ni alloy and Co alloy nanowires depended strongly on the direction of the magnetic field as shown in Figure 15. In the magnetic field direction of 0° (the long axis of nanowires is parallel to the magnetic field), the AMR ratio was almost zero, while a maximum AMR ratio was observed in the magnetic field direction of 90° (the long axis of nanowiresis perpendicular to the magnetic field). Resistance of the nanowires

AMR % 100 (R – R ) / R ( ) = ´ max min min (7)

the anisotropic magnetoresistance (AMR) ratio is defined by the following equation:

tion to the membrane film plan corresponds to the parallel direction to the long axis of nanowires.

behaviour due to the uni-axial shape anisotropy.

114 Nanowires - Recent Advances

**Figure 15.** Magnetoresistance curves of pure Ni, Ni-1.5%Co, Ni-0.8%Fe, pure Co, Co-0.9%Ni and Co-0.1%Fe alloy nanowires electrodeposited into anodized aluminum oxide templates with channel-diameter of 60 nm.Magnetic field was applied to in-plan direction and perpendicular direction to the membrane film plan. The perpendicular direction to the membrane film plan corresponds to the parallel direction to the long axis of nanowires.

## **4. Electrodeposition and Magnetoresistance of Co/Cu Multilayered Nanowires**

#### **4.1. Electrodeposition of Co/Cu Multilayered Nanowires**

For growing Co/Cu multilayered nanowires, the exfoliated anodized aluminum oxide nano‐ channels were used as templates. Aqueous solution containing CoSO4 (120 g/L), CuSO4 (1.6 g/L) and H3BO3 (45 g/L) was used as electrolyte for electrodeposition of Co/Cu multilayered nanowires. A cathode polarization curve was measured over a wide range of cathode poten‐ tial to determine the optimum potential for Cu and Co deposition. Figure 16 shows cathode polarization curves for electrodeposition of Cu and Co from the mixed solution (containing Cu2+ and Co2+ ions) and the solution containing only Co2+ ions. The equilibrium potentials of Cu/Cu2+ and Co/Co2+ are estimated to be around +0.05 V and –0.48 V (vs.Ag/AgCl) on the basis of Nernst equation.The cathode current occurs at the potential region close to the equi‐ librium potential of Cu as shown in Figure 16. It is well-known that Cu2+ ions begin to elec‐ trodeposit without accompanying overpotential from the aqueous solution. Therefore, this cathode current corresponds to the deposition current of Cu. With increasing cathode cur‐ rent, at around 10-2A, the potential significantly polarizes to the less-noble region. This po‐ larization would be caused by the diffusion limit of Cu2+ ions. In the potential region which is less-nobler than the equilibrium potential of Co, the cathode current increases again at around –0.7 V. It is also well-known that the electrodeposition of iron-group metals such as Ni, Co, and Fe is accompanied by the overpotential due to the rate determination of multistep reduction process. Therefore, this increase in cathode current would be mainly caused by the reduction current of Co2+ ions.

**Figure 16.** Cathode polarization curves for electrodeposition of Cu and Co from the mixed solution (containing Cu2+ and Co2+ ions) and the solution containing only Co2+ ions.

Growth rates of nanowires were estimated by the channel filling time, which was deter‐ mined from the time dependence of deposition current at each potential. Figure 17 shows the effect of cathode potential on the time dependence of cathodic current during the elec‐ trodeposition of Co nanowires. The cathode potentials were fixed to -0.80,-0.85, -0.90, -0.95 and -1.0 V. To determine the nanowire growth rate, the channel-filling time was estimat‐ ed by monitoring the deposition current. When the nanowires reach the membrane sur‐ face, the current will increase drastically due to the formation of hemispherical caps. The growth rates were calculated dividing the channel length by the channel-filling time. For ex‐ ample, at –1.0 V, the channel-filling time is around 40 s and the deposition rate is estimat‐ ed to be ca. 150nm/s. On the basis of the results shown in Figure 16 and Figure 17, the optimum deposition potentials of Cu and Co are determined to be about –0.4 and –1.15 V (vs. Ag/AgCl) which is the potential region nobler than the diffusion limit potential of each metal ion. Typical deposition rates of Cu and Co were roughly 10 nm/s (at –0.4 V) and 200 nm s-1 (at –1.15 V).

nanowires. A cathode polarization curve was measured over a wide range of cathode poten‐ tial to determine the optimum potential for Cu and Co deposition. Figure 16 shows cathode polarization curves for electrodeposition of Cu and Co from the mixed solution (containing Cu2+ and Co2+ ions) and the solution containing only Co2+ ions. The equilibrium potentials of Cu/Cu2+ and Co/Co2+ are estimated to be around +0.05 V and –0.48 V (vs.Ag/AgCl) on the basis of Nernst equation.The cathode current occurs at the potential region close to the equi‐ librium potential of Cu as shown in Figure 16. It is well-known that Cu2+ ions begin to elec‐ trodeposit without accompanying overpotential from the aqueous solution. Therefore, this cathode current corresponds to the deposition current of Cu. With increasing cathode cur‐ rent, at around 10-2A, the potential significantly polarizes to the less-noble region. This po‐ larization would be caused by the diffusion limit of Cu2+ ions. In the potential region which is less-nobler than the equilibrium potential of Co, the cathode current increases again at around –0.7 V. It is also well-known that the electrodeposition of iron-group metals such as Ni, Co, and Fe is accompanied by the overpotential due to the rate determination of multistep reduction process. Therefore, this increase in cathode current would be mainly caused

**Figure 16.** Cathode polarization curves for electrodeposition of Cu and Co from the mixed solution (containing Cu2+

Growth rates of nanowires were estimated by the channel filling time, which was deter‐ mined from the time dependence of deposition current at each potential. Figure 17 shows the effect of cathode potential on the time dependence of cathodic current during the elec‐

by the reduction current of Co2+ ions.

116 Nanowires - Recent Advances

and Co2+ ions) and the solution containing only Co2+ ions.

**Figure 17.** Effect of cathode potential on the time dependence of cathodic current during the electrodeposition of Co nanowires.

Co/Cu multilayered nanowires were synthesized alternatingly switching cathode poten‐ tial from -0.4 V (for Cu layer) to -1.15 V (for Co layer) as shown in Figure 18. Accord‐ ing to this figure, when the potential is switched from -1.15 V to -0.4 V, anodic current is observed. This is resulting from the dissolution of electrodeposited Co, because -0.4V is more nobler than the equilibrium potential of Co. At this potential, it is estimated that the Cu deposition and Co dissolution will proceed simultaneously. According to the timedependence of cathodic current during electrodeposition of Co/Cu multilayered nano‐ wires as shown in Figure 19, filling time was around 3500 s and the deposition rate was estimated to be about 17 nm/s.

**Figure 18.** Growing process of Co/Cu multilayered nanowires synthesized alternatingly switching cathode potential from -0.4 V (for Cu layer) to -1.15 V (for Co layer).

**Figure 19.** Time dependence of observed-current during the electrodeposition of Co/Cu multilayered nanowires.

#### **4.2. Structure of Co/Cu Multilayered Nanowires**

TEM bright field images of electrodeposited Co/Cu multilayered nanowires were investigat‐ ed as shown in Figure 20. Shape of the nanowires was almost cylindrical and the multilay‐ ered structure was also observed as shown in this images. In the structure, several stacking faults were also observed. There would be significant strain energy in the interphase boun‐ dary between Co and Cu layer. The strain energy could be stored as the stacking fault ener‐ gy. It is well known that the stacking fault structure is usually observed in fcc crystals such as Au, Ag, Cu, Ni, Al, etc. Therefore, the stacking fault structure would be introduced to the Cu layer from the interface between Co and Cu layer.

**Figure 20.** TEM bright field images of Co/Cu multilayered nanowires electrodeposited in nano-channel membrane fil‐ ters.

#### **4.3. Magnetic Properties of Co/Cu Multilayered Nanowires**

**Figure 18.** Growing process of Co/Cu multilayered nanowires synthesized alternatingly switching cathode potential

**Figure 19.** Time dependence of observed-current during the electrodeposition of Co/Cu multilayered nanowires.

TEM bright field images of electrodeposited Co/Cu multilayered nanowires were investigat‐ ed as shown in Figure 20. Shape of the nanowires was almost cylindrical and the multilay‐ ered structure was also observed as shown in this images. In the structure, several stacking faults were also observed. There would be significant strain energy in the interphase boun‐ dary between Co and Cu layer. The strain energy could be stored as the stacking fault ener‐ gy. It is well known that the stacking fault structure is usually observed in fcc crystals such

from -0.4 V (for Cu layer) to -1.15 V (for Co layer).

118 Nanowires - Recent Advances

**4.2. Structure of Co/Cu Multilayered Nanowires**

Typical perpendicular magnetization behavior was observed from the magnetic hysteresis loops of electrodeposited Co/Cu multilayered nanowires with diameter of 60 nm and length of 60μm as shown in Figure 21-(a). Coercive force of Co/Cu multilayered nanowires was ca. 1 kOe. This is resulting from the shape anisotropy of nanowires with super large aspect ratio of ca. 1000. The magnetoresistance curves revealed that 10.5 % of GMR effect was obtained in the multilayered nanowires with Co layer 10nm, Cu layer 10 nm and 3000 bi-layers as shown in Figure 21-(b).

**Figure 21.** Magnetic hysteresis loops (a) and magnetoresistance curves (b) of Co/Cu multilayered nanowires electro‐ deposited into anodized aluminum oxide templates with channel-diameter of 60 nm. Magnetic field was applied to in-plan direction and perpendicular direction to the membrane film plan. The perpendicular direction to the mem‐ brane film plan corresponds to the parallel direction to the long axis of nanowires.

## **5. Summary**

1 kOe. This is resulting from the shape anisotropy of nanowires with super large aspect ratio of ca. 1000. The magnetoresistance curves revealed that 10.5 % of GMR effect was obtained in the multilayered nanowires with Co layer 10nm, Cu layer 10 nm and 3000 bi-layers as

**Figure 21.** Magnetic hysteresis loops (a) and magnetoresistance curves (b) of Co/Cu multilayered nanowires electro‐ deposited into anodized aluminum oxide templates with channel-diameter of 60 nm. Magnetic field was applied to in-plan direction and perpendicular direction to the membrane film plan. The perpendicular direction to the mem‐

brane film plan corresponds to the parallel direction to the long axis of nanowires.

shown in Figure 21-(b).

120 Nanowires - Recent Advances

Anodized aluminum oxide films with the thickness ranging from 2μm to 200 μm were syn‐ thesized using a bipolar continuous electrolysis process with anodic oxidation and cathodic exfoliation as well as barrier layer thinning process using chemical etching technique. 3.2 % of AMR effect was observed in Ni-1.5at.% Co alloy nanowires with diameter of 60 nm and length of 60μm. Co/Cu multilayered nanowires with Co-10 nm, Cu-10 nm and 3000 bi-layers were synthesized using a pulsed current electrodeposition technique. 10.5 % of GMR effect was observed in Co/Cu multilayered nanowires electrodeposited into anodized aluminum oxide template with super large aspect ratio of ca. 1000.

## **Author details**

Takeshi Ohgai\*

Address all correspondence to: ohgai@nagasaki-u.ac.jp

Division of Chemistry and Materials Science, Nagasaki University, 1-14 Bunkyo-machi, Na‐ gasaki 852-8521, Japan

## **References**


drical Nano-pores on the Surface of Anodized Aluminum. *Journalof Applied Electro‐ chemistry*, 35(5), 479-485.


[20] Kawai, S., & Ishiguro, I. (1976). Recording Characteristics of Anodic Oxide Films on Aluminum Containing Electrodeposited Ferromagnetic Metals and Alloys. *Journal ofThe Electrochemical Society*, 123(7), 1047-1051.

drical Nano-pores on the Surface of Anodized Aluminum. *Journalof Applied Electro‐*

[8] Katkar, R. A., & Tait, G. B. (2007). The Effect of Stationary Ultraviolet Excitation on the Optical Properties of Electrochemically Self-assembled Semiconductor Nano‐

[9] Inguanta, R., Sunseri, C., & Piazza, S. (2007). Photoelectrochemical Characterization of Cu2O-Nanowire Arrays Electrodeposited into Anodic Alumina Membranes. Elec‐

[10] Kim, D. J., Seol, J. K., Lee, M. R., Hyung, J. H., Kim, G. S., Ohgai, T., & Lee, S. K. (2012). Ferromagnetic Nickel Silicide Nanowires for Isolating Primary CD4+

[11] Martin, C. R. (1991). Template Synthesis of Polymeric and Metal Microtubules. *Ad‐*

[12] Spohr, R., Zet, C., Fischer, B. E., Kiesewetter, H., Apel, P., Gunko, I., Ohgai, T., & Westerberg, L. (2010). Controlled Fabrication of Ion Track Nanowires and Channels.

[13] Whitney, T. M., Jiang, J. S., Searson, P. C., & Chien, C. L. (1993). Fabrication and Mag‐ netic Properties of Arrays of Metallic Nanowires. *Science*, 261(5126), 1316-1319.

[14] Piraux, L., George, J. M., Despres, J. F., Leroy, C., Ferain, E., Legras, R., Ounadjela, K., & Fert, A. (1994). Giant Magnetoresistance in Magnetic Multilayered Nanowires. *Ap‐*

[15] Blondel, A., Meier, J. P., Doudin, B., & Ansermet, J. P. (1994). Giant Magnetoresist‐ ance of Nanowires of Multilayers. *Applied Physics Letters*, 65(23), 3019-3021.

[16] Ohgai, T., Enculescu, I., Zet, C., Westerberg, L., Hjort, K., Spohr, R., & Neumann, R. (2006). Magneto-sensitive Nickel Nanowires Fabricated by Electrodeposition into Multi- and Single-ion Track Templates. *Journalof Applied Electrochemistry*, 36(10),

[17] Ohgai, T., Hjort, K., Spohr, R., & Neumann, R. (2008). Electrodeposition of Cobalt Based Ferromagnetic Metal Nanowires in Polycarbonate Films with Cylindrical Nanochannels Fabricated by Heavy-ion-track Etching. *Journalof Applied Electrochemis‐*

[18] Kawai, S., & Ueda, R. (1975). Magnetic Properties of Anodic Oxide Coatings on Alu‐ minum Containing Electrodeposited Co and Co-Ni. *Journal ofThe Electrochemical Soci‐*

[19] Kawai, S. (1975). Retentivities of Anodic Oxide Coatings on Aluminum Containing Electrodeposited Cobalt, Nickel, and Cobalt-Nickel. *Journal ofThe Electrochemical Soci‐*

*Nuclear Instruments and Methods in Physics Research B*, 268(6), 676-686.

T Lym‐

*chemistry*, 35(5), 479-485.

122 Nanowires - Recent Advances

phocytes. *Applied Physics Letters*.

*vanced Materials*, 3(9), 457-459.

*plied Physics Letters*, 65(19), 484-486.

1157-1162.

*try*, 38(5), 713-719.

*ety*, 122(1), 32-36.

*ety*, 122(8), 1026-1029.

wires. *Journal of Applied Physics*, 101(5), 053508.

trochemical and Solid-State Letters ., 10(12), K63-K66.


[45] Li, A. P., Müller, F., Birner, A., Nielsch, K., & Gösele, U. (1998). Hexagonal Pore Ar‐ rays with a 50-420 nm Interpore Distance Formed by Self-organization in Anodic Alumina. *Journal of Applied Physics*, 84(11), 6023-6026.

[32] Ohgai, T., Hoffer, X., Gravier, L., & Ansermet, J. P. (2004). Electrochemical Surface Modification of Aluminum Sheets for Application to Nanoelectronic Devices: Anodi‐ zation Aluminum and Electrodeposition of Cobalt-Copper. *Journalof Applied Electro‐*

[33] Wu, M. T., Leu, I. C., Yen, J. H., & Hon, M. H. (2004). Preparation of Ni Nanodot and Nanowire Arrays Using Porous Alumina on Silicon as a Template without a Conduc‐

[34] Chu, S. Z., Inoue, S., Wada, K., Kanke, Y., & Kurashima, K. (2005). Fabrication and Characterization of Integrated Ultrahigh-density Fe-Pt Alloy Nanowire Arrays on

[35] Wang, X. W., Fei, G. T., Chen, L., X.u, X. J., & Zhang, L. D. (2007). Orientation-con‐ trollable Growth of Ni Nanowire Arrays with Different Diameters. *Electrochemical*

[36] Friedman, A. L., & Menon, L. (2007). Optimal Parameters for Synthesis of Magnetic Nanowires in Porous Alumina Templates. *Journal ofThe Electrochemical Society*, 154(4),

[37] Li, F., & Metzger, R. M. (1997). Activation Volume of α-Fe Particles in Alumite Films.

[38] Zeng, H., Zheng, M., Skomski, R., Sellmyer, D. J., Liu, Y., Menon, L., & Bandyopad‐ hyay, S. (2000). Magnetic Properties of Self-assembled Co Nanowires of Varying

[39] Peng, Y., Zhang, H. L., Pan, S. L., & Li, H. L. (2000). Magnetic Properties and Magnet‐ ization Reversal of α-Fe Nanowires Deposited in Alumina Film. *Journal of Applied*

[40] Yin, A. J., Li, J., Jian, W., Bennett, A. J., & Xu, J. M. (2001). Fabrication of Highly Or‐ dered Metallic Nanowire Arrays by Electrodeposition. *Applied Physics Letters*, 79(7),

[41] Masuda, H., & Fukuda, K. (1995). Ordered Metal Nanohole Arrays Made by a Two-Step Replication of Honeycomb Structures of Anodic Alumina. *Science*, 268(5216),

[42] Masuda, H., & Satoh, M. (1996). Fabrication of Gold Nanodot Array Using Anodic Porous Alumina as an Evaporation Mask. *Japanese Journal of Applied Physics*, 35, L126-

[43] Masuda, H., Yamada, H., Satoh, M., Asoh, H., Nakao, M., & Tamamura, T. (1997). Highly Ordered Nanochannel-array Architecture in Anodic Alumina. *Applied Physics*

[44] Jessensky, O., Müller, F., & Gösele, U. (1998). Self-organized Formation of Hexagonal Pore Arrays in Anodic Alumina. *Applied Physics Letters*, 72(10), 1173-1175.

Length and Diameter. *Journal of Applied Physics*, 87(9), 4718-4720.

tive Interlayer. *Electrochemical and Solid-State Letters*, 7(5), C61-C63.

Glass. *Journal ofThe Electrochemical Society*, 152(1), C42-C47.

*chemistry*, 34(10), 1007-1012.

*and Solid-State Letters*, 10(4), E 1-E3.

*Journal of Applied Physics*, 81(8), 3806-3808.

*Physics*, 87(10), 7405-7408.

*Letters*, 71(19), 2770-2772.

E 68-E70.

124 Nanowires - Recent Advances

1039-1041.

1466-1468.

L129.


## **Nanowires on a Film for Photoelectrochemical Water Splitting**

Miao Zhong, Mario Kurniawan, Aleksandra Apostoluk, Bertrand Vilquin and Jean-Jacques Delaunay

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52593

## **1. Introduction**

Nanowires-on-a-film structure (NFS) consisting of a dense array of semiconductor nano‐ wires on a conductive film has recently attracted much interest due to their wide range of applications in electronic, optoelectronic, electromechanic and electrochemical devices [15, 5, 25]. Thanks to their 3D geometry, NFSs provide various morphology-induced advantages resulting in improved devices performances. In addition, as the array of semiconductor NFSs can be grown on a conductive film, which realizes electrical connection between the nanowires, the integration of this nanostructured layer into devices is greatly facilitated.

In this chapter, we review the applications of NFSs as efficient photoanodes in photoelectro‐ chemical (PEC) water splitting cells. Recently, the efficiency of the PEC cells has been report‐ ed to be enhanced with the use of semiconductor NFSs [23, 10, 30, 32]. The well-aligned geometry of a nanowire array serves as an anti-reflector in order to enhance light absorption in the structure. Also, an effective separation of the photo-generated carriers is achieved along the transverse direction of the nanowires due to the drastically reduced transport dis‐ tance of minority carriers. Furthermore, the nanowire array possesses large surface areas that greatly enhance the chemical reaction of water splitting as compared to its smooth film counterpart. Lastly, the underneath film realizes electrical connection between the top nano‐ wires. As a consequence, the semiconductor NFS offers a great potential for PEC water split‐ ting applications. However, the fabrication of high quality NFS with functional semiconductor materials still remains a challenge. Single crystalline functional semiconduc‐

© 2012 Zhong et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Zhong et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

tor materials will ensure the photoanodes to have adequate semiconducting and electro‐ chemical properties, leading to fulfill the PEC water splitting requirements [14, 2], including:


Furthermore, a controllable approach for the fabrication of the NFSs that enables production of large-area photoanodes at a low cost is not far from being achieved [34].

In the following sections, emphases are placed on the synthesis, property and PEC perform‐ ance of the NFSs fabricated using various materials. The materials under review include TiO2, Fe2O3 and ZnO, as they have the potential to achieve good performance in PEC water splitting. In addition, surface coating of the ZnO NFS with anti-photocorrosive ZnGa2O4 photocatalyst layer is discussed as a mean to improve the stability of the ZnO based photo‐ anodes. Finally, the remaining challenges for the NFS to be used as efficient photoelectrodes for PEC water splitting are presented. It should be noted that the NFS described here are based on arrays of one-dimensional (1D) nanowires, however, nanostructured photoanodes based on nanotubes, nanoplates and other nanostructures on a film should also provide im‐ proved PEC device performance.

### **2. TiO2 NFSs for PEC water splitting**

Since the phenomenon of water electrolysis on a titanium dioxide (TiO2) electrode under UV light illumination was discovered by [9], TiO2 has been considered as one of the most prom‐ ising photocatalysts for energy conversion applications. Later in 1977, Frank and Bard ach‐ ieved the decomposition of cyanide in water using TiO2 material under UV light illumination [7]. This achievement further confirmed the photocatalytic behavior of TiO2 material and leaded for the use of TiO2 in environmental and pollution-degradation applica‐ tions. Recently, extensive research has been conducted on the TiO2 nanostructures in order to enhance device performances [3]. Versatile morphologies including TiO2 nanowires, nanosheets, and nanotubes have been fabricated through different chemical and physical approaches. In many reports, the vertical alignment of the TiO2 nano-architectures perpen‐ dicular to the substrates is found to be a crucial factor necessary to improve device perform‐ ances [8, 10]. In this section, we focus on the synthesis, characterizations and PEC performances of TiO2 NFSs.

## **2.1. Synthesis of the TiO2 NFSs**

tor materials will ensure the photoanodes to have adequate semiconducting and electro‐ chemical properties, leading to fulfill the PEC water splitting requirements [14, 2], including:

Furthermore, a controllable approach for the fabrication of the NFSs that enables production

In the following sections, emphases are placed on the synthesis, property and PEC perform‐ ance of the NFSs fabricated using various materials. The materials under review include TiO2, Fe2O3 and ZnO, as they have the potential to achieve good performance in PEC water splitting. In addition, surface coating of the ZnO NFS with anti-photocorrosive ZnGa2O4 photocatalyst layer is discussed as a mean to improve the stability of the ZnO based photo‐ anodes. Finally, the remaining challenges for the NFS to be used as efficient photoelectrodes for PEC water splitting are presented. It should be noted that the NFS described here are based on arrays of one-dimensional (1D) nanowires, however, nanostructured photoanodes based on nanotubes, nanoplates and other nanostructures on a film should also provide im‐

Since the phenomenon of water electrolysis on a titanium dioxide (TiO2) electrode under UV light illumination was discovered by [9], TiO2 has been considered as one of the most prom‐ ising photocatalysts for energy conversion applications. Later in 1977, Frank and Bard ach‐ ieved the decomposition of cyanide in water using TiO2 material under UV light illumination [7]. This achievement further confirmed the photocatalytic behavior of TiO2 material and leaded for the use of TiO2 in environmental and pollution-degradation applica‐ tions. Recently, extensive research has been conducted on the TiO2 nanostructures in order to enhance device performances [3]. Versatile morphologies including TiO2 nanowires, nanosheets, and nanotubes have been fabricated through different chemical and physical approaches. In many reports, the vertical alignment of the TiO2 nano-architectures perpen‐ dicular to the substrates is found to be a crucial factor necessary to improve device perform‐ ances [8, 10]. In this section, we focus on the synthesis, characterizations and PEC

of large-area photoanodes at a low cost is not far from being achieved [34].

**•** Efficient light absorption

128 Nanowires - Recent Advances

proved PEC device performance.

performances of TiO2 NFSs.

**2. TiO2 NFSs for PEC water splitting**

**•** Effective carrier separation and transport

**•** Suitable energy band position for redox reaction of water

**•** Anti-photocorrosive property in the presence of water.

Titanium (Ti) is a relatively refractory metal with a high melting temperature of about 1670 °C under air condition. This high melting temperature results in a low Ti vapor pressure of ~ 10−3 Torr at 1577 °C, which hinders the TiO2 nucleation in the direct Ti vapor-phase deposi‐ tion processes such as the physical vapor deposition (PVD) and the thermal chemical vapor deposition (CVD). In order to increase the Ti precursor concentration for the growth of 1D TiO2 nanostructures, hydrothermal growth, sol-gel growth and metal organic chemical va‐ por deposition (MOCVD) using Ti organic compounds as precursors are often employed [3].

The fabrication of TiO2 NFS via the hydrothermal growth method has been proved to be ef‐ ficient due to its low-temperature processing (which helps to protect the bottom conductive film during the growth of 1D nanowires), mass production of the nanowires, and their easi‐ ly controllable alignment. As reported by [10], a densely packed, vertically aligned and sin‐ gle crystalline TiO2 rutile nanowire array was synthesized on a transparent conductive layer consisting of fluorine-doped tin oxide (FTO) via the hydrothermal method. The fabrication process is described in the following.

The FTO coated glass substrate was first cleaned by sonication in a mixture of ethanol and water for 30 minutes in order to remove the impurities on the surface of the substrate. Then, a seed layer of TiO2 was synthesized on the cleaned FTO coated glass substrate by a firststep hydrothermal growth. The FTO substrate was soaked in a 0.025 M TiCl4 with n-hexane solution for 30 minutes. The sample was then rinsed with ethanol and finally annealed at 500° C for 30 minutes in air to form a thin layer of TiO2 on the surface of the substrate. The formation of the TiO2-seeded layer on the FTO substrate improved the control and the integ‐ rity of the later TiO2 nanowires growth. Moreover, the TiO2-seeded layer on the FTO layer prevented the formation of electrical short circuits between the electrode and the electrolyte when used as photoanode in the PEC water splitting cell, which guaranteed the photocata‐ lytic function of the TiO2 photoanode. In the next fabrication step, the TiO2 nanowires were grown on the TiO2-seeded FTO substrate with a typical hydrothermal growth procedure. The seeded FTO substrate was placed in sealed Teflon-lined autoclave containing 50 mL of n-hexane, 5 mL of HCL and 2.5-5 mL of titanium isopropoxide. After the reaction was per‐ formed at 150°C for a certain amount of time, the autoclave was cooled down to room tem‐ perature and finally the TiO2 NFS was synthesized.

### **2.2. Characterization of the TiO2 NFSs**

Cross-sectional and top view scanning electron microscope (SEM) images of the synthesized TiO2 nanowires array are shown in Fig. 1 a and 1b [10]. The TiO<sup>2</sup> nanowires array consisting of vertically aligned and tetragonal shaped nanowires is highly uniform and densely packed on the FTO coated glass substrate. The average diameter of the nanowires is ~5 nm and their length is up to ~4.5 μm. The length of the nanowires can be tuned from ~0.5 to ~5 μm, by modifying the synthesis conditions including the precursors ratio (titanium isopropoxide/nhexane), the reaction time and the thickness of TiO2 seed layer. [10]

**Figure 1.** Vertically aligned rutile TiO2 nanowires. (a) Cross-sectional SEM image, (b) top view SEM image, (c) HRTEM image and (d) GIXRD pattern. Reprinted with permission from [Son 10]. Copyright © 2012, American Chemical Society.

The synthesized TiO2 nanowires array was characterized by grazing incidence angle X-ray diffraction (GIXRD) technique as shown in Fig. 1 d. The TiO2 nanowires structure was iden‐ tified as a rutile crystal from the reference XRD data measured with standard rutile TiO2 powder. Furthermore, the high resolution transmission electron (HRTEM) image in Fig. 1d showed that the observed rutile phase TiO2 nanowire is of a single crystal quality, having clear (110) lattice fringes.

#### **2.3. PEC performance of the TiO2 NFSs**

The PEC performance of the synthesized TiO2 NFS (as previously described) was examined in a three-electrode PEC water splitting cell [10]. The TiO2 NFS was used as the working electrode, a saturated Ag/AgCl electrode as the reference electrode and a platinum wire as the counter electrode. The electrolyte consisted of one mol/liter KOH solution with pH = 13.5. For the PEC test, TiO2 nanowires with an average diameter of ~5 nm and average length of ~1.5 μm [10] were studied. The linear sweep voltammetry measurement was per‐ formed under a 100 W Xenon lamp illumination (300 nm < wavelength λ < 800 nm) with a bias swept from -0.8 to 0.6 V (vs. Ag/AgCl), as shown in Fig. 2 a. A photocurrent density of ~0.35 mA/cm2 was obtained at a bias of 1.23 VRHE [10].

**Figure 2.** a) Linear sweep voltammetry measurements of the TiO2 NFS with and without Co treatment, (b) Chronoamperometry measurements of the TiO2 NFS, Co treated TiO2 NFS and Ag treated TiO2 NFS at 1.23 VRHE. Reprinted with permission from [Son 10]. Copyright © 2012, American Chemical Society.

As reported by Tilley et al. [23, 10], decoration of a photoelectrode with metal nanoparticles as co-catalysts exhibited an enhanced yield in hydrogen generation. [10], reported treat‐ ments by cobalt (Co) and silver (Ag) performed on the TiO2 NFS samples. The photocurrent density of ~0.35 mA/cm2 measured on the undoped TiO2 NFS increased to ~0.45 mA/cm2 for the Co and Ag coated TiO2 NFSs, as shown in Fig. 2 b. The enhancement of the photocurrent after Co and Ag treatments was likely related to the improvement in the extraction of photo‐ generated holes from the surface of the TiO2 NFS to the electrolyte, through the Co and Ag nanoparticles [23, 10].

### **3. Hematite NFS for PEC water splitting**

**Figure 1.** Vertically aligned rutile TiO2 nanowires. (a) Cross-sectional SEM image, (b) top view SEM image, (c) HRTEM image and (d) GIXRD pattern. Reprinted with permission from [Son 10]. Copyright © 2012, American Chemical Society.

The synthesized TiO2 nanowires array was characterized by grazing incidence angle X-ray diffraction (GIXRD) technique as shown in Fig. 1 d. The TiO2 nanowires structure was iden‐ tified as a rutile crystal from the reference XRD data measured with standard rutile TiO2 powder. Furthermore, the high resolution transmission electron (HRTEM) image in Fig. 1d showed that the observed rutile phase TiO2 nanowire is of a single crystal quality, having

The PEC performance of the synthesized TiO2 NFS (as previously described) was examined in a three-electrode PEC water splitting cell [10]. The TiO2 NFS was used as the working electrode, a saturated Ag/AgCl electrode as the reference electrode and a platinum wire as the counter electrode. The electrolyte consisted of one mol/liter KOH solution with pH = 13.5. For the PEC test, TiO2 nanowires with an average diameter of ~5 nm and average length of ~1.5 μm [10] were studied. The linear sweep voltammetry measurement was per‐ formed under a 100 W Xenon lamp illumination (300 nm < wavelength λ < 800 nm) with a bias swept from -0.8 to 0.6 V (vs. Ag/AgCl), as shown in Fig. 2 a. A photocurrent density of

was obtained at a bias of 1.23 VRHE [10].

clear (110) lattice fringes.

130 Nanowires - Recent Advances

~0.35 mA/cm2

**2.3. PEC performance of the TiO2 NFSs**

Hematite (α-Fe2O3 or iron oxide) has been widely investigated as a photocatalyst candidate for PEC water splitting [13, 14, 30] because of its abundance on Earth, its non-toxicity to en‐ vironment and its stability in water under light illumination. In addition, a band gap be‐ tween 2.1 and 2.2 eV of the hematite enables visible light absorption and thus makes this material promising for water splitting under solar light illumination. However, hematite has also some disadvantages such as: small light absorption coefficient, short lifetime of the ex‐ cited-state carrier (10-12 s), short hole diffusion length (2-4 nm), inefficient oxygen evolution reaction kinetics and improper position of the conduction band for hydrogen generation [23, 16], limiting the hematite PEC performance. Different research groups made many efforts to address these issues [23, 14, 16, 10]. To date, improved performances have been realized in the following areas:


In the following section, we summarized the results obtained with the silicon-doped hema‐ tite nanowire-like array on a conductive film structure (Si-doped NFS-like hematite) deco‐ rated with IrO2 nanoparticles [23] for PEC water splitting.

### **3.1. Synthesis of the Si-doped NFS-like hematite with IrO2 surface decoration**

Similarly to titanium, iron (Fe) has a relatively high melting temperature of ~1550 °C under air condition and thus hematite nanostructures growth by simple thermal CVD or PVD using solid-phase Fe-compound precursors remains difficult. Instead, solution-phase precursors, such as Fe(CO)5, Fe(AcAc)3 and FeCl3, are used to provide sufficient amount of Fe precur‐ sors in the growth phase for the synthesis of hematite nanostructures via atmospheric pres‐ sure chemical vapor deposition (APCVD) or hydrothermal growth methods [23, 14, 16, 10].

However, the undoped hematite nanostructure is inactive or very inefficient when used as a photoanode for PEC water splitting, because of its low conductivity and improper conduc‐ tion band position for hydrogen generation. To improve the conductivity, hematite has been doped by different elements, such as Si, Ti, Sn, Al, Mg and Zn, leading to increase the elec‐ tron density. To address the issue of improper conduction band position of hematite, hema‐ tite surface decoration treatments with co-catalysts nanoparticles, such as IrO2 nanoparticles [18], were applied to increase the chemical reactions of water oxidation.

As reported by Tilley et al. [23], Si-doped NFS-like hematite sample coated with IrO2 nano‐ particles was fabricated and a large photocurrent density of ~3 mA/cm2 was obtained at a bias of 1.23 VRHE under AM 1.5 illumination condition. In their synthesis process, a FTOcoated glass was used as the substrate for the NFS growth and put on an aluminum block heater. The substrate was maintained at 415 °C during the APCVD growth process. The pre‐ cursors Fe(CO)5 and tetraethoxysilane (TEOS) were carried to the FTO coated glass substrate by bubbling argon gas at fixed flow rates, calculated to be 3.5 mg/min for Fe(CO)5 and 0.32 mg/min for TEOS. These two streams were mixed with an air flow of 2 L/min and vertically sprayed to the substrate. After 5 minutes, NFS-like Si-doped hematite was synthesized. Ac‐ cording to Tilley's reports, the preparation of homogeneous films on large substrates re‐ quired a continuously rotating substrate in a rotating belt APCVD reactor. Moreover, the improved PEC performance with Si-doped NFS-like hematite photoanode needed a first growth of an ultrathin SiO2 interfacial layer of ~1 nm between hematite and FTO. It is very likely that this silica layer worked as an insulator layer to prevent the current leaking from the FTO layer to the electrolyte during the PEC experiments. Finally, the fabricated Si-doped NFS-like hematite was coated with ~2 nm IrO2 nanoparticles using an electrophoresis depo‐ sition process. An increased photocurrent of ~3 mA/cm2 was achieved with the IrO2 nano‐ particles coating, to be compared with a photocurrent of ~2 mA/cm2 obtained for the Sidoped NFS-like hematite without co-catalysts [23].

### **3.2. Characterization of the Si-doped NFS-like hematite with IrO2 decoration**

In the following section, we summarized the results obtained with the silicon-doped hema‐ tite nanowire-like array on a conductive film structure (Si-doped NFS-like hematite) deco‐

Similarly to titanium, iron (Fe) has a relatively high melting temperature of ~1550 °C under air condition and thus hematite nanostructures growth by simple thermal CVD or PVD using solid-phase Fe-compound precursors remains difficult. Instead, solution-phase precursors, such as Fe(CO)5, Fe(AcAc)3 and FeCl3, are used to provide sufficient amount of Fe precur‐ sors in the growth phase for the synthesis of hematite nanostructures via atmospheric pres‐ sure chemical vapor deposition (APCVD) or hydrothermal growth methods [23, 14, 16, 10].

However, the undoped hematite nanostructure is inactive or very inefficient when used as a photoanode for PEC water splitting, because of its low conductivity and improper conduc‐ tion band position for hydrogen generation. To improve the conductivity, hematite has been doped by different elements, such as Si, Ti, Sn, Al, Mg and Zn, leading to increase the elec‐ tron density. To address the issue of improper conduction band position of hematite, hema‐ tite surface decoration treatments with co-catalysts nanoparticles, such as IrO2 nanoparticles

As reported by Tilley et al. [23], Si-doped NFS-like hematite sample coated with IrO2 nano‐ particles was fabricated and a large photocurrent density of ~3 mA/cm2 was obtained at a bias of 1.23 VRHE under AM 1.5 illumination condition. In their synthesis process, a FTOcoated glass was used as the substrate for the NFS growth and put on an aluminum block heater. The substrate was maintained at 415 °C during the APCVD growth process. The pre‐ cursors Fe(CO)5 and tetraethoxysilane (TEOS) were carried to the FTO coated glass substrate by bubbling argon gas at fixed flow rates, calculated to be 3.5 mg/min for Fe(CO)5 and 0.32 mg/min for TEOS. These two streams were mixed with an air flow of 2 L/min and vertically sprayed to the substrate. After 5 minutes, NFS-like Si-doped hematite was synthesized. Ac‐ cording to Tilley's reports, the preparation of homogeneous films on large substrates re‐ quired a continuously rotating substrate in a rotating belt APCVD reactor. Moreover, the improved PEC performance with Si-doped NFS-like hematite photoanode needed a first growth of an ultrathin SiO2 interfacial layer of ~1 nm between hematite and FTO. It is very likely that this silica layer worked as an insulator layer to prevent the current leaking from the FTO layer to the electrolyte during the PEC experiments. Finally, the fabricated Si-doped NFS-like hematite was coated with ~2 nm IrO2 nanoparticles using an electrophoresis depo‐

particles coating, to be compared with a photocurrent of ~2 mA/cm2 obtained for the Si-

was achieved with the IrO2 nano‐

**3.1. Synthesis of the Si-doped NFS-like hematite with IrO2 surface decoration**

[18], were applied to increase the chemical reactions of water oxidation.

sition process. An increased photocurrent of ~3 mA/cm2

doped NFS-like hematite without co-catalysts [23].

rated with IrO2 nanoparticles [23] for PEC water splitting.

132 Nanowires - Recent Advances

The cross-sectional SEM images of the synthesized Si-doped NFS-like hematite are shown in Fig. 3 a and 3b [23]. The hematite nanowires with an average height of ~ 500 nm were dense‐ ly packed on the FTO-coated glass substrate. The diameter of the hematite nanowires at the top area varied from 75 to 250 nm. It is observed that the surfaces of hematite nanowires were very rough and composed of nanoparticles with diameters as small as ~ 10-20 nm. As suggested by Tilley et al., the rough surfaces further increased the nanowire/electrolyte in‐ terface areas and also shortened the diffusion length of the photogenerated holes to the in‐ terface areas, contributing to the increase of PEC water splitting performance [23].

**Figure 3.** Cross-sectional SEM images of the dense and rough Si-doped NSF-like hematite. [23] Copyright John Wiley and Sons.

The microstructure and crystal quality of the NFS-like hematite samples with/without Si doping were further studied by XRD 2theta-omega scan technique. The XRD results are shown in Fig. 4. [13]. The undoped and Si-doped NFS-like hematite samples did not con‐ tain any other iron oxide phases than hematite, indicating a good experimental control over the hematite synthesis. The fact that the synthesized nanostructures are made of a singlephase hematite is important for the PEC water splitting applications, because other iron oxide phases such as Fe3O4 always contain strong electron-hole recombination centers, thus leading to lifetime degradation of the holes used in the oxidation of water [13]. Another important feature shown in the XRD results is that the crystal orientation of the synthe‐ sized hematite varied with the synthesis conditions. In the case of the undoped NSF-like hematite sample, there is no dominant crystal orientation. The intensities of the different Xray diffraction peaks obtained with the undoped NSF-like hematite sample (see Fig. 4b) are similar to the XRD reference data measured on the randomly oriented hematite powders (see Fig. 4c). In contrast, the Si-doped NFS-like hematite sample showed a clear preferen‐ tial [110] crystal orientation. These [110] oriented hematite nanowires likely have an en‐ hanced conductivity compared to the randomly oriented hematite nanowires. In fact, the conductivity along the hematite [110] orientation was calculated to be of roughly four or‐ ders of magnitude higher than the conductivity along the other crystal orientations perpen‐ dicular to the [110] in hematite [12]. Therefore, the Si-doped NFS-like hematite with a preferential [110] orientation reported by Tilley et al. achieved a benchmark high photocur‐ rent for PEC water splitting.

**Figure 4.** X-ray diffraction pattern of (a) Si-doped NFS-like hematite, (b) undoped NFS-like hematite, (c) standard pow‐ ders of the hematite (black lines) and the SnO2 (blue lines). Reprinted with permission from [13]. Copyright © 2006, American Chemical Society.

Finally, the synthesized Si-doped NFS-like hematite was coated with IrO2 nanoparticles (~2 nm in diameter) by the electrophoresis deposition process [23]. The X-ray photoelectron spectroscopy (XPS) results showed an iridium concentration of ~1% on the surface of the IrO2 coated sample, confirming the presence of IrO2 [23].

### **3.3. PEC performance of the Si-doped NFS-like Hematite with IrO2 decoration**

The PEC performance of the IrO2 decorated Si-doped NFS-like hematite was studied using a three-electrode PEC cell [23]. In the reported experiment, the Si-doped NFS-like hematite sample was used as the working electrode, an Ag/AgCl in a saturated KCl solution electrode was used as the reference electrode and a platinum wire was used as the counter electrode. One mole/liter NaOH solution with pH = 13.6 was used as the electrolyte. The sample was irradiated with the standard AM1.5 G 100 mW/cm2 solar simulator.

The chrono-amperometry measurement of the hematite photoanode at a bias of 1.23 VRHE with light on-off cycles shows that an on-off current behavior corresponding to the light onoff cycles is clearly observed (Fig. 5). The low dark current revealed that no/few chemical reactions occurred on the hematite photoanode without light illumination. A high photocur‐ rent density of ~3 mA/cm2 was obtained at a bias of 1.23 VRHE, as shown in Fig. 5. This photo‐ current decreased within the 200-second chrono-amperometry scan. As suggested by Tilley et al., the photocurrent decrease was related to the detachment of the IrO2 nanoparticles from the surfaces of the Si-doped NFS-like hematite due to a weak chemical bonding be‐ tween the IrO2 and the hematite. More robust method to attach the IrO2 nanoparticles on hematite should further improve the PEC results, as was suggested by Tilley et al. [23]

**Figure 5.** Chrono-amperometry measuremens of the IrO2 decorated Si-doped NFS-like hematite photoanode at a bias of 1.23 VRHE under AM 1.5 G 100 mW/cm2 light illumination [23] Copyright, John Wiley and Sons.

## **4. ZnO NFS for PEC water splitting**

**Figure 4.** X-ray diffraction pattern of (a) Si-doped NFS-like hematite, (b) undoped NFS-like hematite, (c) standard pow‐ ders of the hematite (black lines) and the SnO2 (blue lines). Reprinted with permission from [13]. Copyright © 2006,

Finally, the synthesized Si-doped NFS-like hematite was coated with IrO2 nanoparticles (~2 nm in diameter) by the electrophoresis deposition process [23]. The X-ray photoelectron spectroscopy (XPS) results showed an iridium concentration of ~1% on the surface of the

The PEC performance of the IrO2 decorated Si-doped NFS-like hematite was studied using a three-electrode PEC cell [23]. In the reported experiment, the Si-doped NFS-like hematite sample was used as the working electrode, an Ag/AgCl in a saturated KCl solution electrode was used as the reference electrode and a platinum wire was used as the counter electrode. One mole/liter NaOH solution with pH = 13.6 was used as the electrolyte. The sample was

The chrono-amperometry measurement of the hematite photoanode at a bias of 1.23 VRHE with light on-off cycles shows that an on-off current behavior corresponding to the light onoff cycles is clearly observed (Fig. 5). The low dark current revealed that no/few chemical reactions occurred on the hematite photoanode without light illumination. A high photocur‐

current decreased within the 200-second chrono-amperometry scan. As suggested by Tilley et al., the photocurrent decrease was related to the detachment of the IrO2 nanoparticles from the surfaces of the Si-doped NFS-like hematite due to a weak chemical bonding be‐ tween the IrO2 and the hematite. More robust method to attach the IrO2 nanoparticles on hematite should further improve the PEC results, as was suggested by Tilley et al. [23]

solar simulator.

was obtained at a bias of 1.23 VRHE, as shown in Fig. 5. This photo‐

**3.3. PEC performance of the Si-doped NFS-like Hematite with IrO2 decoration**

IrO2 coated sample, confirming the presence of IrO2 [23].

irradiated with the standard AM1.5 G 100 mW/cm2

American Chemical Society.

134 Nanowires - Recent Advances

rent density of ~3 mA/cm2

Due to their unique optoelectronic and semiconducting properties, single crystal ZnO densely packed nanowire arrays have been intensively studied in the past few decades for their potential use in electronic, electromechanical, and photonic devices [29, 20, 5, 26, 4, 32]. Recently, much research effort has been devoted to the fabrication of a high quality ZnO nanowire arrays and their application in PEC water splitting with the aim of taking advant‐ age of ZnO material properties such as:


For PEC water splitting applications, an efficient means permitting a good electrical connec‐ tion of the nanowires through a conductive film, namely a ZnO NFS, is required. Yang et al. presented a detailed study on the fabrication and PEC performance of ZnO NFS (ZnO nano‐ wire arrays on ITO film) with nitrogen doping [30]. An increased PEC photocurrent was ob‐ tained with the nitrogen-doped ZnO NFS sample. Wang et al. reported advanced PEC performances with CdS and CdSe quantum dots loaded ZnO NFSs under the conditions of sacrificial reagents [24]. However promising these results may be, further efforts to increase the stability of the ZnO NFSs in water under light illumination are required.

Since ZnO is chemically unstable in water under light irradiation, surface coating of the ZnO NFS with anti-photocorrosive and photocatalytic shell layers is one of the strategies to im‐ prove its chemical stability. However, the search for suitable anti-photocorrosive photocata‐ lysts, which can grow easily on the ZnO nanowire surfaces with a high crystal quality, is a major concern. One possible candidate for the shell material is ZnGa2O4. Indeed, ZnGa2O4 is conductive, has anti-photocorrosive ability in water and photocatalytic property for PEC water splitting [11]. In addition, single crystal quality ZnGa2O4 can be synthesized via hightemperature chemical reactions using ZnO as the precursor material. Thus, surface coating of the ZnO NFS with conductive and anti-photocorrosive ZnGa2O4 shells to form ZnO-ZnGa2O4 core-shell NFS has been investigated [33]. With this structure, stable and efficient water splitting was achieved [32]. Further improvements are expected, including enhanced visible light absorption with doping, and efficient charge transfer to the electrolyte with the co-catalyst decoration of the ZnO nanowires.

In this section, we first review the synthesis of high crystal quality ZnO NFSs using different approaches. The realization of high quality ZnO nanowire arrays with efficient electrical connection is crucial for their subsequent use in high performance PEC cells. Then, the sur‐ face coating of the anti-photocorrosive ZnGa2O4 shells on the ZnO NFS is presented. Finally, the electronic properties and a stable PEC performance of the single crystal ZnO-ZnGa2O4 core-shell NFS are reported in detail.

#### **4.1. Synthesis of the ZnO NFSs**

A variety of physical and chemical approaches have been used to synthesize dense and sin‐ gle crystal ZnO nanowire arrays, including laser interference lithography [27], template-as‐ sisted wet-chemical growth [30, 4], MOCVD [21] and CVD [29, 5, 34]. As evidenced by these reports, the choice of a suitable substrate is of highest importance to synthesize single crystal ZnO nanowires with an aligned geometry.

Sapphire is a chemically stable and relatively cheap ceramic material and thus has been widely used as substrates for the growth of ZnO nanowire arrays in various fabrication processes. Erdélyi, et al. reported the synthesis of a single crystal quality and mono-oriented ZnO nanowire array on a *c*-plane sapphire by the pulsed laser deposition [4]. In their experi‐ ment, the pre-growth of a high quality and size controlled ZnO nanoseeds is needed for the growth of the nanowires. The top-view SEM image of the ZnO nanowire array is shown in Fig. 6 a. Because sapphire is an insulator, the as-grown ZnO nanowires on the sapphire sub‐ strate are not electrically connected, preventing their efficient use in optoelectronic applica‐ tions. Thus, high quality ZnO nanowire arrays grown on conductive films are a prerequisite to the development of advanced nanowire-based materials.

One strategy is to form ZnO NFS on metal substrates. However, the growth of nanowires on metal substrate presents many difficulties. Three major issues should be addressed to realize the functionality of the ZnO NFS using metal substrates. They are as follows:


vertically aligned ZnO nanowires. As a result, the fabrication process becomes intricate, hard to control and expensive.

major concern. One possible candidate for the shell material is ZnGa2O4. Indeed, ZnGa2O4 is conductive, has anti-photocorrosive ability in water and photocatalytic property for PEC water splitting [11]. In addition, single crystal quality ZnGa2O4 can be synthesized via hightemperature chemical reactions using ZnO as the precursor material. Thus, surface coating of the ZnO NFS with conductive and anti-photocorrosive ZnGa2O4 shells to form ZnO-ZnGa2O4 core-shell NFS has been investigated [33]. With this structure, stable and efficient water splitting was achieved [32]. Further improvements are expected, including enhanced visible light absorption with doping, and efficient charge transfer to the electrolyte with the

In this section, we first review the synthesis of high crystal quality ZnO NFSs using different approaches. The realization of high quality ZnO nanowire arrays with efficient electrical connection is crucial for their subsequent use in high performance PEC cells. Then, the sur‐ face coating of the anti-photocorrosive ZnGa2O4 shells on the ZnO NFS is presented. Finally, the electronic properties and a stable PEC performance of the single crystal ZnO-ZnGa2O4

A variety of physical and chemical approaches have been used to synthesize dense and sin‐ gle crystal ZnO nanowire arrays, including laser interference lithography [27], template-as‐ sisted wet-chemical growth [30, 4], MOCVD [21] and CVD [29, 5, 34]. As evidenced by these reports, the choice of a suitable substrate is of highest importance to synthesize single crystal

Sapphire is a chemically stable and relatively cheap ceramic material and thus has been widely used as substrates for the growth of ZnO nanowire arrays in various fabrication processes. Erdélyi, et al. reported the synthesis of a single crystal quality and mono-oriented ZnO nanowire array on a *c*-plane sapphire by the pulsed laser deposition [4]. In their experi‐ ment, the pre-growth of a high quality and size controlled ZnO nanoseeds is needed for the growth of the nanowires. The top-view SEM image of the ZnO nanowire array is shown in Fig. 6 a. Because sapphire is an insulator, the as-grown ZnO nanowires on the sapphire sub‐ strate are not electrically connected, preventing their efficient use in optoelectronic applica‐ tions. Thus, high quality ZnO nanowire arrays grown on conductive films are a prerequisite

One strategy is to form ZnO NFS on metal substrates. However, the growth of nanowires on metal substrate presents many difficulties. Three major issues should be addressed to realize

**1.** In order to avoid the formation of a Schottky barrier at the interfaces between the ZnO nanowires and the metal substrate, the work function of the metal should be lower than the Fermi level of the ZnO. This requirement severely limits the possible candidates for

**2.** Metal substrates usually have large lattice mismatches with ZnO and, therefore, a pregrowth of lattice matched ZnO nanoseeds on the metal substrate is necessary to obtain

the functionality of the ZnO NFS using metal substrates. They are as follows:

co-catalyst decoration of the ZnO nanowires.

core-shell NFS are reported in detail.

ZnO nanowires with an aligned geometry.

to the development of advanced nanowire-based materials.

**4.1. Synthesis of the ZnO NFSs**

136 Nanowires - Recent Advances

the metal substrates.


In summary, the ZnO nanowire array on a metal substrate is still difficult to fabricate with adequate functionalities for use as a photoanode in a PEC cell.

**Figure 6.** Top-view SEM images of vertically aligned ZnO nanowires on the (a) *c*-plane sapphire. Reprinted with per‐ mission from [4]. Copyright © 2011, American Chemical Society. (b) GaN substrate. Reprinted with permission from [27]. Copyright © 20101, American Chemical Society. (c) GaN substrate. [27] Copyright, John Wiley and Sons. (d) high quality ZnO film using *a*-plane sapphire as substrate. [34]

A different strategy is to grow ZnO nanowire arrays on conductive semiconductor sub‐ strates. GaN matches the in-plane geometry and lattice constants with ZnO and, therefore, has been employed as a high-quality epitaxial substrate for the growth of ZnO nanowire ar‐ rays [5, 27]. Fan et al. and Wei et al. synthesized orderly aligned, size-controlled and single crystalline quality ZnO nanowire arrays using single crystal GaN as substrates via a tem‐ plate-assisted growth process. In Fan's method, a porous anodic alumina membrane was used as a template mask to control the gold catalysts deposition on the GaN substrate, whereas in Wei's method, laser interference lithography technique was used for a pregrowth of high quality ZnO nanoseeds on the GaN substrate. According to their reports, the diameters and distributions of the synthesized ZnO nanowires are governed by the parame‐ ters of the porous alumina membrane and the laser interference lithography techniques, re‐ spectively. The top-view SEM images of the synthesized ZnO nanowires are shown in Fig 6 b and 6c. Advanced piezo-electric performances have been reported for the fabricated ZnO NFSs. For the use in PEC applications, special attention should be paid to the interface de‐ fects and the barrier formation between the ZnO and the GaN in order to realize an efficient electrical charge collection.

The fabrication of a high quality ZnO nanowire array on a ZnO film (ZnO NFS) using a sin‐ gle-step CVD process has been recently reported by our group [34]. The synthesis process enables the growth of a dense ZnO nanowires array on a ZnO film. The nanowires, as well as the film, have a single crystal domain texture quality and the nanowires are perfectly aligned over a large area, which comes as the result of the single crystal domain texture. The top-view SEM image of the fabricated array is shown in Fig 6 d. Efficient electrical connec‐ tion is realized because nanowires and the underneath film layer are made of the same ma‐ terial ZnO [34]. Compared to other growth processes of ZnO NFSs, our one-step CVD process is easily controlled and cheap. However, the diameters and distributions of the ZnO nanowires cannot be as precisely controlled as in the template-assisted growth processes. The synthesized ZnO NFS offers an ideal template for the subsequent synthesis of a single crystal ZnO-ZnGa2O4 core-shell NFS, which provides anti-photocorrosive property for the PEC applications.

## **4.2. Synthesis of the ZnO-ZnGa2O4 core-shell NFSs**

ZnGa2O4 is a ternary metal oxide compound with a complicated crystal structure: one con‐ ventional ZnGa2O4 unit cell is composed of 8 Zn atoms, 16 Ga atoms and 32 O atoms, ar‐ ranged into a spinel structure [6]. The direct fabrication of ZnGa2O4 nanowire array with a single crystal quality is very difficult and has not been reported yet. Thus, a two-step proc‐ ess to grow dense ZnO-ZnGa2O4 core-shell NFS has been proposed recently [33]:


A high quality ZnO NFS consisting of dense and vertically aligned nanowires on the ZnO film was first fabricated using an *a*-plane sapphire substrate through a simple thermal CVD process [32]. In this process, Au-coated *a*-plane sapphire was selected as the growth sub‐ strate. ZnO powder and graphite powder mixed with a 2:1 weight ratio were used as source material and placed at 1 cm upstream position away from the substrate in a tubular furnace. The furnace was operated at 1000 ºC for 30 minutes with the argon and oxygen gas mixture (5:1 in volume ratio) flowing through the tube at the working pressure of 50 mbar. At the end of the synthesis, the furnace cooled down naturally to the room temperature. The syn‐ thesized product consisted of a dense nanowires array on a ZnO film (ZnO NFS). Such a structure offers several advantages, namely,


In a second step, a CVD growth was performed on the ZnO NFS sample. The sample was put in a furnace at a temperature of 900ºC. Mixed precursors of Ga2O3, ZnO and graphite were placed at 11 cm upstream position from the sample. The temperature of the precursor zone was set to 1180 ºC. The carrier gases consisted of the argon and oxygen (10:1 in volume ratio) at the working pressure of 50 mbar. After a 30 minutes-long growth, the ZnO-ZnGa2O4 core-shell NFS was obtained [32].

## **4.3. Characterization of ZnO-ZnGa2O4 core-shell NFSs**

used as a template mask to control the gold catalysts deposition on the GaN substrate, whereas in Wei's method, laser interference lithography technique was used for a pregrowth of high quality ZnO nanoseeds on the GaN substrate. According to their reports, the diameters and distributions of the synthesized ZnO nanowires are governed by the parame‐ ters of the porous alumina membrane and the laser interference lithography techniques, re‐ spectively. The top-view SEM images of the synthesized ZnO nanowires are shown in Fig 6 b and 6c. Advanced piezo-electric performances have been reported for the fabricated ZnO NFSs. For the use in PEC applications, special attention should be paid to the interface de‐ fects and the barrier formation between the ZnO and the GaN in order to realize an efficient

The fabrication of a high quality ZnO nanowire array on a ZnO film (ZnO NFS) using a sin‐ gle-step CVD process has been recently reported by our group [34]. The synthesis process enables the growth of a dense ZnO nanowires array on a ZnO film. The nanowires, as well as the film, have a single crystal domain texture quality and the nanowires are perfectly aligned over a large area, which comes as the result of the single crystal domain texture. The top-view SEM image of the fabricated array is shown in Fig 6 d. Efficient electrical connec‐ tion is realized because nanowires and the underneath film layer are made of the same ma‐ terial ZnO [34]. Compared to other growth processes of ZnO NFSs, our one-step CVD process is easily controlled and cheap. However, the diameters and distributions of the ZnO nanowires cannot be as precisely controlled as in the template-assisted growth processes. The synthesized ZnO NFS offers an ideal template for the subsequent synthesis of a single crystal ZnO-ZnGa2O4 core-shell NFS, which provides anti-photocorrosive property for the

ZnGa2O4 is a ternary metal oxide compound with a complicated crystal structure: one con‐ ventional ZnGa2O4 unit cell is composed of 8 Zn atoms, 16 Ga atoms and 32 O atoms, ar‐ ranged into a spinel structure [6]. The direct fabrication of ZnGa2O4 nanowire array with a single crystal quality is very difficult and has not been reported yet. Thus, a two-step proc‐

ess to grow dense ZnO-ZnGa2O4 core-shell NFS has been proposed recently [33]:

**2.** surface coating of the ZnO NFS with the ZnGa2O4 shells (Fig. 7).

**Figure 7.** Schematic illustration of the fabrication process of the ZnO-ZnGa2O4 core-shell NFS. [32]

electrical charge collection.

138 Nanowires - Recent Advances

PEC applications.

**4.2. Synthesis of the ZnO-ZnGa2O4 core-shell NFSs**

**1.** synthesis of the high quality ZnO NFS and

The morphologies of the synthesized ZnO and ZnO-ZnGa2O4 core-shell NFSs are presented in Fig. 8. The tilted-angle SEM image of Fig. 8a shows a highly uniform and densely ZnO nanowire array grown over a large area on the Au-coated *a*-plane sapphire substrate. The inset in Fig. 8a is an enlarged SEM image revealing the vertically aligned nanowires. The vertical alignment and smooth sidewalls of the nanowires can be clearly observed. The aver‐ age diameter of the nanowires is ~70 nm and the average length is about 1 μm. Energy-dis‐ persive X-ray analyses (EDX) performed on the nanowire array gave a Zn/O atomic ratio of ~1/1, indicating the formation of the ZnO nanowires [34].

The tilted-angle SEM image of Fig. 8b shows a densely packed array of ZnO-ZnGa2O4 coreshell nanowires homogeneously arranged over large areas. The inset in Fig. 8b is an en‐ larged SEM image of the vertically aligned ZnO-ZnGa2O4 core-shell nanowires. The initial vertical alignment of the ZnO nanowires is well maintained in the ZnO-ZnGa2O4 core-shell nanowire sample. The average diameter of the ZnO-ZnGa2O4 core-shell nanowires is ~100 nm. EDX analysis performed on the surfaces of the core-shell nanowire gave a Zn/Ga/O atomic ratio of ~14/28/58 (~1/2/4), indicating the formation of the ZnGa2O4 shells. [32] In this work, a nanowires-on-a-film structure

**Figure 8.** Tilted-angle SEM image of the dense and vertically aligned (a) ZnO and (b) ZnO-ZnGa2O4 nanowire arrays on the *a*-plane sapphire substrate. Insets in (a) and (b) are close-up images of the vertically aligned ZnO and ZnO-ZnGa2O4 nanowires. [32 and 34]

TEM and electron diffraction (ED) were used to analyze the structure and the crystalline quality of the ZnO and ZnO-ZnGa2O4 core-shell NFSs. The cross-sectional high-angle annular darkfield (HAADF) TEM image of the ZnO NFS is shown in Fig. 9a (left image). The average length of the nanowires is about 1 μm and the thickness of the film is about 2.5 μm. Such a thick ZnO bottom layer realized the electrical connection of the nanowire array. The energy dispersive spectroscopy (EDS) mapping images presented in Fig. 9a (the central image) clearly reveal that the NFS is made of ZnO and fabricated on the *a*-plane sapphire substrate.

ED analyses were further performed to characterize the crystal quality and the crystallo‐ graphic orientation of the ZnO NFS sample. The ED pattern with the incident zone axis of [1-100] (see the right image in Fig. 9a) shows that the ZnO nanowire array and ZnO thick film are all single crystalline and well-aligned along the wurtzite [0001] direction. Such a single crystal and unique [0001] oriented ZnO NFS offers an ideal substrate for the subse‐ quent growth of the ZnO-ZnGa2O4 core-shell NFS.

The TEM and ED results of the ZnO-ZnGa2O4 core-shell NFS are shown in Fig. 9b [33]. The bright-field TEM image reveals the nanowire core-shell structure with a different brightness contrast. The ED pattern of the shell region marked as the square A in the fig. 9b presented in the inset A of this figure shows that the shell part of the nanowire consists of a single-crystal‐ line spinel ZnGa2O4 with a [111] direction parallel to the long axial direction of the nano‐ wire. The ED pattern of the core region of B (square B, the ED pattern is shown in the inset B) confirms that the core part of the nanowire is a single crystal wurtzite ZnO with a [0002] direction parallel to the long axial direction of the nanowire. Furthermore, the ED pattern taken on the interfacial region of C (square C, the ED pattern is shown in the inset C) clearly shows two sets of single-crystalline electron diffraction dot patterns, indexed to the spinel ZnGa2O4 and the wurtzite ZnO. From the electron diffraction pattern, it is found that the ZnGa2O4 [111] direction is parallel to the ZnO [0002] direction and the ZnGa2O4 [2-20] direction is parallel to the ZnO [-2110] direction. An epitaxial relationship between the ZnGa2O4 (11-2) plane and the ZnO (1-100) plane is therefore predicted. Thus the subsequent growth of ZnGa2O4 on the ZnO NFS results in a hetero-epitaxial growth of the ZnGa2O4 shells on the ZnO cores. The final product is well-aligned ZnO-ZnGa2O4 core-shell nanowires.

**Figure 8.** Tilted-angle SEM image of the dense and vertically aligned (a) ZnO and (b) ZnO-ZnGa2O4 nanowire arrays on the *a*-plane sapphire substrate. Insets in (a) and (b) are close-up images of the vertically aligned ZnO and ZnO-ZnGa2O4

TEM and electron diffraction (ED) were used to analyze the structure and the crystalline quality of the ZnO and ZnO-ZnGa2O4 core-shell NFSs. The cross-sectional high-angle annular darkfield (HAADF) TEM image of the ZnO NFS is shown in Fig. 9a (left image). The average length of the nanowires is about 1 μm and the thickness of the film is about 2.5 μm. Such a thick ZnO bottom layer realized the electrical connection of the nanowire array. The energy dispersive spectroscopy (EDS) mapping images presented in Fig. 9a (the central image) clearly reveal

ED analyses were further performed to characterize the crystal quality and the crystallo‐ graphic orientation of the ZnO NFS sample. The ED pattern with the incident zone axis of [1-100] (see the right image in Fig. 9a) shows that the ZnO nanowire array and ZnO thick film are all single crystalline and well-aligned along the wurtzite [0001] direction. Such a single crystal and unique [0001] oriented ZnO NFS offers an ideal substrate for the subse‐

The TEM and ED results of the ZnO-ZnGa2O4 core-shell NFS are shown in Fig. 9b [33]. The bright-field TEM image reveals the nanowire core-shell structure with a different brightness contrast. The ED pattern of the shell region marked as the square A in the fig. 9b presented in the inset A of this figure shows that the shell part of the nanowire consists of a single-crystal‐ line spinel ZnGa2O4 with a [111] direction parallel to the long axial direction of the nano‐ wire. The ED pattern of the core region of B (square B, the ED pattern is shown in the inset B) confirms that the core part of the nanowire is a single crystal wurtzite ZnO with a [0002] direction parallel to the long axial direction of the nanowire. Furthermore, the ED pattern taken on the interfacial region of C (square C, the ED pattern is shown in the inset C) clearly shows two sets of single-crystalline electron diffraction dot patterns, indexed to the spinel ZnGa2O4 and the wurtzite ZnO. From the electron diffraction pattern, it is found that the ZnGa2O4 [111] direction is parallel to the ZnO [0002] direction and the ZnGa2O4 [2-20] direction is parallel to the ZnO [-2110] direction. An epitaxial relationship between the ZnGa2O4 (11-2) plane and the ZnO (1-100) plane is therefore predicted. Thus the subsequent growth of ZnGa2O4 on the ZnO NFS results in a hetero-epitaxial growth of the ZnGa2O4 shells on the ZnO cores. The final

that the NFS is made of ZnO and fabricated on the *a*-plane sapphire substrate.

quent growth of the ZnO-ZnGa2O4 core-shell NFS.

product is well-aligned ZnO-ZnGa2O4 core-shell nanowires.

nanowires. [32 and 34]

140 Nanowires - Recent Advances

**Figure 9.** TEM, EDX-mapping and ED analyses of the dense and vertically aligned (a) ZnO and (b) ZnO-ZnGa2O4 nano‐ wire arrays. [33 and 34]

Finally, standard X-ray diffraction (XRD) 2θ-ω scans were performed to examine the crystal structure and crystal orientations of the ZnO NFS and ZnO-ZnGa2O4 core-shell NFS over large areas [32 and 34]. Only two main diffraction peaks indexed to the wurtzite ZnO (0002) and (0004) planes were observed on the ZnO NFS sample, indicating a unique out-of-plane crystallographic orientation of the synthesized ZnO NFS over large areas. Furthermore, Zhong et al. performed XRD pole figure and rocking curve measurements. A single crystal domain texture quality (a unique in-plane and out-of-plane orientation over large areas) of the synthesized ZnO NFS was concluded. Such a high quality ZnO NFS was synthesized thanks to a new epitaxial process involving two growth steps. An epitaxial synthesis of the (0001) ZnO on the (11-20) sapphire substrate likely occurred at the beginning of the growth, helping to lock the in-plane orientation of the (0001) ZnO crystal. The Zn atoms then dif‐ fused into the Al2O3 substrate to form a single crystal, resulting into an intermediate layer of [0-21] oriented ZnAl2O4. This layer provides an improved in-plane geometry and lattice con‐ stant matching over the *a* plane of sapphire and, therefore, helps sustaining the growth of the high quality (0001) ZnO [34]. The fabricated high quality ZnO NFS can be seen as an ide‐ al template to support the synthesis of a single crystal quality ZnO-ZnGa2O4 core-shell NFS. The XRD 2θ-ω scan result of the core-shell NFS shows only four diffraction peaks. Two dif‐ fraction peaks are indexed to the wurtzite ZnO (0002) and (0004) of the cores, and the other two diffraction peaks are indexed to the spinel ZnGa2O4 (111) and (222) of the shells. This result is strong evidence that the ZnO cores and ZnGa2O4 shells were both of a single crystal quality and that the crystallographic axes of [0001] ZnO and [111] ZnGa2O4 are well aligned over large areas.

#### **4.4. Electronic properties of ZnO-ZnGa2O4 core-shell NFSs**

It is of first importance for the development of PEC water splitting cells to understand the electronic properties of the ZnO-ZnGa2O4 core-shell NFS photoanode in the electrolyte solu‐ tion. The carrier density and the flatband potential of the ZnO-ZnGa2O4 core-shell NFS were measured by electrochemical impedance spectroscopy [32]. For this purpose, a solution of 0.5 M NaClO4 buffered (pH of 7.0) was used as an electrolyte, Ag/AgCl electrode in saturat‐ ed KCl solution as the reference electrode, a Pt wire as the counter electrode and the ZnO-ZnGa2O4 core-shell NFS sample as a working electrode. Prior to the electrochemical impedance measurement, N2 gas was bubbled for 10 minutes to get rid of O2 in the electro‐ lyte solution. The electrochemical impedance analysis was then performed for a bias rang‐ ing from - 1 to 1 V (versus Ag/AgCl), at the frequency of 1 kHz under dark condition. The results of this experiment are presented in a *1/C <sup>2</sup>* vs. *V* plot (Fig. 10a).

The carrier density and the flatband potential of the cylindrical nanowires were quantified using the model proposed by [17]. Solving the Poisson equation with a cylindrical capaci‐ tance approximation at the nanowire/electrolyte interfaces, we obtain:

$$\frac{1}{r}\frac{\partial}{\partial r}\left(r\frac{\partial V}{\partial r}\right) = -\frac{q}{s\varepsilon\_{\circ}}N\_{\omega} \tag{1}$$

$$C\_{\boldsymbol{x}} = \frac{2\boldsymbol{\varepsilon}\boldsymbol{\varepsilon}\_{\circ}\boldsymbol{\boldsymbol{x}}^{\circ}}{\mathcal{R}(\boldsymbol{\mathcal{R}}^{\circ} - \boldsymbol{\boldsymbol{x}}^{\circ})} \tag{2}$$

where *R* is the radius of the nanowires, *x* the central radius of the quasi-neutral region in the nanowires, *C <sup>S</sup>* the capacitance per unit area of the nanowires, *V* the applied bias between the electrodes, *V FB* the flatband potential of the nanowire at the nanowire/electrolyte inter‐ face, *q* the unit electron charge, *ε <sup>0</sup>* the permittivity of vacuum and *ε* the relative dielectric constant of ZnO. The results of the fitting of the measured *1/C <sup>2</sup>* vs. *V* curve according to the Mora-Seró model are shown in Fig. 10 b. The carrier density of the ZnO-ZnGa2O4 nanowires was estimated to be ~1019 cm-3. According to the literature [1, 28, 34], the carrier density of the undoped ZnO nanowires is usually around ~1017 – 1018 cm-3. The increase in the carrier density of the ZnO-ZnGa2O4 NFS sample compared to that of the ZnO nanowires can be un‐ derstood as the formation of conductive ZnGa2O4 shell layers and the introduction of Ga do‐ nors inside the ZnO cores [19, 31]. The increased carrier density in the ZnO-ZnGa2O4 nanowires decreases their electrical resistance and thus contributes to a reduced energy loss for PEC water splitting applications.

The flatband potential of the ZnO-ZnGa2O4 NFS sample at the nanowire/electrolyte inter‐ face is another important parameter when analyzing the photoanodes properties for water splitting in a PEC cell. Typically, a negative flatband potential and an upward band bending occur at the interface between an *n*-type semiconductor and an electrolyte, due to their Fermilevel difference [22]. A schematic illustration of the band bending with a contact energy barrier at an idealized interface between a n-type semiconductor in equilibrium with an electrolyte is shown in Fig. 10 c. The flat-band potential of the ZnO-ZnGa2O4 core-shell NFS sample was estimated to be about - 0.4 V (versus NHE). The cathodic flatband potential of the ZnO-ZnGa2O4 NFS sample provides a suitable energy band position for PEC water splitting.

**Figure 10.** a) Electrochemical impedance plot of the ZnO-ZnGa2O4 NFS sample at the frequency of 1 kHz in dark condi‐ tion with a bias varying from -0.8 to 1.2 V (versus NHE); (b) Fitted curves of ZnO-ZnGa2O4 NFS sample with an assumed ε value of 10 for the ZnO-ZnGa2O4 nanowires; (c) A schematic of the surface band bending at an idealized interface between an n-type semiconductor and an electrolyte in equilibrium. qVbi is the energy barrier height at the interface. [32] - Reproduced by permission of The Royal Society of Chemistry.

### **4.5. PEC performances of ZnO-ZnGa2O4 core-shell NFSs**

quality and that the crystallographic axes of [0001] ZnO and [111] ZnGa2O4 are well aligned

It is of first importance for the development of PEC water splitting cells to understand the electronic properties of the ZnO-ZnGa2O4 core-shell NFS photoanode in the electrolyte solu‐ tion. The carrier density and the flatband potential of the ZnO-ZnGa2O4 core-shell NFS were measured by electrochemical impedance spectroscopy [32]. For this purpose, a solution of 0.5 M NaClO4 buffered (pH of 7.0) was used as an electrolyte, Ag/AgCl electrode in saturat‐ ed KCl solution as the reference electrode, a Pt wire as the counter electrode and the ZnO-ZnGa2O4 core-shell NFS sample as a working electrode. Prior to the electrochemical impedance measurement, N2 gas was bubbled for 10 minutes to get rid of O2 in the electro‐ lyte solution. The electrochemical impedance analysis was then performed for a bias rang‐ ing from - 1 to 1 V (versus Ag/AgCl), at the frequency of 1 kHz under dark condition. The

The carrier density and the flatband potential of the cylindrical nanowires were quantified using the model proposed by [17]. Solving the Poisson equation with a cylindrical capaci‐

*V q r N*

¶ ¶ = - ¶ ¶ æ ö ç ÷

0

ee

2 0 2 2 2 ( ) *<sup>S</sup> <sup>x</sup> <sup>C</sup> RR x* ee

where *R* is the radius of the nanowires, *x* the central radius of the quasi-neutral region in the nanowires, *C <sup>S</sup>* the capacitance per unit area of the nanowires, *V* the applied bias between the electrodes, *V FB* the flatband potential of the nanowire at the nanowire/electrolyte inter‐ face, *q* the unit electron charge, *ε <sup>0</sup>* the permittivity of vacuum and *ε* the relative dielectric

Mora-Seró model are shown in Fig. 10 b. The carrier density of the ZnO-ZnGa2O4 nanowires was estimated to be ~1019 cm-3. According to the literature [1, 28, 34], the carrier density of the undoped ZnO nanowires is usually around ~1017 – 1018 cm-3. The increase in the carrier density of the ZnO-ZnGa2O4 NFS sample compared to that of the ZnO nanowires can be un‐ derstood as the formation of conductive ZnGa2O4 shell layers and the introduction of Ga do‐ nors inside the ZnO cores [19, 31]. The increased carrier density in the ZnO-ZnGa2O4 nanowires decreases their electrical resistance and thus contributes to a reduced energy loss

The flatband potential of the ZnO-ZnGa2O4 NFS sample at the nanowire/electrolyte inter‐ face is another important parameter when analyzing the photoanodes properties for water

*d*

tance approximation at the nanowire/electrolyte interfaces, we obtain:

1

constant of ZnO. The results of the fitting of the measured *1/C <sup>2</sup>*

for PEC water splitting applications.

*rr r*

=

vs. *V* plot (Fig. 10a).

è ø (1)


vs. *V* curve according to the

**4.4. Electronic properties of ZnO-ZnGa2O4 core-shell NFSs**

results of this experiment are presented in a *1/C <sup>2</sup>*

over large areas.

142 Nanowires - Recent Advances

The PEC performance of the ZnO-ZnGa2O4 NFS sample used as a photoanode was studied in a 0.1 M Na2SO4 electrolyte solution at pH = 6 under illumination with a 300 W Xenon lamp [32]. An Ag/AgCl electrode in a saturated KCl solution was used as the reference electrode and a Pt wire used as the counter electrode. Fig. 11a shows the current-voltage curve for the ZnO-ZnGa2O4 nanowire array at a bias ranging from -0.8 V to 1 V (versus Ag/AgCl) with light on-off cycles, indicating a good photoresponsibility of the NFS sample used as photoanode. The flatband potential of the NFS sample is equivalent to the onset potential of the anodic photocurrent, as shown to be ~ -0.35 V vs. NHE with an estimated error of ±0.1 V in the inset of Fig. 11a [32]. This value of the onset potential agreed well with the flatband potential of ~ -0.4 V estimated from the fit of the electrochemical impedance data in Fig. 10 b.

The amperometric study of the ZnO-ZnGa2O4 core-shell nanowire array was performed at a fixed bias of 1.23 VRHE under on-off illumination cycles. The obtained data are plotted in the form of current-time curve (Fig. 11b). The result shows a very low current of < 10-4 mA/cm2 under dark conditions, indicating the absence of any chemical reaction at the ZnO-ZnGa2O4 NFS anode for an applied bias of 1.23 VRHE under dark conditions. Under illumination, a large photocurrent with a steady value of 1.2 mA/cm2 was obtained, indicating a stable and efficient photoelectrolysis of water at the surface of the ZnO-ZnGa2O4 NFS photoanode. As a comparison, the photocurrent measured on ZnO or *n*-doped ZnO nanowire array always deteriorates with time under continuous light illumination, confirming that ZnO is chemi‐ cally unstable in water under illumination (Li et al. 2009). After switching off the light illu‐ minating the ZnO-ZnGa2O4 NFS anode, a fast recovery of the current to the dark current value was clearly observed. The on-off current cycles presented a very good reproducibility and anti-photocorrosion stability as shown in Fig. 11 b. Therefore, the core-shell NFS dem‐ onstrated an improved stability and a large photocurrent of 1.2 mA/cm2 when used as the photoanode in the PEC water splitting cell [32].

**Figure 11.** a) Current-voltage curves of the ZnO-ZnGa2O4 NFS sample in an electrolyte of 0.1 M Na2SO4 at pH = 6 under illumination with a 300 W Xenon lamp. Sweep rate of the voltage was 0.5 mV/s. (b) Current-time curve of the ZnO-ZnGa2O4 NFS when used as a photoanode at an applied bias of 1.23 VRHE with light on-off cycles. [32] - Reproduced by permission of The Royal Society of Chemistry.

#### **5. Conclusions**

Recent advances in the development of densely packed and single crystalline quality semi‐ conductor NFSs with a controllable doping level offer promising opportunities for the fu‐ ture PEC water splitting applications. In this type of nanostructured photoanodes, an improved light absorption, enhanced charge separation and an increased surface area for chemical reaction are obtained due to the shell layer of the semiconductor nanowire array of the NFS. Further, the film underneath the nanowires realizes an efficient electrical connec‐ tion of the semiconductor nanowire array, thus facilitating the direct use of the NFS as the photoanode in PEC water splitting applications. However, many difficulties still remain to be resolved. Two essential issues related to the reliability and the efficiency needs to be ad‐ dressed for all the PEC water splitting systems. In this Chapter, the NFS-based photoanodes made of different semiconductor photocatalysts have been reviewed and their advantages and the remaining challenges have been discussed in detail. The TiO2 NFS photoanode has an excellent chemical stability in water, but its large bandgap and inefficient doping limit the efficiency in the PEC applications under solar illumination. The hematite NFS photoano‐ des have a bandgap which permits a good photoresponsivity under the illumination with the visible part of the solar spectrum and a good chemical stability in water. However, the large resistance and inadequate energy band position of the hematite render this NFS ineffi‐ cient for the PEC applications without any further efficient doping and the co-catalysts dec‐ oration. An efficient doping and the co-catalyst decoration of the hematite NFS have not yet been achieved to improve its performances and obtain a stable operation in water splitting PEC. The ZnO-ZnGa2O4 core-shell NFS photoanode offers an enhanced conductivity over TiO2 and hematite, as well as anti-photocorrosive property in water. Thus this material of‐ fers stable and efficient operation of water splitting PEC. However, the ZnO-ZnGa2O4 coreshell NFS only responds to the UV light, which results in a low efficiency under solar light illumination. Further work is now in progress to achieve an efficient and stable solar water splitting with doped ZnO-ZnGa2O4 core-shell NFSs. Besides the photocatalyst materials al‐ ready fabricated in the form of NFSs reviewed in this Chapter, it is worth noting that a large variety of alternative photocatalysts exist, which can permit to grow nanostructures for high efficiency PEC applications. Finally, still more progress is expected to be made to realize an efficient and stable solar water splitting system permitting to achieve a new, sustainable and environmentally friendly hydrogen production method, contributing to a better daily life of human beings in the near future.

## **Author details**

cally unstable in water under illumination (Li et al. 2009). After switching off the light illu‐ minating the ZnO-ZnGa2O4 NFS anode, a fast recovery of the current to the dark current value was clearly observed. The on-off current cycles presented a very good reproducibility and anti-photocorrosion stability as shown in Fig. 11 b. Therefore, the core-shell NFS dem‐ onstrated an improved stability and a large photocurrent of 1.2 mA/cm2 when used as the

**Figure 11.** a) Current-voltage curves of the ZnO-ZnGa2O4 NFS sample in an electrolyte of 0.1 M Na2SO4 at pH = 6 under illumination with a 300 W Xenon lamp. Sweep rate of the voltage was 0.5 mV/s. (b) Current-time curve of the ZnO-ZnGa2O4 NFS when used as a photoanode at an applied bias of 1.23 VRHE with light on-off cycles. [32] - Reproduced by

Recent advances in the development of densely packed and single crystalline quality semi‐ conductor NFSs with a controllable doping level offer promising opportunities for the fu‐ ture PEC water splitting applications. In this type of nanostructured photoanodes, an improved light absorption, enhanced charge separation and an increased surface area for chemical reaction are obtained due to the shell layer of the semiconductor nanowire array of the NFS. Further, the film underneath the nanowires realizes an efficient electrical connec‐ tion of the semiconductor nanowire array, thus facilitating the direct use of the NFS as the photoanode in PEC water splitting applications. However, many difficulties still remain to be resolved. Two essential issues related to the reliability and the efficiency needs to be ad‐ dressed for all the PEC water splitting systems. In this Chapter, the NFS-based photoanodes made of different semiconductor photocatalysts have been reviewed and their advantages and the remaining challenges have been discussed in detail. The TiO2 NFS photoanode has an excellent chemical stability in water, but its large bandgap and inefficient doping limit the efficiency in the PEC applications under solar illumination. The hematite NFS photoano‐ des have a bandgap which permits a good photoresponsivity under the illumination with the visible part of the solar spectrum and a good chemical stability in water. However, the large resistance and inadequate energy band position of the hematite render this NFS ineffi‐ cient for the PEC applications without any further efficient doping and the co-catalysts dec‐

photoanode in the PEC water splitting cell [32].

permission of The Royal Society of Chemistry.

**5. Conclusions**

144 Nanowires - Recent Advances

Miao Zhong1 , Mario Kurniawan1 , Aleksandra Apostoluk2\*, Bertrand Vilquin2 and Jean-Jacques Delaunay1\*

\*Address all correspondence to: jean@mech.t.u-tokyo.ac.jp

1 School of Engineering, The University of Tokyo, Japan

2 INL, CNRS UMR-5270, Lyon University, France

## **References**


[19] Omata, T., Ueda, N., & Ueda, K. (1994). New ultraviolet‐transport electroconductive oxide, ZnGa2O4 spinel. *Appl. Phys. Lett.*, 64, 1077-1078.

[5] Fan, H. J., Lee, W., Hauschild, R., Alexe, M., Rhun, G. L., Scholz, R., Dadgar, A., Nielsch, K., Kalt, H., Krost, A., Zacharias, M., & Gösele, U. (2006). Template-assisted large-scale ordered arrays of ZnO pillars for optical and piezoelectric applications.

[6] Fan, H. J., Yang, Y., & Zacharias, M. (2009). ZnO-based ternary compound nanotubes

[7] Frank, S. N., & Bard, A. J. (1977). Heterogeneous photocatalytic oxidation of cyanide ion in aqueous solutions at titanium dioxide powder. *J. Am. Chem. Soc.*, 99, 303-304.

[8] Feng, X., Shankar, K., Varghese, O. K., Paulose, M., Latempa, T. J., & Grimes, C. A. (2008). Vertically aligned single crystal TiO2 nanowire arrays grown directly on transparent conducting oxide coated glass: synthesis details and applications. *Nano.*

[9] Fujishima, A., & Honda, K. (1972). Electrochemical photolysis of water at a semicon‐

[10] Hoang, S., Guo, S., Hahn, N. T., Bard, A. J., & Mullins, C. B. (2011). Visible light driv‐ en photoelectrochemical water oxidation on nitrogen-modified TiO2 nanowire. *Nano.*

[11] Ikarashi, K., Sato, J., Kobayashi, H., Saito, N., Nishiyama, H., & Inoue, Y. (2002). Pho‐ tocatalysis of water decomposition by RuO2-Dispersed ZnGa2O4 with d10 configura‐

[12] Iordanova, N., Dupuis, M., & Rosso, K. M. (2005). Charge transport in metal oxides:

[13] Kay, A., Cesar, I., & Grätzel, M. (2006). New benchmark for water photooxidation by

[14] Krol, R. V. D., Liang, Y., & Schoonman, J. (2008). Solar hydrogen production with

[15] Law, M., Greene, L. E., Johnson, J. C., Saykally, R., & Yang, P. (2005). Nanowire dye-

[16] Ling, Y., Wang, G., Wheeler, A. D., Zhang, J. Z., & Li, Y. (2011). Sn-doped hematite nanostructures for photoelectrochemical water splitting. *Nano. Lett.*, 11, 119-125.

[17] Mora-Seró, I., Fabregat-Santiago, F., Denier, B., & Bisquert, J. (2006). Determination of carrier density of ZnO nanowires by electrochemical techniques. *Appl. Phys. Lett.*, 89,

[18] Nakagawa, T., Beasley, C. A., & Murray, R. W. (2009). Efficient electro-oxidation of water near its reversible potential by a mesoporous IrO x nanoparticle film. *J. Phys.*

A theoretical study of hematite α-FeO. *J. Chem. Phys.*, 122, 144305-1-10.

nanostructured α-Fe2O3 films. *J. Am. Chem. Soc.*, 128, 15714-15721.

nanostructured metal oxide. *J. Mater. Chem.*, 18, 2311-2320.

*Small*, 2, 561-568.

146 Nanowires - Recent Advances

*Lett.*, 8, 3781-3786.

*Lett.*, 12, 26-32.

203117-1-3.

*Chem. C*, 113, 12958-12961.

and nanowires. *J. Mater. Chem.*, 19, 885-900.

ductor electrode. *Nature*, 283, 37-38.

tion. *J. Phys. Chem. B*, 106, 9048-9053.

sensitized solar cells. *Nature*, 4, 455-459.


**Section 2**

**Silicon Nanowires**

[33] Zhong, M., Li, Y., Tokizono, T., Zheng, M., Yamada, I., Delaunay, J., & , J. (2012b). Vertically aligned ZnO-ZnGa2O4 core-shell nanowires: from synthesis to optical

[34] Zhong, M., Sato, Y., Kurniawan, M., Apostoluk, A., Masenelli, B., Maeda, E., Ikuhara, Y., & Delaunay, J. J. (2012c). ZnO dense nanowire array on a film structure in a single crystal domain texture for optical and photoelectrochemical applications. *Nanotech‐*

properties. *J. Nanopart. Res.*, 14, 804-814.

*nology*, accepted.

148 Nanowires - Recent Advances

## **Chapter 8**

## **Silicon Nanowire FinFETs**

C. Mukherjee and C. K. Maiti

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52591

## **1. Introduction**

Some of the fundamental problems of ultra-small MOSFETs beyond sub-10nm channel length are the electrostatic limits, source-to-drain tunnelling, carrier mobility degradation, process variations, and static leakage. The trend toward ultra-short gate length MOSFETs re‐ quires a more and more effective control of the channel by the gate leading to new device architecture. It appears that non-classical device architectures can extend the CMOS lifetime and provide solutions to continue scaling. In case of silicon-based CMOS technologies, pla‐ nar MOSFETs are limited to scaling beyond 15 nm technology node. As simple scaling of silicon CMOS becomes increasingly complex and expensive, there is considerable interest in increasing performance by using strained channels which can improve carrier mobility and drive current in a device. Multi-gate MOSFETs based on the concept of volume inversion are widely recognized as one of the most promising solutions for meeting the ITRS roadmap requirements. A wide variety of multi-gate architectures, including Double-Gate (DG), Gate-All-Around (GAA), Pi-FET and Fin Field-Effect Transistors (FinFETs), rectangular or cylin‐ drical nanowire MOSFETs has been proposed in the literature. In all cases, these structures exhibit a superior control of short channel effects resulting from an exceptional electrostatic coupling between the conduction channel and the surrounding gate electrode. The nanowire (NW) transistors can be seen as the ultimate integration of the innovative nanodevices and is one of the candidates which have gained significant attention from both the device and cir‐ cuit developers because of its potential for building highly dense and high performance electronic circuits. Recent advances in nanoscale fabrication techniques have shown that semiconductor nanowires may become the candidate for next generation technologies. Si and Ge nanowire transistors are also important because of their compatibility with the CMOS technology.

Nanowires and FinFETs have attracted considerable attention due to their proven robust‐ ness against Short-Channel Effects (SCE) and relatively simple fabrication. Silicon nanowire

© 2012 Mukherjee and Maiti; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Mukherjee and Maiti; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

FinFET (SNWFT) is being considered as the candidate for CMOS scaling beyond the 32 nm node due to its high performance, excellent gate control and enhanced carrier transportation properties. Thin and multi-gate controlled body provide FinFETs a superior short-channel effect control, electrostatic shielding from the body bias, relaxing channel doping or pocket implants, commonly needed in planar technologies to avoid threshold voltage (VTH) roll-off [1]. Adequate device VTH is now being achieved by using gate stack engineering. Metal-Gates with High-k (HKMG) dielectric stacks provide the desired VTH by work-function tun‐ ing as well as preserving low gate leakage [2]. Various process options are being attempted to affect carrier transport in a different manner. The use of metal gates as replacement for Poly-Si stacks eliminates the Poly-depletion effect, benefiting effective carrier mobility by re‐ ducing the transverse field [1]. Additionally, use of undoped channels improves low-field *μ eff* due to the reduction in the substrate impurity scattering [3]. Mobility degradation due to Coulomb scattering in short-channel devices should further be reduced in the absence of pocket implants [4]. However, high-k dielectrics are known to degrade mobility as a result of a combination of Coulomb and phonon scattering mechanisms [5]. In order to account for inversion layer mobility degradation mechanisms for FinFET devices, a robust *μ eff* extrac‐ tion algorithm is necessary. The mobility extraction requires accurate measurement of both gate to channel capacitance and channel current, together with reliable estimations for the parasitic Source-Drain series resistance (RSD) and the effective channel length [4]. Unfortu‐ nately, the capacitance of short-channel length devices in the presence of large gate leakage is no longer characterized trivially. In multi-gate architectures like the FinFET, the carrier transport occurs in different crystallographic planes. For standard substrates, the current flow occurs in the (100)/(110) for the top surface and (110)/(110) for the sidewalls. The trans‐ port in these crystal planes and directions are characterized by different mobility primarily due to the anisotropy of the effective masses.

The most promising among various Si multi-gate MOSFET architectures such as double-gate and tri-gate FinFETs, are nanowire FinFETs due to their superior electrostatic control through gate-all-around structure. Superior gate control, immunity of threshold voltage from substrate bias and excellent carrier transport properties along with more aggressive channel length scaling possibility have made GAA architecture with semiconductor nano‐ wire channel a potential candidate for post-planar transistor design. Two approaches are generally used to fabricate Si NWs as well as other semiconductor NWs: bottom-up and topdown. In the first method, NWs are usually grown using a metallic catalyst on a separate substrate, usually through a Vapor-Liquid-Solid (VLS) growth mechanism. After a chemical or mechanical separation step, the NWs are harvested and transferred to another substrate [6]. In the top down-approach, the NWs are fabricated using a CMOS compatible technolo‐ gy, such as lithography-based patterning and etching [7]. Unlike the bottom-up approach where the NWs are randomly distributed, the top-down method enables accurate position‐ ing of the NWs across the wafer and facilitates the ultra-large-scale-integration (ULSI) for high performance nano-electronic circuits. Moreover, due to processing difficulties related to the length of grown NWs, NW release and gate-etch process, most of the VLS grown NW transistors have omega-shaped gate (Ω-gate) geometry and are thus not full gate-all-around [6]. Among all the promising post-CMOS non-conventional structures, the silicon nanowire transistor has the most unique advantage – it is based on silicon.

FinFET (SNWFT) is being considered as the candidate for CMOS scaling beyond the 32 nm node due to its high performance, excellent gate control and enhanced carrier transportation properties. Thin and multi-gate controlled body provide FinFETs a superior short-channel effect control, electrostatic shielding from the body bias, relaxing channel doping or pocket implants, commonly needed in planar technologies to avoid threshold voltage (VTH) roll-off [1]. Adequate device VTH is now being achieved by using gate stack engineering. Metal-Gates with High-k (HKMG) dielectric stacks provide the desired VTH by work-function tun‐ ing as well as preserving low gate leakage [2]. Various process options are being attempted to affect carrier transport in a different manner. The use of metal gates as replacement for Poly-Si stacks eliminates the Poly-depletion effect, benefiting effective carrier mobility by re‐ ducing the transverse field [1]. Additionally, use of undoped channels improves low-field *μ eff* due to the reduction in the substrate impurity scattering [3]. Mobility degradation due to Coulomb scattering in short-channel devices should further be reduced in the absence of pocket implants [4]. However, high-k dielectrics are known to degrade mobility as a result of a combination of Coulomb and phonon scattering mechanisms [5]. In order to account for inversion layer mobility degradation mechanisms for FinFET devices, a robust *μ eff* extrac‐ tion algorithm is necessary. The mobility extraction requires accurate measurement of both gate to channel capacitance and channel current, together with reliable estimations for the parasitic Source-Drain series resistance (RSD) and the effective channel length [4]. Unfortu‐ nately, the capacitance of short-channel length devices in the presence of large gate leakage is no longer characterized trivially. In multi-gate architectures like the FinFET, the carrier transport occurs in different crystallographic planes. For standard substrates, the current flow occurs in the (100)/(110) for the top surface and (110)/(110) for the sidewalls. The trans‐ port in these crystal planes and directions are characterized by different mobility primarily

The most promising among various Si multi-gate MOSFET architectures such as double-gate and tri-gate FinFETs, are nanowire FinFETs due to their superior electrostatic control through gate-all-around structure. Superior gate control, immunity of threshold voltage from substrate bias and excellent carrier transport properties along with more aggressive channel length scaling possibility have made GAA architecture with semiconductor nano‐ wire channel a potential candidate for post-planar transistor design. Two approaches are generally used to fabricate Si NWs as well as other semiconductor NWs: bottom-up and topdown. In the first method, NWs are usually grown using a metallic catalyst on a separate substrate, usually through a Vapor-Liquid-Solid (VLS) growth mechanism. After a chemical or mechanical separation step, the NWs are harvested and transferred to another substrate [6]. In the top down-approach, the NWs are fabricated using a CMOS compatible technolo‐ gy, such as lithography-based patterning and etching [7]. Unlike the bottom-up approach where the NWs are randomly distributed, the top-down method enables accurate position‐ ing of the NWs across the wafer and facilitates the ultra-large-scale-integration (ULSI) for high performance nano-electronic circuits. Moreover, due to processing difficulties related to the length of grown NWs, NW release and gate-etch process, most of the VLS grown NW transistors have omega-shaped gate (Ω-gate) geometry and are thus not full gate-all-around

due to the anisotropy of the effective masses.

152 Nanowires - Recent Advances

In a MOSFET, the carriers encounter various scattering mechanisms on their way towards the drain terminal. Carrier mobility is a well known benchmark to judge the intrinsic per‐ formance of a long channel MOSFET. The state-of-the art short channel devices do not oper‐ ate in the fully ballistic regime (they are at roughly 60% of the ballistic limit) and mobility is related to velocity through effective mass and ballistic ratio. Therefore, understanding the carrier mobility is beneficial to design and engineer new devices for future CMOS genera‐ tions. The presence of significant resistive and capacitive parasitics as well as lack of large capacitance due to the small size of the NW channel adds complexities to the extraction of the intrinsic NW device characteristics. Simplified assumptions or incomplete device charge simulations generally lead to inaccurate and unreasonable mobility extraction [9]. In the ab‐ sence of a well-behaved top-down process (as compared to the bottom-up-fabricated NWs with smooth and near ideal sidewalls) results in significant mobility degradation for smaller NW sizes [8].

Noise is known as a fundamental problem in various fields such as telecommunication, nanoelectronics, and biological systems. The low-frequency noise, or 1/f noise, is the excess noise at low frequencies whose power spectral density (PSD) approximately depends inver‐ sely on the frequency and therefore escalates at low frequencies. The 1/f noise originating from the transistors is a major issue in analog circuit design. The 1/f noise is, for example, up converted to undesired phase noise in voltage controlled oscillator (VCO) circuits, which can limit the information capacity of communication systems.

In this chapter we shall present a detailed framework on gate-all-around nanowires and multi-gate FinFETs which will include process and device design and characterization, sim‐ ulation, and low-frequency noise and tunnelling spectroscopy analyses. Carrier mobility ex‐ traction in nanowire and FinFETs using split-CV technique will be described. Channel orientation dependence of the electrical parameters such as current, low-frequency noise will be presented. Self-heating effects in nanowire transistors are also discussed. In Section 2, the growth and fabrication of silicon nanowire transistors and multi-gate FinFET devices will be covered. The detailed fabrication process steps for GAA nanowires as well as tri-gate FinFETs are discussed. 3D device simulation and electrical characterization will also be cov‐ ered in this Section. Split CV measurement technique is employed for mobility calculation and detailed model equations are formulated for FinFET and nanowire structures. In Section 3, the low-frequency noise characteristics of the nanowire FinFETs will be taken up. Voltage dependence of flicker noise (1/*f*) and power spectral densities and corner frequencies will be discussed based on low-frequency noise measurements. Time-domain random telegraph signal (RTS) and existence of two-level or multi-level RTS noise in nanowire transistors will also be taken up. In section 4, the orientation dependence will be discussed and current-volt‐ age characteristics are compared to show the effect of orientation on the drain current. The effect of channel orientation is also discussed for the nanowire FinFET noise spectral densi‐ ties for (100) and (110) channel orientations. In section 5, the lattice structures of Si, SiO2 in the nanowire FinFET are analyzed via inelastic electron tunneling spectroscopy (IETS). The lattice dynamics and other molecular vibrations in silicon nanowire FinFETs are discussed using the IETS spectra.

## **2. Multi-gate FinFET and Nanowire Transistors**

In order to control short channel effects in planar SOI MOSFETs and maintain reasonable electrostatics at a given gate length, LG, a critical Si thickness, tSi on the order of LG/3 is re‐ quired [10]. Multi-gate devices have been suggested to significantly relax the critical Si di‐ mension in fully depleted SOI (FD-SOI) device architectures and process technology. For this purpose, planar double-gate, vertical double-gate (FinFET), triple-gate (Tri-gate), ome‐ ga-shaped gate (Ω-gate), penta-gate, inverted T-channel FET, Φ-FET, and GAA nanowires have been proposed by various research groups over the past decade.

While most of these multi-gate technologies have been experimentally shown to benefit from excellent electrostatics and short-channel effects, they commonly suffer from a compli‐ cated fabrication process technology compared to planar bulk or SOI technology. Figure 1. shows the device schematic of


The critical Si dimensions to sustain acceptable electrostatics have been reported [11]. It can be seen that the criterion of Si film thickness may be relaxed for the nanowire MOSFET, which is naturally evolved from double-gate FinFET technology.

**Figure 1.** Schematics of device structures of (a) double-gate FinFET (b) tri-gate and (c) GAA NW MOSFETs.

#### **2.1. 3D Quantum Simulation of Multi-Gate FinFETs**

lattice dynamics and other molecular vibrations in silicon nanowire FinFETs are discussed

In order to control short channel effects in planar SOI MOSFETs and maintain reasonable electrostatics at a given gate length, LG, a critical Si thickness, tSi on the order of LG/3 is re‐ quired [10]. Multi-gate devices have been suggested to significantly relax the critical Si di‐ mension in fully depleted SOI (FD-SOI) device architectures and process technology. For this purpose, planar double-gate, vertical double-gate (FinFET), triple-gate (Tri-gate), ome‐ ga-shaped gate (Ω-gate), penta-gate, inverted T-channel FET, Φ-FET, and GAA nanowires

While most of these multi-gate technologies have been experimentally shown to benefit from excellent electrostatics and short-channel effects, they commonly suffer from a compli‐ cated fabrication process technology compared to planar bulk or SOI technology. Figure 1.

The critical Si dimensions to sustain acceptable electrostatics have been reported [11]. It can be seen that the criterion of Si film thickness may be relaxed for the nanowire MOSFET,

**Figure 1.** Schematics of device structures of (a) double-gate FinFET (b) tri-gate and (c) GAA NW MOSFETs.

**2. Multi-gate FinFET and Nanowire Transistors**

have been proposed by various research groups over the past decade.

which is naturally evolved from double-gate FinFET technology.

using the IETS spectra.

154 Nanowires - Recent Advances

shows the device schematic of

**a.** double-gate FinFET

**c.** GAA NW MOSFETs.

**b.** tri-gate and

The simulation of tri-gate FinFET is performed using SILVACO ATLAS3D quantum simula‐ tor framework. The Bohm Quantum Potential (BQP) model is included to model the con‐ finement effects in the simulation. Also, calibration of the BQP model against the 2D Schrödinger-Poisson simulation is done for describing the quantum effects in small geome‐ try Si FinFET and nanowire transistors. To model the effects of quantum confinement, AT‐ LAS allows the solution of Schrödinger's equation along with the fundamental device equations. The solution of Schrödinger's equation gives a quantized description of the den‐ sity of states in the presence of quantum-mechanical confining potential variations. When the quantum confinement is in one dimension (along y-axis), the calculation of the quantum electron density relies upon a solution of a 1D Schrödinger equation solved for eigen-state energies Eiv (x) and wave-functions Ψiv (x, y) at each slice perpendicular to x-axis and for each electron valley (or hole band) *v* as described below:

$$-\frac{h^2}{2}\frac{\partial}{\partial \mathbf{y}} \left(\frac{1}{m\_{\mathbf{y}}^{\prime}(\mathbf{x}, \mathbf{y})} \frac{\partial \boldsymbol{\nu}\_{\mathrm{iv}}}{\partial \mathbf{y}}\right) + E\_{\varepsilon}(\mathbf{x}, \mathbf{y}) \boldsymbol{\nu}\_{\mathrm{iv}} = E\_{\mathrm{iv}} \boldsymbol{\Psi}\_{\mathrm{iv}} \tag{1}$$

Here, *m<sup>v</sup> y* (*x*, *y*)is a spatially dependent effective mass in y-direction for the *v*-th valley and EC (x, y) is a conduction band edge. The equation for holes is obtained by substituting hole effective masses instead of that of electrons and valence band edge -EV(x,y) instead of EC (x,y). ATLAS solves the one-dimensional Schrödinger's equation along a series of slices in the y direction relative to the device. The location of the slices in the y direction is developed in two ways. For rectangular ATLAS-defined meshes, the slices will automatically be taken along the existing mesh lines in the ATLAS mesh. If the mesh is non-rectangular or not an ATLAS defined mesh or both, a rectangular mesh must be specified. The potential derived from the solution of Poisson's equation is substituted back into Schrödinger's equation. This solution process (alternating between Schrödinger's and Poisson's equations) continues until convergence and a self-consistent solution of Schrödinger's and Poisson's equations is reached.

For the BQP model, there are two advantages over the density gradient method. First, it has better convergence properties in many situations. Secondly, one can calibrate it against re‐ sults from the Schrödinger-Poisson equation under conditions of negligible current flow. The model introduces a position dependent Quantum Potential, Q, which is added to the Potential Energy of a given carrier type. This quantum potential is derived using the Bohm interpretation of quantum mechanics and takes the following form [12]

$$Q = -\frac{h^2}{2} \frac{\mathcal{V}\underline{\Delta}(M^{-1}\underline{\Delta}(n''))}{n''} \tag{2}$$

where γ and α are two adjustable parameters, M-1 is the inverse effective mass tensor and n is the electron (or hole) density. The first part of the calibrating the BQP model against the Schrödinger-Poisson (S-P) Model, is to choose a suitable bias for the device. There should be negligible current flow and quantum confinement effects that manifest at the chosen biases. The second part of the calibration is to set the appropriate BQP parameters. The third part of calibration is to choose the quantity to compare with S-P results. The BQP equation is cou‐ pled with Poisson's equation using the charge density terms:

$$n = N\_c \exp\left(-\frac{E\_c + qQ}{kT}\right),\\p = N\_r \exp\left(-\frac{qQ - E\_r}{kT}\right) \tag{3}$$

#### **2.2. Tri-gate FinFETs**

Typical p-type tri-gate FinFET is fabricated on 1000 Ǻ SOI layer. The FinFET has a fin height of 30 nm. The fin width and fin length are varied from 60-90 nm and 60 nm to 100 μm, re‐ spectively. A 50 Ǻ SiO2 gate oxide is grown. Polysilicon gate was deposited of thickness 1500 Ǻ, followed by pocket implantation and a 300 Ǻ SiO2 spacer deposition. A deep source/drain implant is followed. Silicidation is done for contact formation by 300 Ǻ Ni/400 Ǻ TiN depo‐ sitions with RTA at 550-600°C for 1 minute. A 700 Ǻ - 1 μm thick Al-pad is deposited for contact formation. The final device is annealed in forming gas at 420°C for 30 minutes.

#### *3D Device Simulation*

The 3D device structure simulated in ATLAS3D framework [13] of the tri-gate bulk Si-Fin‐ FET with fin height of 30 nm, fin width and fin length of 60 nm and 160 nm, is shown in Fig. 2.The simulation takes into account the quantum effects using the BQP model and the Schrödinger-Poisson solver. The tri-gate structure includes a poly-Si gate and a Si-Fin as a channel on the buried oxide (BOX). A gate oxide thickness of 5 nm is used for device simula‐ tion. The energy band diagram is shown in Fig. 3. depicting the conduction and valence band energies as a function of the distance along the Si-fin channel.

**Figure 2.** Simulated 3D device structure of a silicon tri-gate FinFET.

**Figure 3.** Energy band diagram of a tri-gate Si-FinFET.

#### *Current-Voltage Characteristics*

negligible current flow and quantum confinement effects that manifest at the chosen biases. The second part of the calibration is to set the appropriate BQP parameters. The third part of calibration is to choose the quantity to compare with S-P results. The BQP equation is cou‐

exp , exp *C V*

Typical p-type tri-gate FinFET is fabricated on 1000 Ǻ SOI layer. The FinFET has a fin height of 30 nm. The fin width and fin length are varied from 60-90 nm and 60 nm to 100 μm, re‐ spectively. A 50 Ǻ SiO2 gate oxide is grown. Polysilicon gate was deposited of thickness 1500 Ǻ, followed by pocket implantation and a 300 Ǻ SiO2 spacer deposition. A deep source/drain implant is followed. Silicidation is done for contact formation by 300 Ǻ Ni/400 Ǻ TiN depo‐ sitions with RTA at 550-600°C for 1 minute. A 700 Ǻ - 1 μm thick Al-pad is deposited for

contact formation. The final device is annealed in forming gas at 420°C for 30 minutes.

band energies as a function of the distance along the Si-fin channel.

**Figure 2.** Simulated 3D device structure of a silicon tri-gate FinFET.

The 3D device structure simulated in ATLAS3D framework [13] of the tri-gate bulk Si-Fin‐ FET with fin height of 30 nm, fin width and fin length of 60 nm and 160 nm, is shown in Fig. 2.The simulation takes into account the quantum effects using the BQP model and the Schrödinger-Poisson solver. The tri-gate structure includes a poly-Si gate and a Si-Fin as a channel on the buried oxide (BOX). A gate oxide thickness of 5 nm is used for device simula‐ tion. The energy band diagram is shown in Fig. 3. depicting the conduction and valence

èø èø (3)

*C V E qQ qQ E n N p N kT kT* æö æö + - =- =- ç÷ ç÷

pled with Poisson's equation using the charge density terms:

**2.2. Tri-gate FinFETs**

156 Nanowires - Recent Advances

*3D Device Simulation*

Figure 4. shows the measured ID-VGS characteristics of the p-type tri-gate FinFET. Typical de‐ vice dimensions used for measurements are fin length of 160 nm, and width of 60 nm and oxide thickness of ~5 nm. The device displays excellent performance in terms of near ideal subthreshold slope (SS) (~82 mV/dec), and high Ion/Ioff ratios (~106 ). Figure 5. compares the measured ID-VDS with the TCAD simulation results which shows an excellent agreement showing the need of quantum 3D device simulation.

**Figure 4.** Typical measured ID-VGS characteristics of a tri-gate p-type Si-FinFET for two drain voltages; the inset shows the transconductance (*g <sup>m</sup>*) plot as a function of gate voltage.

**Figure 5.** Comparison of ID-VDS characteristics of a tri-gate p-type Si-FinFET for results of both measured data and TCAD simulation.

#### **2.3. Gate-All-Around Nanowire Transistors**

GAA MOSFET is one of the most promising multi-gate structures to extend the scaling of the CMOS devices as it provides the best channel electrostatic control, which improves fur‐ ther with the shrinking of channel thickness. GAA MOSFETs have potential to offer en‐ hanced carrier transport properties compared to planar devices, because of the high carrier mobility on sidewall planes. Careful design and fabrication are critical for GAA MOSFETs for obtaining better transport properties in such structures. The basic nanowire transistor fabrication process flow is illustrated in Fig. 6.

**Figure 6.** Typical nanowire fabrication process flow (After P. Hashemi, Gate-All-Around Silicon Nanowire MOSFETs: Top-down Fabrication and Transport Enhancement Techniques, PhD dissertation, Massachusetts Institute of Technol‐ ogy, 2010 [14]).

After definition of e-beam lithography (EBL) alignment marks, Si nanowires along <110> di‐ rection and S/D pads are defined by hybrid lithography. After reactive-ion etching (RIE) of Si and stripping the resist, nanowires are locally released using a protective photo resist mask and a buffered-oxide etch bath. The suspended nanowires are then RCA cleaned and are subjected to hydrogen anneal process. Although thermally grown oxide benefits from excellent interface with Si nanowires, the facet dependence of the growth usually results in a non-uniform oxide thickness for various crystallographic planes. In addition, thermally grown oxide can induce a significant amount of stress in the nanowires which can alter their intrinsic carrier transport properties [15]. The ALD dielectric benefits from scalability to thinner effective-oxide-thicknesses (EOT) and its excellent conformality ensures that the en‐ tire perimeter of the Si nanowire is uniformly gated. The overlapped gate structures are de‐ signed for two reasons. First, it helps to minimize the S/D external resistance as the ionimplanted nanowires may exhibit very large series resistance. Second, it is technologically difficult to remove the metal gate stringers under the nanowires if the gate is under lapped. Moreover, any gate-first process to be used after vertical etching of the gate requires an iso‐ tropic etch of the gate metal stringers under the nanowires which also laterally etches the bottom gate and results in high extension resistance. After contact via opening and Ti/Al deposition and patterning, the fabrication is completed by a forming gas annealing at 450°C for 30 minutes.

The 3D device simulation is performed in ATLAS3D framework of SILVACO. The simulat‐ ed device structure is shown in Fig. 7. depicting a cylindrical GAA FinFET structure with fin diameter ~10 nm and fin length of 100 nm wih a surrounding gate oxide thickness of 10 nm. Gate electrode is defined with 70 nm overlap on source/drain. (100) SOI wafer (p-type, 1015 cm-3) was the starting material with 200 nm Si on 150 nm buried oxide (BOX). The gate elec‐ trode deposition included 130 nm amorphous silicon (*α*-Si).

**Figure 7.** Simulated GAA nanowire FinFET structure.

**Figure 5.** Comparison of ID-VDS characteristics of a tri-gate p-type Si-FinFET for results of both measured data and

GAA MOSFET is one of the most promising multi-gate structures to extend the scaling of the CMOS devices as it provides the best channel electrostatic control, which improves fur‐ ther with the shrinking of channel thickness. GAA MOSFETs have potential to offer en‐ hanced carrier transport properties compared to planar devices, because of the high carrier mobility on sidewall planes. Careful design and fabrication are critical for GAA MOSFETs for obtaining better transport properties in such structures. The basic nanowire transistor

**Figure 6.** Typical nanowire fabrication process flow (After P. Hashemi, Gate-All-Around Silicon Nanowire MOSFETs: Top-down Fabrication and Transport Enhancement Techniques, PhD dissertation, Massachusetts Institute of Technol‐

After definition of e-beam lithography (EBL) alignment marks, Si nanowires along <110> di‐ rection and S/D pads are defined by hybrid lithography. After reactive-ion etching (RIE) of

TCAD simulation.

158 Nanowires - Recent Advances

ogy, 2010 [14]).

**2.3. Gate-All-Around Nanowire Transistors**

fabrication process flow is illustrated in Fig. 6.

**Figure 8.** a): ID-VDS characteristics of a GAA Si-nanowire; (b): ID-VGS characteristics of a GAA Si-nanowire at two differ‐ ent drain voltages.

Figures 8 (a) and (b)show the typical Id-Vds and Id-Vgs characteristics of a GAA n-FinFET, re‐ spectively. Typical device dimensions used for measurements are fin length of 100 nm, and diameter of 10 nm and oxide thickness of ~10 nm. The excellent subthreshold performance include a high subthreshold slope (SS) of ~90 mV/dec, and Ion/Ioff ratios ~106 .

#### **2.4. Carrier Mobility in FinFETs**

Effective mobility extraction for the FinFET devices was based on an improved split C-V method, similar to planar device mobility extraction [4]. This technique relies on the accu‐ rate computation of the inversion charge (*Q <sup>i</sup>* ), as well as on the correct calculation of the channel conductance (*g <sup>c</sup>*) along the different operation regimes. Effective channel mobility is defined as [16]:

$$
\mu\_{\rm eff} = \frac{L\_{\rm eff}}{W\_{\rm eff}} \cdot \frac{\mathbf{g}\_c}{\mathbf{Q}\_c} \tag{4}
$$

*Weff* =(*W fin* + 2 \* *H fin*)\* *N fins*for a tri-gate structure and *Weff* =2 \*(*W fin* + *H fin*)\* *N fins*for a GAA structure; *Nfins* is the total number of fins, *H fin* is the fin height. The inversion charge is obtained by integration of the accurate gate to channel capacitance data (Cgc).

$$
\underline{Q}\_i(V\_g) = \int\_{V\_G \circ V\_r}^{V\_G} C\_{g\varepsilon}(V)dV\tag{5}
$$

The transfer characteristics and the gate-to-channel capacitance were measured using the split C–V technique to calculate the effective mobility of devices with fin width (Wfin) being 60 nm and the height of the fins (Hfin) being 30 nm. The effective mobility can be expressed in terms of top, bottom and sidewall mobilities; μtop, μbottom, μsidewall, in tri-gate and GAA structures, respectively, as:

#### Silicon Nanowire FinFETs http://dx.doi.org/10.5772/52591 161

$$\begin{aligned} \mu\_{\text{eff}} W\_{\text{eff}} &= \mu\_{\text{top}} W\_{\text{fin}} + 2 \, ^\ast \mu\_{\text{silcalall}} H\_{\text{fin}} \\ \mu\_{\text{eff}} W\_{\text{eff}} &= \mu\_{\text{top}} W\_{\text{fin}} + \mu\_{\text{bonon}} W\_{\text{fin}} + 2 \, ^\ast \mu\_{\text{silcalall}} H\_{\text{fin}} \end{aligned} \tag{6}$$

**Figure 9.** Extracted effective mobility in a p-type tri-gate Si-FinFET with gate length 160 nm, and gate width 60 nm; inset shows the inversion charge as a function of gate voltage.

A linear regression can be used to extract the sidewall and top-surface mobility. Eqs. (4) and (5) are used for calculating the effective mobility in the FinFET device from split C-V meas‐ urements. The inset of Fig. 9 shows inversion charge density as a function of gate voltage, as obtained from Eq. (5) using split C-V gate-to-channel capacitance. Figure 9 depicts the effec‐ tive mobility of the FinFET as calculated using Eq. (4) as a function of inversion charge. The extracted mobility values agree well with the results of [16] for a p-FinFET of with gate length 160 nm.

#### **2.5. Self Heating Effects in Nanowire Transistors**

**Figure 8.** a): ID-VDS characteristics of a GAA Si-nanowire; (b): ID-VGS characteristics of a GAA Si-nanowire at two differ‐

Figures 8 (a) and (b)show the typical Id-Vds and Id-Vgs characteristics of a GAA n-FinFET, re‐ spectively. Typical device dimensions used for measurements are fin length of 100 nm, and diameter of 10 nm and oxide thickness of ~10 nm. The excellent subthreshold performance

Effective mobility extraction for the FinFET devices was based on an improved split C-V method, similar to planar device mobility extraction [4]. This technique relies on the accu‐

channel conductance (*g <sup>c</sup>*) along the different operation regimes. Effective channel mobility is

. *eff <sup>c</sup>*

*eff i L g W Q*

*Weff* =(*W fin* + 2 \* *H fin*)\* *N fins*for a tri-gate structure and *Weff* =2 \*(*W fin* + *H fin*)\* *N fins*for a GAA structure; *Nfins* is the total number of fins, *H fin* is the fin height. The inversion charge is

The transfer characteristics and the gate-to-channel capacitance were measured using the split C–V technique to calculate the effective mobility of devices with fin width (Wfin) being 60 nm and the height of the fins (Hfin) being 30 nm. The effective mobility can be expressed in terms of top, bottom and sidewall mobilities; μtop, μbottom, μsidewall, in tri-gate and GAA

.

), as well as on the correct calculation of the

= (4)

*i g gc V V Q V C V dV* <sup>&</sup>lt; <sup>=</sup> ò (5)

include a high subthreshold slope (SS) of ~90 mV/dec, and Ion/Ioff ratios ~106

*eff*

obtained by integration of the accurate gate to channel capacitance data (Cgc).

( ) () *<sup>G</sup> G T V*

m

ent drain voltages.

160 Nanowires - Recent Advances

defined as [16]:

structures, respectively, as:

**2.4. Carrier Mobility in FinFETs**

rate computation of the inversion charge (*Q <sup>i</sup>*

One of the significant challenges in future semiconductor device design is excessive power dissipation and rise in device temperature. With the introduction of new geometrically con‐ fined device structures like SOI, FinFETs, nanowires and incorporation of new materials with poor thermal conductivities in the device active region, the device thermal problem is expected to become more challenging in coming years. There is considerable degradation in the ON current due to self-heating effects in silicon nanowire transistors. However, insignif‐ icant current degradation is observed in nanowire transistors because of pronounced veloci‐ ty overshoot effect. Placements of the source and drain contacts also play a significant role on the magnitude of self-heating effect in nanowire transistors. Self-heating and random charge effects simultaneously influence the magnitude of the ON current for both positively and negatively charged single charges. The self-heating affects the ON current in two ways:


There is larger current degradation because of self-heating due to decreased thermal con‐ ductivity. Crystallographic direction dependent thermal conductivity is also an important aspect of self heating effects. Larger degradation is observed in the current along the [100] direction when compared to the [110] direction.

In SOI devices, the low thermal conductivity of the underlying silicon dioxide layer, which is about two orders of magnitude less than that of silicon inhibits cooling in SOI devices and causes severe self-heating, resulting in a higher channel operating temperature. The temper‐ ature rise is significant and depends on the buried oxide thickness, silicon thickness, and channel/metal contact separation. The device mobility is reduced as a result of the elevated temperature and results in reduced maximum drain saturation current. Also, high channel temperature leads to increase in interconnect temperature at the silicon-metal contact and make conduction cooling through the source, drain, and interconnects important. It should also be noted that the thermal conductivity of the silicon films decreases as the film thick‐ ness is reduced due to boundary scattering of phonons which further exacerbates self-heat‐ ing and hence device performance. In brief, as semiconductor technology approaches to two-dimensional and three-dimensional transistor structures and are more isolated from the substrate self-heating effects will become increasingly important.

### **3. Low-Frequency Noise in Nanowire Transistors**

Among various options for multi-gate device architecture, such as double-gate, tri-gate, etc., the nanowire (NW) channel with a wrap-around gate, GAA, has the largest advantage in terms of electrostatic integrity. However, several undesired effects become prominent from the miniaturization of the device dimensions. One such unwanted effect is a strong increase in the low-frequency noise generated in the transistor as the size of the device decreases. With dimension-scaling, it is necessary to also scale the power supplies. Noise cannot be completely eliminated and with small signal strength, the accuracy and measurements are limited in electronic circuits in presence of the dominant noise sources. The low-frequency noise, or 1/*f* noise, is the excess noise at low frequencies whose power spectral density ap‐ proximately depends inversely on the frequency and becomes pronounced at low frequen‐ cies. The 1/*f* noise originating from the transistors is a severe obstacle in analog circuit design. For example, it can be up converted to undesired phase noise in voltage controlled oscillator circuits, which can limit the information capacity of communication systems. The problems in down-scaled devices such as the nanowire FinFETs compel one to study noise sources, mechanisms and their physical origins in detail. In this Section, we shall briefly dis‐ cuss the two fundamental low-frequency noise characteristics, random telegraph signal noise and 1/*f* noise.

#### **3.1. Low-Frequency Noise Measurement Setup**

charge effects simultaneously influence the magnitude of the ON current for both positively and negatively charged single charges. The self-heating affects the ON current in two ways: **1.** by lowering the barrier at the source end of the channel, thus allowing more carriers to

There is larger current degradation because of self-heating due to decreased thermal con‐ ductivity. Crystallographic direction dependent thermal conductivity is also an important aspect of self heating effects. Larger degradation is observed in the current along the [100]

In SOI devices, the low thermal conductivity of the underlying silicon dioxide layer, which is about two orders of magnitude less than that of silicon inhibits cooling in SOI devices and causes severe self-heating, resulting in a higher channel operating temperature. The temper‐ ature rise is significant and depends on the buried oxide thickness, silicon thickness, and channel/metal contact separation. The device mobility is reduced as a result of the elevated temperature and results in reduced maximum drain saturation current. Also, high channel temperature leads to increase in interconnect temperature at the silicon-metal contact and make conduction cooling through the source, drain, and interconnects important. It should also be noted that the thermal conductivity of the silicon films decreases as the film thick‐ ness is reduced due to boundary scattering of phonons which further exacerbates self-heat‐ ing and hence device performance. In brief, as semiconductor technology approaches to two-dimensional and three-dimensional transistor structures and are more isolated from the

Among various options for multi-gate device architecture, such as double-gate, tri-gate, etc., the nanowire (NW) channel with a wrap-around gate, GAA, has the largest advantage in terms of electrostatic integrity. However, several undesired effects become prominent from the miniaturization of the device dimensions. One such unwanted effect is a strong increase in the low-frequency noise generated in the transistor as the size of the device decreases. With dimension-scaling, it is necessary to also scale the power supplies. Noise cannot be completely eliminated and with small signal strength, the accuracy and measurements are limited in electronic circuits in presence of the dominant noise sources. The low-frequency noise, or 1/*f* noise, is the excess noise at low frequencies whose power spectral density ap‐ proximately depends inversely on the frequency and becomes pronounced at low frequen‐ cies. The 1/*f* noise originating from the transistors is a severe obstacle in analog circuit design. For example, it can be up converted to undesired phase noise in voltage controlled oscillator circuits, which can limit the information capacity of communication systems. The problems in down-scaled devices such as the nanowire FinFETs compel one to study noise sources, mechanisms and their physical origins in detail. In this Section, we shall briefly dis‐

go through and

162 Nanowires - Recent Advances

**2.** via the screening effect of the Coulomb potential.

substrate self-heating effects will become increasingly important.

**3. Low-Frequency Noise in Nanowire Transistors**

direction when compared to the [110] direction.

The standard noise measurement setup included an E5263A 2-channel high speed source monitor unit, a SR 570 low noise amplifier (LNA) and a 35670A dynamic signal analyzer. E5263A 2-channel high speed source monitor unit provided the necessary gate-source and drain-source biases as shown in Fig. 10. The minute fluctuations in the drain-source voltage were amplified to the measurable range using low noise amplifier. The output of the ampli‐ fier is fed to 35670A dynamic signal analyzer that performs the fast Fourier transform on the time domain signal to yield the voltage noise power spectral density (SV) in the 1-100 kHz range after correcting for amplifier gain. In order to obtain a stable spectrum, the number of averages was set at 100 and a 90% sampling window overlap was used for optimal real time processing. A computer interface is connected with the measuring system through GPIB connection to control the dynamic signal analyzer and for noise data collection.

**Figure 10.** Low-frequency noise measurement setup.

#### **3.2. Random Telegraph Signal (RTS)**

RTS noise, also known as burst noise or popcorn noise is observed as random switching events in the time domain voltage or current signal. In a MOSFET, if a carrier is trapped in a single trap or localized defect state, the current or voltage signal displays a random shift in the level denoting a change in the channel resistance. The two-level RTS signal signifies only one active trap. However, when multiple traps are involved, the current (or voltage) can switch between two or more states resembling a RTS waveform due to random trapping and de-trapping of carriers and the phenomenon is much more difficult to explain in order to identify the trapping/de-trapping process. For simple two-level RTS pulses with equal height ∆*I* and Poisson distributed mean time durations in the lower state *τ <sup>l</sup>* and in the high‐ er state *τ <sup>h</sup>*, the PSD of the current fluctuations is derived as [17].

$$S\_{l}(f) = \frac{4(\Delta \mathcal{I})^2}{(\overline{\tau\_l} + \overline{\tau\_h})\left[\left(1/\overline{\tau\_l} + 1/\overline{\tau\_h}\right)^2 + \left(2\pi f\right)^2\right]}\tag{7}$$

Mainly two types of traps are identified depending on the nature of trapping mechanism. They are donor and acceptor traps. The donor trap is charged when it emits and electron (i.e. empty) and is neutral when it captures. The acceptor trap is, contrary to the donor trap, charged when it captures an electron and neutral when empty. In MOSFETs, the channel re‐ sistance increases with the charged-trap state changing the current (or voltage) to a high state. Clearly, the donor trap causes high current level after emission of carriers, and the ac‐ ceptor trap causes the high current level when it captures an electron.

Depending on the values of the mean time constants of the RTS, the traps can be character‐ ized of two types. These are the slow traps with high values of time constants, and the fast traps, with the time constants being very small (order of few 0.1 milliseconds). From RTS noise characterizations, interesting information about the trap energy, capture and emission kinetics and spatial location of the traps inside the semiconductor device can be acquired. The multi-level RTS is due to the activation of multiple traps near the quasi-Fermi level. With smaller area devices, only single traps are active as number of traps is less and the RTS becomes a simple two-level signal, with a Lorentzian PSD (1/*f* <sup>2</sup> ).

#### **3.3. 1/f Noise**

1/*f* noise, also called flicker noise or pink noise, is the low-frequency noise with fluctuations with a PSD proportional to 1/*f* <sup>γ</sup> with *γ* close to 1, usually in the range 0.7-1.3. The PSD for 1/*f* noise takes the general form

$$S\_{\nu}(f) = \frac{KI^{\beta}}{f^{\prime}} \tag{8}$$

where *K* is a constant and *β* is current exponent. There are various theories regarding the 1/*f* noise mechanisms of which most prominent are the carrier number fluctuation (surface phe‐ nomenon) and mobility fluctuation theories (bulk phenomenon). The mechanisms behind flicker noise are still a topic of research. The generally accepted origins of the 1/*f* noise are attributed to conductivity fluctuations, damage in crystal structures, and traps due to de‐ fects in semiconductors. 1/*f* fluctuations in the conductance have been observed in the lowfrequency part of the spectrum (10-6 to 106 Hz) in most conducting materials and a wide variety of semiconductor devices [18, 19]. There are essentially two physical mechanisms be‐ hind any fluctuations in the current: fluctuations in the mobility or fluctuations in the num‐ ber of carriers. In 1957, McWorther presented a 1/*f* noise model based on quantum mechanical tunneling transitions of electrons between traps in the gate oxide and the chan‐ nel [20]. The tunneling time varies exponentially with distance from the trap and the 1/*f* noise is obtained for a trap density that is uniform in both energy and distance from the channel interface. The McWorther model is widely accepted for simplicity and its excellent agreement with experimental data, especially for nMOS transistors. However, the mobility fluctuation noise model explains the 1/*f* noise in pMOS transistors better [21]. It was later explained by the unified flicker noise theory that a trapped carrier also affects the surface mobility through Coulombic interaction. This correlated mobility fluctuation model gave a correction to the number fluctuation noise model which resolves the deviations of the theory for pMOS devices. However, the correction factor was debated for being very high since screening was not accounted for. Also, the carrier mobility at the surface is reduced com‐ pared to the bulk mobility due to additional surface scattering (by acoustic phonons and surface roughness), which has an impact on the mobility fluctuations. Moreover, the Hooge mobility noise is sensitive to the crystalline quality, which is deteriorated close to the inter‐ face. The most feasible explanation for the higher 1/*f* noise with the carriers being in close proximity of the gate oxide surface is increased mobility fluctuation noise.

#### **3.4. Low-Frequency Noise**

( ) ( )

 t

DI <sup>=</sup> é ù + ++ ê ú ë û

Mainly two types of traps are identified depending on the nature of trapping mechanism. They are donor and acceptor traps. The donor trap is charged when it emits and electron (i.e. empty) and is neutral when it captures. The acceptor trap is, contrary to the donor trap, charged when it captures an electron and neutral when empty. In MOSFETs, the channel re‐ sistance increases with the charged-trap state changing the current (or voltage) to a high state. Clearly, the donor trap causes high current level after emission of carriers, and the ac‐

Depending on the values of the mean time constants of the RTS, the traps can be character‐ ized of two types. These are the slow traps with high values of time constants, and the fast traps, with the time constants being very small (order of few 0.1 milliseconds). From RTS noise characterizations, interesting information about the trap energy, capture and emission kinetics and spatial location of the traps inside the semiconductor device can be acquired. The multi-level RTS is due to the activation of multiple traps near the quasi-Fermi level. With smaller area devices, only single traps are active as number of traps is less and the RTS

1/*f* noise, also called flicker noise or pink noise, is the low-frequency noise with fluctuations with a PSD proportional to 1/*f* <sup>γ</sup> with *γ* close to 1, usually in the range 0.7-1.3. The PSD for 1/*f*

g

where *K* is a constant and *β* is current exponent. There are various theories regarding the 1/*f* noise mechanisms of which most prominent are the carrier number fluctuation (surface phe‐ nomenon) and mobility fluctuation theories (bulk phenomenon). The mechanisms behind flicker noise are still a topic of research. The generally accepted origins of the 1/*f* noise are attributed to conductivity fluctuations, damage in crystal structures, and traps due to de‐ fects in semiconductors. 1/*f* fluctuations in the conductance have been observed in the low-

variety of semiconductor devices [18, 19]. There are essentially two physical mechanisms be‐ hind any fluctuations in the current: fluctuations in the mobility or fluctuations in the num‐ ber of carriers. In 1957, McWorther presented a 1/*f* noise model based on quantum mechanical tunneling transitions of electrons between traps in the gate oxide and the chan‐ nel [20]. The tunneling time varies exponentially with distance from the trap and the 1/*f*

b

( ) *<sup>I</sup> KI S f <sup>f</sup>* 2 2

(7)

 p*f*

).

= (8)

Hz) in most conducting materials and a wide

2

4( ) ( ) ( ) 1/ 1/ 2 *<sup>I</sup> lh l h*

 t

tt

ceptor trap causes the high current level when it captures an electron.

becomes a simple two-level signal, with a Lorentzian PSD (1/*f* <sup>2</sup>

**3.3. 1/f Noise**

noise takes the general form

frequency part of the spectrum (10-6 to 106

*S f*

164 Nanowires - Recent Advances

Figure 11 shows the RTS of a p-FinFET in time domain in a large span of 60 ms time for *V <sup>d</sup>* = −50 mV and |*V <sup>g</sup> − V th*| = 0.1 V. For this particular RTS, two distinct sets of current switching levels are observed, each of which is consistent with trapping and de-trapping at a single discrete trapping site. A segment of the RTS is enlarged in the inset of Fig. 11 showing a dis‐ tinct two-level RTS. This shows a fast varying RTS (wit smaller time constants) is superim‐ posed on a slow-varying RTS (with higher time constants) which generates the four-level RTS shown in the figure. This RTS is due to a slow (with higher time constant) and a fast trap (with lower time constant). The corresponding drain current spectral density is plotted in Fig. 12, with a corner frequency *f <sup>c</sup>* and a Lorentzian 1/*f* <sup>2</sup> nature. The traps are either donor or acceptor types as described in section 3.2. The high level in current corresponds to the charged trap state. With increasing gate bias, the trap occupancy increases [22]. Also, the time in the high current level increases with gate voltage. So the high level is the occupied charged-trap state or the emission time. The low-current level is the capture time from the same interpretation. The charged trap state is the occupied state which makes the trap to be an acceptor trap. The emission and capture time constants are thus identified in Fig. 12. The mean emission (*τ <sup>e</sup>*) and capture time (*τ <sup>c</sup>*) constants of the RTS in Fig. 11 are shown in Fig. 13 as a function of gate bias. The ratio of capture to emission times (*τ <sup>c</sup> / τ <sup>e</sup>*) shows a gradual decrease with gate voltage. The high current time being *τ <sup>e</sup>* and the low current time as *τ <sup>c</sup>* The following can be written [22]:

$$\ln\left(\frac{\mathbf{r}\_c}{\mathbf{r}\_c}\right) = -\frac{1}{kT}\left| \begin{pmatrix} (E\_{cox} - E\_T) - (E\_c - E\_F) - \phi\_0 \\ + q\boldsymbol{\wp}\_s + \frac{q\boldsymbol{\wp}\_T}{t\_{ox}}(V\_{GS} - V\_{FB} - \boldsymbol{\wp}\_s) \end{pmatrix} \right| \tag{9}$$

where *E Cox* is the conduction band edge of the oxide, EC is the conduction band edge of the silicon, *Φ <sup>0</sup>* is the difference between the electron affinities of Si and SiO2, *t ox* is the oxide thickness, *V GS* is the gate bias, *V FB* is the flat band voltage, *ψ <sup>s</sup>* is the surface potential and *x <sup>T</sup>* is the position of the trap measured from the Si/SiO2 interface. Differentiating in terms of the gate bias we can obtain the position of traps as:

$$\frac{d}{dV\_{GS}}\ln\left(\frac{\tau\_c}{\tau\_e}\right) = -\frac{1}{kT}\left[\frac{q\mathbf{x}\_T}{t\_{ox}}\right] \tag{10}$$

From Fig. 13, using Eq. (10), the trap depth is calculated as 0.172 nm from the Si/SiO2 inter‐ face. A Lorentzian expression, as expected for an RTS [23, 24], is obtained as

$$\text{WL}\frac{S\_{I\_d}(f)}{\left(\int\_d^2} = \frac{A}{\left[1 + \left(f \,/\, f\_c\right)^2\right]}\tag{11}$$

where *A* is a constant independent of frequency, *f <sup>c</sup>* = (1/2*π*)(1/*τ <sup>c</sup>* + 1/*τ <sup>e</sup>*), and *τ <sup>c</sup>* and *τ <sup>e</sup>* are the capture and emission times, respectively, of a single carrier by an interface trap. It can be seen that the corner frequency *f <sup>c</sup>* ∼ 1 kHz, which is consistent with the observed switching times in the order of 0.4–1.4 ms.

**Figure 11.** RTS of a p-FinFET in time domain for *Vds* = −50 mV and |*V gs* − V *th*| = 0.1 V, depicting four levels with at least two traps; the inset shows a zoom of the fast RTS with the capture and emission time constants shown in the waveform.

where *E Cox* is the conduction band edge of the oxide, EC is the conduction band edge of the silicon, *Φ <sup>0</sup>* is the difference between the electron affinities of Si and SiO2, *t ox* is the oxide thickness, *V GS* is the gate bias, *V FB* is the flat band voltage, *ψ <sup>s</sup>* is the surface potential and *x <sup>T</sup>* is the position of the trap measured from the Si/SiO2 interface. Differentiating in terms of the

> <sup>1</sup> ln *<sup>c</sup> <sup>T</sup> GS e ox d qx dV kT t* t

æö é ù

From Fig. 13, using Eq. (10), the trap depth is calculated as 0.172 nm from the Si/SiO2 inter‐

( ) <sup>2</sup> <sup>2</sup>

*d c*

*I f f* <sup>=</sup> é ù <sup>+</sup> ë û

1 /

where *A* is a constant independent of frequency, *f <sup>c</sup>* = (1/2*π*)(1/*τ <sup>c</sup>* + 1/*τ <sup>e</sup>*), and *τ <sup>c</sup>* and *τ <sup>e</sup>* are the capture and emission times, respectively, of a single carrier by an interface trap. It can be seen that the corner frequency *f <sup>c</sup>* ∼ 1 kHz, which is consistent with the observed switching

**Figure 11.** RTS of a p-FinFET in time domain for *Vds* = −50 mV and |*V gs* − V *th*| = 0.1 V, depicting four levels with at least two traps; the inset shows a zoom of the fast RTS with the capture and emission time constants shown in the waveform.

ç ÷ = - ê ú èø ë û (10)

(11)

t

face. A Lorentzian expression, as expected for an RTS [23, 24], is obtained as

( )

*S f <sup>A</sup> WL*

*dI*

gate bias we can obtain the position of traps as:

166 Nanowires - Recent Advances

times in the order of 0.4–1.4 ms.

**Figure 12.** Drain current noise power spectral density with a corner frequency *fc* and a Lorentzian nature (1/*f* 2).

**Figure 13.** The mean emission (τ*e*) and capture time (τ*c*) constants and the ratio of τ*c*/τ*<sup>e</sup>* of the RTS are shown as a func‐ tion of gate bias.

The entire FinFET channel is very thin and close to the interface and thus will be highly sensitive to interface defects. The defect potential perturbation, on the order of several nanometres, will have an effect on a significant cross section of the 10-nm-wide fin, hence producing a large current change when the trap is filled. The relative amplitude Δ*I <sup>d</sup>*/*I <sup>d</sup>* can be written as [25]

$$\frac{\Delta I\_d}{I\_d} = \frac{1}{WL} \left[ \frac{1}{N} \pm \alpha \mu \right] \tag{12}$$

where *N* is the carrier density, *α* is the scattering coefficient, and *μ* is the carrier mobility. Using the double lateral channel FinFET approximation [26], *N = 2 C ox |V <sup>G</sup> − V TH*|/*q* and *μ = g <sup>m</sup> L/*(*2WC ox V <sup>d</sup>*). The mobility limited by Coulomb scattering is calculated as *μ <sup>C</sup>* = 1/*αN <sup>t</sup>* , where *N <sup>t</sup>* is the density of occupied traps [25].

#### **4. Orientation Dependence: Nanowire FinFETs**

Theory and measurements on planar transistors show a large difference in mobility for elec‐ trons and holes depending on the interface orientation and the direction of current flow in MOSFETs [27, 28, 29]. Figure 14 illustrates the measured drain currents (ID-VDS) for two dif‐ ferent channel orientations; (100) and (110). It is observed that the current is higher in (100) orientation compared to the (110) orientation. The change in channel crystallographic direc‐ tion (and thereby a change in the direction of current flow) greatly influences the carrier mo‐ bilities due to direction dependency of the heavy hole effective mass affecting the low field hole mobility. The effect of channel orientation is dominant in p-type devices due to differ‐ ent band curvatures along the <100> and <110> channel directions.

**Figure 14.** Comparison of the measured drain currents (ID-VDS) for two different channel orientations; (100) and (110) of a p-type tri-gate Si-FinFET.

1 1 *<sup>d</sup>*

<sup>D</sup> é ù = ± ê ú

where *N* is the carrier density, *α* is the scattering coefficient, and *μ* is the carrier mobility. Using the double lateral channel FinFET approximation [26], *N = 2 C ox |V <sup>G</sup> − V TH*|/*q* and *μ = g <sup>m</sup> L/*(*2WC ox V <sup>d</sup>*). The mobility limited by Coulomb scattering is calculated as *μ <sup>C</sup>* = 1/*αN <sup>t</sup>*

Theory and measurements on planar transistors show a large difference in mobility for elec‐ trons and holes depending on the interface orientation and the direction of current flow in MOSFETs [27, 28, 29]. Figure 14 illustrates the measured drain currents (ID-VDS) for two dif‐ ferent channel orientations; (100) and (110). It is observed that the current is higher in (100) orientation compared to the (110) orientation. The change in channel crystallographic direc‐ tion (and thereby a change in the direction of current flow) greatly influences the carrier mo‐ bilities due to direction dependency of the heavy hole effective mass affecting the low field hole mobility. The effect of channel orientation is dominant in p-type devices due to differ‐

**Figure 14.** Comparison of the measured drain currents (ID-VDS) for two different channel orientations; (100) and (110)

am

ë û (12)

,

*I WL N*

*d I*

is the density of occupied traps [25].

**4. Orientation Dependence: Nanowire FinFETs**

ent band curvatures along the <100> and <110> channel directions.

where *N <sup>t</sup>*

168 Nanowires - Recent Advances

of a p-type tri-gate Si-FinFET.

**Figure 15.** Comparison of the measured gate voltage noise power spectral density for two different channel orienta‐ tions; (100) and (110) of a p-type tri-gate Si-FinFET.

The input referred gate voltage noise power spectral density (PSD) of the FinFET is shown in Fig. 15 for two different channel orientations ((100) and (110)) as a function of the gatesource voltage. It is observed that the gate voltage PSDs are comparable in magnitude for both orientations. It is well known that technological factors such as gate oxide quality, gate oxide material, channel material, conduction path, annealing, etc. can be of great importance for the low-frequency noise properties [30, 31, 32, 33]. The gate oxide quality and other tech‐ nological factors can be considered as unchanged by the channel rotation, which explains why the orientation displays a negligible influence on the low-frequency noise. If the mobili‐ ty fluctutation noise becomes dominant, the channel orientation may play significant part in determining the noise PSD. This is due to the fact that the carrier mobiliy depends on chan‐ nel orientation. The difference in noise PSD for the two orientation is large in the subthres‐ hold region. This is due to the fact that the mobility-fluctuation noise generation mechanism is dominant in the subthreshold region, since in the number-fluctuation model, there is no dependence of noise PSD on mobility at a fixed drain-current level. The channel orientation dependence of noise is more dominant in pMOS devices as also reported in reference 34. Stronger current and mobility enhancement in certain orientations may also cause a higher LFN level in the corresponding orientation.

#### **5. Lattice Dynamics: Nanowire FinFETs**

Lattice vibrations in small geometry devices and its orientation-dependence are important, which however, have not been studied in detail. Inelastic electron tunneling spectroscopy (IETS) has gained importance for characterization of small-geometry devices and ultra-thininsulators due to its high resolution and sensitivity and has been used by many previous re‐ searchers for studying defects and molecular vibrations in metal-insulator-semiconductor (MIS) structures [35, 36, 37, 38, 39, 40, 41].

The principle of IETS depends on the tunneling current through the ultra-thin dielectric lay‐ er of the MIS structures and becomes more sensitive as the tunneling current increases. Tun‐ neling spectroscopy detects interaction of tunneling electrons (through the dielectric) with lattice vibrations of the substrate, dielectric, electrodes or other molecular species as inelastic peaks in the second derivative of the current-voltage characteristics. The application of tun‐ neling spectroscopy was first demonstrated by Hall [36]. The technique of inelastic electron tunneling spectroscopy was properly introduced in 1966 by Jaklevic and Lambe [37]. With high sensitivity and resolution, IETS has been used for various applications including the determination of density of states, molecular vibrations, band gap of superconductors and semiconductors, defects in semiconductors and insulators [38, 39]. IETS has also been used for studying the traps and molecular vibrations in ultra thin oxides [40] and high-k gate die‐ lectrics [41]. A tunneling spectrum reveals mainly the vibration energies of the molecules be‐ tween the electrodes and phonon modes. Yanson *et al.* have shown that the intensity of the inelastic tunneling process is characterized by the relative change in Δ*g*/*g* for conductance *g* at energy *ħω*. The quantity Δ*g*/*g* is related to the observed I-V characteristics by the follow‐ ing expression [42]:

$$\frac{\Delta \mathbf{g}}{\mathbf{g}} = V \int\_{V\_1}^{V\_2} \frac{1}{I} \frac{d^2 I}{dV^2} dV \tag{13}$$

where, *V <sup>1</sup>* and *V <sup>2</sup>* are the starting and ending voltages of the vibration band under consider‐ ation. This relation can be used to quantify the strength of the vibration modes, i.e. the change in the Δg/g ratio indicates the strength of the interactions. In the following, we present the experimental tunneling spectra of Si, SiO2 phonon modes obtained for p-type Al/SiO2/Si tri-gate FinFETs and study the influence of crystal orientation. A comparison of lattice vibration modes and dispersion characteristics for (100) and (110) silicon orientation in tri-gate FinFET is made. It is observed that the lattice dynamics of silicon and SiO2 changes for different crystal orientations.

The low temperature (77K) inelastic electron tunneling spectroscopy measurement setup in‐ cludes an E5263A 2-channel high speed source monitor unit for dc biasing, and a SR830 DSP lock-in amplifier for capturing the tunneling spectra by measuring the second derivatives of the current-voltage characteristics. The setup is shown in Fig. 16. AC modulation signal gen‐ erated from oscillator output of the lock-in amplifier with *f* = 890 Hz, peak amplitude of 2 mV, is superimposed on a slowly varying dc gate voltage. A time constant of 3 sec was used. A notch filter in the lock-in amplifier was used to remove the unwanted harmonic frequen‐ cies. A computer interface is connected with the measuring system through GPIB connection to control the dynamic signal analyzer and for noise data collection.

**Figure 16.** IETS measurement setup.

searchers for studying defects and molecular vibrations in metal-insulator-semiconductor

The principle of IETS depends on the tunneling current through the ultra-thin dielectric lay‐ er of the MIS structures and becomes more sensitive as the tunneling current increases. Tun‐ neling spectroscopy detects interaction of tunneling electrons (through the dielectric) with lattice vibrations of the substrate, dielectric, electrodes or other molecular species as inelastic peaks in the second derivative of the current-voltage characteristics. The application of tun‐ neling spectroscopy was first demonstrated by Hall [36]. The technique of inelastic electron tunneling spectroscopy was properly introduced in 1966 by Jaklevic and Lambe [37]. With high sensitivity and resolution, IETS has been used for various applications including the determination of density of states, molecular vibrations, band gap of superconductors and semiconductors, defects in semiconductors and insulators [38, 39]. IETS has also been used for studying the traps and molecular vibrations in ultra thin oxides [40] and high-k gate die‐ lectrics [41]. A tunneling spectrum reveals mainly the vibration energies of the molecules be‐ tween the electrodes and phonon modes. Yanson *et al.* have shown that the intensity of the inelastic tunneling process is characterized by the relative change in Δ*g*/*g* for conductance *g* at energy *ħω*. The quantity Δ*g*/*g* is related to the observed I-V characteristics by the follow‐

2

1 *<sup>V</sup>*

2 2

where, *V <sup>1</sup>* and *V <sup>2</sup>* are the starting and ending voltages of the vibration band under consider‐ ation. This relation can be used to quantify the strength of the vibration modes, i.e. the change in the Δg/g ratio indicates the strength of the interactions. In the following, we present the experimental tunneling spectra of Si, SiO2 phonon modes obtained for p-type Al/SiO2/Si tri-gate FinFETs and study the influence of crystal orientation. A comparison of lattice vibration modes and dispersion characteristics for (100) and (110) silicon orientation in tri-gate FinFET is made. It is observed that the lattice dynamics of silicon and SiO2

The low temperature (77K) inelastic electron tunneling spectroscopy measurement setup in‐ cludes an E5263A 2-channel high speed source monitor unit for dc biasing, and a SR830 DSP lock-in amplifier for capturing the tunneling spectra by measuring the second derivatives of the current-voltage characteristics. The setup is shown in Fig. 16. AC modulation signal gen‐ erated from oscillator output of the lock-in amplifier with *f* = 890 Hz, peak amplitude of 2 mV, is superimposed on a slowly varying dc gate voltage. A time constant of 3 sec was used. A notch filter in the lock-in amplifier was used to remove the unwanted harmonic frequen‐ cies. A computer interface is connected with the measuring system through GPIB connection

to control the dynamic signal analyzer and for noise data collection.

<sup>D</sup> <sup>=</sup> ò (13)

1

*V g dI V dV g I dV*

(MIS) structures [35, 36, 37, 38, 39, 40, 41].

ing expression [42]:

170 Nanowires - Recent Advances

changes for different crystal orientations.

The phonon spectra of Si are observed in the 0-70 mV range. Figure 17(a)shows the inelastic peaks in the phonon spectra of Si for FinFET on (100) silicon orientation obtained from the average of six scans after de-convolution of the IETS spectra. The most intense peak is found at 60 mV (wave number 483.9 cm-1). The peak is due to optical phonons at the conduction band minima (X-point) of silicon in transverse optical (TO) mode. Other significant Si-pho‐ nons include longitudinal acoustical (LA) phonon peak towards the X point at 47.2 mV (380.6 cm-1), the longitudinal optical (LO) phonon towards X point at 53 mV (427.4 cm-1). The transverse acoustic (TA) mode was observed at 19.7 mV (158.9 cm-1). Phonons of Al-electro‐ des are observed at 26 and 33 mV. These findings are in well agreement with reported re‐ sults for Si phonons with (100) orientation [35]. The values of phonon frequencies of unstrained silicon were calculated by Sui et al. using a modified Keating model [43]. The ex‐ perimental values are compared with these theoretical values. We suspect the minor shifts in peak locations are due to the inherent strain in the small geometry structure of the Fin‐ FET. The TO and LO mode towards the Γ point, however, was not observed from the IETS. Figure 17(b)shows similar phonon modes of silicon for (110) orientation. From the Brillouin zone of the Si-crystal the dispersion in (110) direction is calculated along the main symmetry directions Γ→X through K point and X→X through W point [44]. The identified peaks in‐ clude TO mode at Γ and W points, respectively, at 68 (548.4 cm-1) and 58 mV (467.7 cm-1). LO and LA mode towards X point are found at 54 (435.5 cm-1) and 45.6 mV (367.7 cm-1), and LA towards K point is observed at 41 mV (330.6 cm-1). The TA modes at X and W points are observed at 19.5 (157.3 cm-1) and 26 mV (209.7 cm-1), respectively. The peaks at W point or K point are mostly of weaker intensity compared to the peaks at Γ and X points, which are the conduction band minima for the lattice. Also, unlike the (100) orientation, the symmetry di‐ rections in (110) orientation reveal inelastic interactions at the K and W points along with the Γ and X points which describes the Brillouin zone for Si in more detail.

**Figure 17.** IETS spectra showing Si-phonons in a Al/SiO2/Si FinFET for (a) (100) channel orientation and (b) (110) chan‐ nel orientation.

Figures 18 (a) and (b) show the SiO2 phonon modes (in the 129-170 mV range) for (100) and (110) orientations, respectively. The peaks exhibit almost the same energy position for both the Si (110) and Si (100) devices. These peaks include, TO3 mode at 134 mV, LO4 at 144.2 mV, TO4 at 149.8 and 153.4 mV for (100) device. The corresponding peaks for (110) devices are observed at 132, 144.2 148 and 159 mV, respectively. The main discrepancy between the two orientations comes in the magnitude of the phonons. The asymmetric stretch TO4 mode is theoretically found at 148.8 mV [45] which appears as a strong peak in the Si (110) device, and as a weak peak in the (100) orientation, as also described by [35]. The ratio of intensities of vibration modes may vary for different orientations of the molecular dipoles in the tunnel oxide barrier for different device orientations.

**Figure 18.** IETS spectra showing SiO2-phonons in a Al/SiO2/Si FinFET for (a) (100) channel orientation and (b) (110) channel orientation.

### **6. Summary**

rections in (110) orientation reveal inelastic interactions at the K and W points along with

**Figure 17.** IETS spectra showing Si-phonons in a Al/SiO2/Si FinFET for (a) (100) channel orientation and (b) (110) chan‐

Figures 18 (a) and (b) show the SiO2 phonon modes (in the 129-170 mV range) for (100) and (110) orientations, respectively. The peaks exhibit almost the same energy position for both the Si (110) and Si (100) devices. These peaks include, TO3 mode at 134 mV, LO4 at 144.2 mV, TO4 at 149.8 and 153.4 mV for (100) device. The corresponding peaks for (110) devices are observed at 132, 144.2 148 and 159 mV, respectively. The main discrepancy between the two orientations comes in the magnitude of the phonons. The asymmetric stretch TO4 mode is theoretically found at 148.8 mV [45] which appears as a strong peak in the Si (110) device, and as a weak peak in the (100) orientation, as also described by [35]. The ratio of intensities of vibration modes may vary for different orientations of the molecular dipoles in the tunnel

**Figure 18.** IETS spectra showing SiO2-phonons in a Al/SiO2/Si FinFET for (a) (100) channel orientation and (b) (110)

the Γ and X points which describes the Brillouin zone for Si in more detail.

nel orientation.

172 Nanowires - Recent Advances

channel orientation.

oxide barrier for different device orientations.

In this chapter, we have discussed silicon nanowire FinFETs in detail. Several important electrical characteristics such as mobility, self-heating, low-frequency noise, impact of chan‐ nel crystallographic orientation and lattice dynamics are presented for nanowire FinFETs. The fabrication process flow of the nanowire and FinFETs are described. Application of split C-V measurement to calculate effective carrier mobilities is shown and mobility models for FinFETs including side and top-wall mobilities are developed. Low-frequency noise measure‐ ments are performed and results of 1/*f* noise and RTS in FinFET devices are presented. Ef‐ fects of channel orientation on current drive and low-frequency noise are discussed. Lattice vibrations and phonon mode of Si and SiO2 are studied for (100) and (110) channel orienta‐ tions and Brillouin zone boundaries of the semiconductor are explored. Crystal dynamics in FinFETs are explained via inelastic electron tunneling spectroscopy. It is expected that the combination of strain effects with NWs can lead to very high performance devices.

## **Author details**

C. Mukherjee and C. K. Maiti

\*Address all correspondence to: ckm@ece.iitkgp.ernet.in

Indian Institute of Technology, Kharagpur, India

## **References**


[17] Machlup, S. (1954). Noise in semiconductors: spectrum of a two-parameter random signal. *J. Appl. Phys.*, 25, 341-343.

[5] Zhu, W., Jin-Ping, H., & Ma, T. P. (2004). Mobility measurement and degradation mechanisms of MOSFETs made with ultrathin high-k dielectrics. *IEEE Trans. Electron*

[6] Cui, Y., Zhong, Z., Wang, D., Wang, W. U., & Lieber, C. M. (2003). High Performance

[7] Suk, S. D., Lee, S., , Y., Kim, S. M., Yoon, E. J., Kim, M. S., Li, M., Oh, C. W., Yeo, K. H., Kim, S. H., Shin, D. S., Lee, K. H., Park, H. S., Han, J. N., Park, C. J., Park, J. B., Kim, D. W., Park, D., & Ryu, B. I. (2005, Dec). High Performance 5nm radius Twin Silicon Nanowire MOSFET (TSNWFET) Fabrication on Bulk Si Wafer, Characteris‐

[8] Chen, J., Saraya, T., Miyaji, K., Shimizu, K., & Hiramoto, T. (2008). Experimental study of mobility in [110] and [100]-directed multiple silicon nanowire GAA MOS‐

[9] Koo, S. M., Fujiwara, A., Han, J. P., Vogel, E. M., Richter, C. A., & Bonevich, J. E. (2004, Sep). High Inversion Current in Silicon Nanowire Field Effect Transistors.

[10] Young, K. K. (1989). Short-channel effect in fully depleted SOI MOSFETs. IEEE

[11] Yang, F. L., Lee, D. H., Chen, H. Y., Chang, C. Y., Liu, S. D., Huang, C. C., Chung, T. X., Chen, H. W., Huang, C. C., Liu, Y. H., Wu, C. C. , Chen, C. C., Chen, S. C., Chen, Y. T., Chen, Y. H., Chen, C. l., Chan, B. W., Hsu, P. F., Shieh, J. H., Tao, H. J., Yeo, Y. C., Li, Y., Lee, J. W., Chen, P., Liang, M. S., & Hu, C. (2004). 5nm-Gate Nanowire Fin‐

[12] Patnaik, V. S., Gheedia, A., & Kumar, M. J. (2008, July, Aug, Sept.,). 3D Simulation of Nanowire FETs using Quantum Models. *SILVACO simulation standard*, 6-12.

[13] SILVACO Inc. (2010). Patrick Henry Drive, Santa Clara, CA 95054 ATLAS User's

[14] Hashemi, P. (2010). Gate-All-Around Silicon Nanowire MOSFETs: Top-down Fabri‐ cation and Transport Enhancement Techniques. *PhD Dissertation, Massachusetts Insti‐*

[15] Najmzadeh, M., Bouvet, D., Dobrosz, P., Olsen, S., & Ionescu, A. M. (2009). Investiga‐ tion of oxidation-induced strain in a top-down Si nanowire platform. *Microelectronic*

[16] Ramos, J., Severi, S., Augendre, E., Kerner, C., Chiarella, T., Nackaerts, A., Hoffmann, T., Collaert, N., Jurczak, M., & Biesemans, S. (2006). Effective Mobility Extraction Based on a Split RF C-V Method for Short-Channel FinFETs. *Proc. Solid-State Device*

Silicon Nanowire Field Effect Transistors. *Nano Lett*, 3(2), 149.

tics, and Reliability. *Proc. IEEE IEDM Tech. Dig*, 735-738.

FETs on (100) SOI. *Proc. VLSI Symp. Tech. Dig*, 32-33.

*Dev*, 51(1), 98-105.

174 Nanowires - Recent Advances

*Nano Lett*, 4(11), 2197-2201.

Manual. , April 20.

*tute of Technology.*

*Engineering*, 86(7-9), 1961-1964.

*Research Conference, (ESSDERC)*, 363-366.

Trans. Electron Dev Feb., 36(2), 399-402.

FET. Proc. VLSI Symp. Tech. Dig. , 196-197.


## **Hybrid Silicon Nanowires: From Basic Research to Applied Nanotechnology**

Muhammad Y. Bashouti, Matthias Pietsch, Kasra Sardashti, Gerald Brönstrup, Sebastian W. Schmitt, Sanjay K. Srivastava, Jürgen Ristein, Jordi Arbiol, Hossam Haick and Silke Christiansen

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54383

## **1. Introduction**

[32] Von, Haartman. M., Lindgren, A. C., Hellstrom, P. E., Malm, B. G., Zhang, S. L., & Ostling, M. (2003). 1/f noise in Si and Si0.7Ge0.3 pMOSFETs. *IEEE Trans. Electron Dev*,

[33] Vandamme, L. K. J., & Oosterhoff, S. (1986). Annealing of ion-implanted resistors re‐

[34] Wei, C., Xiong-Z, Y., Zhou, X., Singh, N., Rustagi, S. C., Lo, G. Q., & Kwong, D. L. (2009, June). Investigation of Low-Frequency Noise in Silicon Nanowire MOSFETs in

[35] Petit, C., Salace, G., & Vuillaume, D. (2004). Aluminum, oxide, and silicon phonons by inelastic electron tunneling spectroscopy on metal-oxide-semiconductor tunnel junctions: Accurate determination and effect of electrical stress. *J. Appl. Phys*, 96(9),

[36] Hall, R. N., Racette, J. H., & Ehrenreich, H. (1960, May). Direct observations of Polar‐ ons and Phonons during tunneling in group 3-5 semiconductor junctions. *Phys. Rev.*

[37] Jaklevic, R. C., & Lambe, J. (1966). Molecular vibration spectra by electron tunneling.

[38] Giaever, I., & Megerle, K. (1961). Study of superconductors by electron tunneling.

[39] Chynoweth, A. G., Logan, R. A., & Thomas, D. E. (1962). Phonon- Assisted Tunneling

[40] Petit, C., & Salace, G. (2003). Inelastic electron tunneling spectrometer to characterize metal-oxide-semiconductor devices with ultrathin oxides. *Rev. Sci. Instrum*, 74, 4462.

[41] He, W., & Ma, T. P. (2003). Inelastic Electron Tunneling Spectroscopy Study of Traps

[42] Yanson, I. K., Bogatina, N. I., Verkin, B. J., & Shklyarevskii, O. I. (1972). Asymmetry of Tunnel Spectrum Intensities of Impurity Organic Molecules. *Sov. Phys. JETP*, 35,

[43] Sui, Z., & Herman, I. P. (1993). Effect of strain on phonons in Si, Ge, and Si/Ge hetero‐

[44] Tubino, R., Piseri, L., & Zerbi, G. (1972). Lattice Dynamics and Spectroscopic Proper‐ ties by a Valence Force Potential of Diamondlike Crystals: C, Si, Ge, and Sn. *J. Chem.*

[45] Kirk, C. T. (1988). Quantitative analysis of the effect of disorder induced mode cou‐

pling on infrared absorption in silica. *Phys. Rev. B*, 38, 1255-1273.

in Ultra-thin High-k Gate Dielectrics. *Appl. Phys. Lett*, 83(26), 5461-5463.

in Silicon and Germanium Esaki Junctions. *Phys. Rev*, 125, 877-881.

the Subthreshold Region. *IEEE Electron Dev. Lett.*, 30(6), 668-671.

duces the 1/f noise. *J. Appl. Phys.*, 59, 3169-3174.

50, 2513-2519.

176 Nanowires - Recent Advances

5042-5049.

540.

*Phys*, 56(3), 1022.

*Lett*, 4, 456-458.

*Phys. Rev. Lett.*, 17, 1139-1140.

structures. *Phys. Rev. B*, 48, 17938-17953.

*Phys. Rev*, 122, 1101-1111.

Silicon nanowires (SiNWs) have attracted particular attention in nanotechnology applica‐ tions due to their unique advantages with respect to electrical, optical and thermoelec‐ tric properties compared to the planar thin films [1-3]. SiNWs show strong absorption of visible light in thin layers, which makes them attractive for the use in novel thin film concepts as in photovoltaics [4]. Like all the one-dimensional (1D) nanostructures, SiNWs have a large surface to volume ratio and thus surface-dominated properties that can be tuned, e.g. by suitable surface functionalization to be applied in sensitive sensors [5, 6]. Surface functionalization affects the overall electrical properties of SiNWs as dop‐ ants surface concentration, surface recombination rate, density of surface states, etc.[2, 7-10]. Atop Silicon atoms on SiNW surfaces can be terminated with a wide variety of molecules bearing covalent interfacial bonds i.e. Si-C [2, 7-10], Si-N [11] and Si-O [12]. Up to now we essentially exploited the Si-C bonds that permit a thorough control of the SiNW surface yielding a strongly reduced tendency to the undesirable oxidation of the SiNWs [2]. With such surface functionalization methods at hand, control and prediction of the electrical transport properties have become so viable that functionalized SiNWs can be considered as auspicious nanoscale building blocks in future high-performance nano-devices in the areas of electronics, opto-electronics, photovoltaics and sensing [13-18]. In this chapter we present a chemically-based surface functionalization method

© 2012 Bashouti et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Bashouti et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

based on Si-C bond surface termination that permits to fine-tune the electrical proper‐ ties of VLS-grown SiNWs [2, 7-10, 19].

To date, electrical properties of SiNWs were controlled by doping with either boron (p-type) or phosphorus (n-type) [20]. Electrical transport measurements indicate a reduced resistivity according to higher doping levels and a tunable conductivity of doped SiNWs depending on the gate potential (Vg) applied. Although changes of Vg provide temporary control of SiNW conductivity, a more permanent and robust method is desirable for SiNWs to be useful in nanoelectronic device applications.

A promising approach to control the electronic properties of SiNWs is to use hybrid mono‐ layers (either polar or nonpolar molecules) that affects the surface state density and displays an electrical potential on the surface/interface which modify the work function, electron af‐ finity, surface Fermi level, and band offset/band bending at an interface [21-23]. This func‐ tionalization effect is a general one and can be obtained with non-molecular treatments as well [24]. Nevertheless, the use of molecules, especially organic ones, allows a systematic tuning of the desired dipole moment by an appropriate choice of their functional groups [25] or the intermolecular interactions between self-assembled molecules [10]. A surface functionalization of SiNWs with organic functionalities can be as important as the effect of the diameter and the wire orientation. In fact, due to the very high surface-to-volume ratio, the functionalization of the surface and the immediate surroundings of the wire is expected to have a dominant impact on its properties. Some computational studies have shown the importance of passiveness in SiNWs [26, 27]. Similarly, the large surface-to-volume ratio in SiNWs can be exploited in SiNW functionalization and used to modify transport properties. SiNWs tend to form a layer of SiO2 with thickness 1–2nm under ambient conditions. Subse‐ quent HF treatment can be used to remove this oxide layer. Different Si surface treatments can be performed to passivate SiNW surfaces with organic functionalities and obtain de‐ sired electrical surface properties [2, 7-9].

## **2. Definition: What hybrid material means?**

Although the term "hybrid material" is used to describe many diverse systems straddling a widespread area of different materials, we restrict it by defining the concept as a composi‐ tion and structure with/without interactions between the inorganic and organic units. In general we can distinguish between two possible classes. *Class I* hybrid materials are those that show weak interactions between the two phases, such as van der Waals, hydrogen bonding or weak electrostatic interactions. *Class II* hybrid materials are those that show strong chemical interactions between the components as in is our case.

#### **2.1. Artificial strategies towards hybrid materials**

There are two main approaches which are used to form hybrid materials:

**1.** Building block approach: A well-defined preformed matrix is applied to react with build‐ ing blocks to form the final hybrid material. Regularly, the matrix consists of at least one functional group that allows an interaction with the building blocks. A representative ex‐ ample of this approach is the bonding of organic molecules to silicon surfaces.

**2.** In situ formation of the components: One or more structural units are formed from the precursors that are transformed into a novel network structure. In these cases well-de‐ fined discrete molecules are transformed to multidimensional structures, which often show totally different properties from the original precursors. Typical examples of this approach are polymer reactions and Sol-Gel techniques.

We use the building block approach since it has a main advantage associated with the *in situ* formation in which the structural unit will remain well-defined and usually does not suffer from significant structural changes during the hybrid-matrix formation. Additionally, the building blocks or the matrix can be designed independently in such a way to give the best performance and to be designed to our desire. For example polar or non-polar molecules can be attached to the silicon matrix for sensing application or for FETs applications.

#### **2.2. Chlorination/Alkylation process**

based on Si-C bond surface termination that permits to fine-tune the electrical proper‐

To date, electrical properties of SiNWs were controlled by doping with either boron (p-type) or phosphorus (n-type) [20]. Electrical transport measurements indicate a reduced resistivity according to higher doping levels and a tunable conductivity of doped SiNWs depending on the gate potential (Vg) applied. Although changes of Vg provide temporary control of SiNW conductivity, a more permanent and robust method is desirable for SiNWs to be useful in

A promising approach to control the electronic properties of SiNWs is to use hybrid mono‐ layers (either polar or nonpolar molecules) that affects the surface state density and displays an electrical potential on the surface/interface which modify the work function, electron af‐ finity, surface Fermi level, and band offset/band bending at an interface [21-23]. This func‐ tionalization effect is a general one and can be obtained with non-molecular treatments as well [24]. Nevertheless, the use of molecules, especially organic ones, allows a systematic tuning of the desired dipole moment by an appropriate choice of their functional groups [25] or the intermolecular interactions between self-assembled molecules [10]. A surface functionalization of SiNWs with organic functionalities can be as important as the effect of the diameter and the wire orientation. In fact, due to the very high surface-to-volume ratio, the functionalization of the surface and the immediate surroundings of the wire is expected to have a dominant impact on its properties. Some computational studies have shown the importance of passiveness in SiNWs [26, 27]. Similarly, the large surface-to-volume ratio in SiNWs can be exploited in SiNW functionalization and used to modify transport properties. SiNWs tend to form a layer of SiO2 with thickness 1–2nm under ambient conditions. Subse‐ quent HF treatment can be used to remove this oxide layer. Different Si surface treatments can be performed to passivate SiNW surfaces with organic functionalities and obtain de‐

Although the term "hybrid material" is used to describe many diverse systems straddling a widespread area of different materials, we restrict it by defining the concept as a composi‐ tion and structure with/without interactions between the inorganic and organic units. In general we can distinguish between two possible classes. *Class I* hybrid materials are those that show weak interactions between the two phases, such as van der Waals, hydrogen bonding or weak electrostatic interactions. *Class II* hybrid materials are those that show

**1.** Building block approach: A well-defined preformed matrix is applied to react with build‐ ing blocks to form the final hybrid material. Regularly, the matrix consists of at least one

ties of VLS-grown SiNWs [2, 7-10, 19].

178 Nanowires - Recent Advances

nanoelectronic device applications.

sired electrical surface properties [2, 7-9].

**2. Definition: What hybrid material means?**

**2.1. Artificial strategies towards hybrid materials**

strong chemical interactions between the components as in is our case.

There are two main approaches which are used to form hybrid materials:

The organic building blocks were connected to SiNW surfaces following the building-block approach. We report on the functionalization of SiNWs with alkyl chains using a versatile two step chlorination/alkylation process [8]. The two step process found to be gentle in the sense that it did not break the matrix (i.e.) SiNWs or change their diameters.

The SiNWs were terminated with alkyl chains using a two-step chlorination/alkylation route (see Figure 1). Before any chemical treatment, each sample was cleaned by N2 flow. After being cleaned, H-terminated SiNWs were prepared by etching the amorphous SiOx coating by exposing the SiNWs to buffered HF solution (pH=5) for 60 s and then NH4F for 30 s. The samples were then rinsed in water for <10 s to limit oxidation, dried in flowing N2(g) for 10 s, and immersed into a saturated solution of PCl5 in C6H5Cl (0.65 M) that contained a few grains of C6H5OOC6H5 that act as a radical initiator. The reaction solution was heated to 90-100 °C for 5-7 min. Then the sample was removed from the reaction solution and rinsed in tetrahydrofuran (THF) followed by rinsing in methanol and then dried under a stream of N2(g). The Chlorine (Cl) terminated SiNWs were alkylated by immersion in 0.5 M alkyl Grignard, R-MgCl, in THF, with R= methyl (CH3; hereinafter C1), ethyl (CH3CH2; hereinafter C2), propyl (CH3(CH2)2; hereinafter C3), butyl (CH3(CH2)3; hereinafter C4), pentyl (CH3(CH2)4; hereinafter C5), or hexyl (CH3(CH2)5; hereinafter C6), or Octyl(CH3(CH2)7; here‐ inafter C8), or Nonyl(CH3(CH2)8; hereinafter C9), or Decyl(CH3(CH2)7; hereinafter C10).

The reaction was performed for 5 to 4320 min (72 h) at 80 °C. At the end of the reaction time, the samples were removed from the reaction solution, rinsed in THF and methanol, and dried under a stream of N2(g). 2D Si (100) surfaces were alkylated following the procedure mentioned above, but with two minor differences:

**i.** the samples were chlorinated (i.e., immersed in the hot solution of PCl5) for 60 min

**ii.** the alkylation time for all samples was 24 h.

These conditions were found to give the maximal coverage of alkyl chain on 2D Si(100). Al‐ kylation of 2D Si(100) for 48 and 72 h gave similar coverage to those processed for 24 h.

**Figure 1.** Scheme for illustrating the functionalization of SiNWs through the two step chlorination/alkylation process.

### **3. Fabrication of silicon nanowires**

The SiNWs discussed in this chapter are grown by the well-known vapor-liquid-solid (VLS) mechanisms and a short description of the SiNWs growth will be given [28]. Si and Au form an eutectic that melts at 363°C while they individually melt at temperatures well above 1000°C. To grow SiNWs small gold droplets are dispensed on a Si wafer and inserted in a chemical vapor deposition (CVD) chamber. Then the samples heated up to 500°C i.e. above the eutectic temperature and below the melting points of each element. Si is supplied in va‐ por phase using silane (SiH4) as the precursor. Si atoms enter the Au droplets causing their supersaturation of Si and the subsequent deposition of excess Si at the droplet-substrate in‐ terface. The VLS growth mechanism has its name due to the different states of the Si during the process. Different doping of the SiNWs can be obtained by introducing doping gases during the growth. Additionally, by the right choice of nanoparticle size, SiNWs diameters ranging from a few hundred nanometers down to a few nanometers can be obtained. The parameters of the CVD processes that are used for the SiNWs discussed in this paper are: growth pressure of 0.5 to 2.0 mbar, heater temperature of between 600°C and 700°C (since the heater was not in direct contact with the sample later temperature was considerably low‐ er) and silane flow rate of 5 sccm using 5 to 10 sccm Ar as carrier gas. The grown SiNWs exhibit random orientation and usually gold nanoparticles on their tips (Figure 2).

These conditions were found to give the maximal coverage of alkyl chain on 2D Si(100). Al‐ kylation of 2D Si(100) for 48 and 72 h gave similar coverage to those processed for 24 h.

**Figure 1.** Scheme for illustrating the functionalization of SiNWs through the two step chlorination/alkylation process.

The SiNWs discussed in this chapter are grown by the well-known vapor-liquid-solid (VLS) mechanisms and a short description of the SiNWs growth will be given [28]. Si and Au form an eutectic that melts at 363°C while they individually melt at temperatures well above 1000°C. To grow SiNWs small gold droplets are dispensed on a Si wafer and inserted in a chemical vapor deposition (CVD) chamber. Then the samples heated up to 500°C i.e. above the eutectic temperature and below the melting points of each element. Si is supplied in va‐ por phase using silane (SiH4) as the precursor. Si atoms enter the Au droplets causing their supersaturation of Si and the subsequent deposition of excess Si at the droplet-substrate in‐ terface. The VLS growth mechanism has its name due to the different states of the Si during the process. Different doping of the SiNWs can be obtained by introducing doping gases during the growth. Additionally, by the right choice of nanoparticle size, SiNWs diameters ranging from a few hundred nanometers down to a few nanometers can be obtained. The parameters of the CVD processes that are used for the SiNWs discussed in this paper are: growth pressure of 0.5 to 2.0 mbar, heater temperature of between 600°C and 700°C (since the heater was not in direct contact with the sample later temperature was considerably low‐

**3. Fabrication of silicon nanowires**

180 Nanowires - Recent Advances

**Figure 2.** SEM image of SiNWs prepared by CVD on Si(111) substrates (upper part) and TEM images of an individual SiNW (lower part). The images indicate that the studied SiNWs consists of a 50±10 nm diameter Si core coated with 1-2nm of native SiOx shell.

The Au nanoparticle from the catalyst is very mobile and diffuses over the SiNW surface (and inside the bulk regime).

**Figure 3.** (a) HAADF-STEM general view of SiNWs, the red lines show the Au nanocluster extension. (b) HRTEM details of the selected area in Figure 3a.

This effect is enhanced in thin NWs and it was found from our previous report that Au nanoparticles can accelerate the oxidation on SiNWs [29]. 25-nm SiNWs show Au in the NW (in brighter contrast in Fig 3). Notice that Au is not only localized on the catalyst tip, but extends up to 300 nm apart from the tip. Figure 3a correspond to the HAADF-STEM general view of the NWs, the red lines show the Au nanocluster extension. Fig 3b corresponds to HRTEM details of the marked area in Fig 3a. In this case we have clearly observed that the presence of Au clusters (darker contrast), enhances oxidation rate in the thinner NWs rela‐ tive to the thicker ones.

## **4. Molecular functionalization of CVD-SiNWs**

Before any chemical treatment, SiNWs show a native oxide as illustrated previously in Fig‐ ure 2. The oxide layer can be detected by X-ray photoemission spectroscopy (XPS) of the ox‐ ide core level spectrum. The Si2p of the oxide peak appears in the 101-104 eV binding energy range. In order to attach molecules to the SiNW surface, the native oxide has to be removed. This can be done by shortly immersing the SiNWs in HF and NH4F solution. The XPS survey of the as-prepared H-SiNWs proved the thorough removal of oxides by the ab‐ sence of any peak in the 101-104 eV as can be clearly seen in Figure 4.

**Figure 4.** Si2p emission of SiO2-SiNW and H-SiNW.

It should be noted that the Si2p emission of SiO2-SiNW is shifted towards higher binding ener‐ gy as will be explained in the solar cell section. The emission of the C1s regime before and after termination is also compared in order to follow the covalent attachment of the molecules on the surfaces. Comparing the two emissions, an additional peak has to be introduced for moleculeterminated SiNW surfaces to obtain an appropriate deconvolution (figure 5). Figure 5 depicts the XPS spectrum of C1s emission region for the hydrogen- and alkyl-terminated SiNWs. H-SiNWs' spectrum was fitted to two C-C at 285.20±0.02 eV, and C-O at 286.69±0.02 eV while the alkyl-SiNW was fitted to three peaks: C-Si at 284.11±0.02 eV, C-C at 285.20±0.02 eV, and C-O at 286.69±0.02 eV. The new peak at C-Si at 284.11±0.02 eV confirm the chemical bonding of organ‐ ic molecules onto the SiNW surface. Peaks were typically adjusted to produce fits that mini‐ mized the difference between the full widths at half-maximum (fwhm).The center-to-center distances were fixed at 1.10±0.10 eV between the C-Si and the C-C emissions and at 2.60±0.10 eV between the C-O and the C-Si emissions. To attain the molecular coverage of surface atomic sites, the area under the C-Si peak is divided by the one under the Si2p peak (sum of Si2p1/2 and Si2p3/2) and normalized by the scan time. The coverage of the alkyl functionalities was com‐ pared with the highest value obtained for methyl (C1) functionality; in this instance "highest" means nearly full coverage of the Si atop sites, achieved after alkylation times of >24 h. This comparison is expressed throughout the text as "(C-Si/Si2p)alkyl/(C-Si/Si2p)max.methyl". Occasion‐ ally, a small signal due to oxygen was observed at 532.02±0.02 eV (O1s) and assigned to adven‐ titious adsorbed hydrocarbons having oxygen bonded to carbon (286.69±0.02 eV) as shown before and after alkyl treatment. It is reasonable to assume that these adventitious hydrocar‐ bons could stem from the wet chemical processing with THF solvent, THF/methanol rinse after functionalization, and/or carbonaceous materials present in the laboratory environment.

This effect is enhanced in thin NWs and it was found from our previous report that Au nanoparticles can accelerate the oxidation on SiNWs [29]. 25-nm SiNWs show Au in the NW (in brighter contrast in Fig 3). Notice that Au is not only localized on the catalyst tip, but extends up to 300 nm apart from the tip. Figure 3a correspond to the HAADF-STEM general view of the NWs, the red lines show the Au nanocluster extension. Fig 3b corresponds to HRTEM details of the marked area in Fig 3a. In this case we have clearly observed that the presence of Au clusters (darker contrast), enhances oxidation rate in the thinner NWs rela‐

Before any chemical treatment, SiNWs show a native oxide as illustrated previously in Fig‐ ure 2. The oxide layer can be detected by X-ray photoemission spectroscopy (XPS) of the ox‐ ide core level spectrum. The Si2p of the oxide peak appears in the 101-104 eV binding energy range. In order to attach molecules to the SiNW surface, the native oxide has to be removed. This can be done by shortly immersing the SiNWs in HF and NH4F solution. The XPS survey of the as-prepared H-SiNWs proved the thorough removal of oxides by the ab‐

It should be noted that the Si2p emission of SiO2-SiNW is shifted towards higher binding ener‐ gy as will be explained in the solar cell section. The emission of the C1s regime before and after termination is also compared in order to follow the covalent attachment of the molecules on the surfaces. Comparing the two emissions, an additional peak has to be introduced for molecule-

tive to the thicker ones.

182 Nanowires - Recent Advances

**4. Molecular functionalization of CVD-SiNWs**

**Figure 4.** Si2p emission of SiO2-SiNW and H-SiNW.

sence of any peak in the 101-104 eV as can be clearly seen in Figure 4.

**Figure 5.** XPS data from the C1s emission region of the hydrogen and alkyl-terminated SiNWs. (a) H-SiNWs show two peaks: C-C at 285.20±0.02 eV, and C-O at 286.69±0.02 eV. (b) Alky-SiNWs, show three peaks: C-Si at 284.11±0.02 eV, C-C at 285.20±0.02 eV, and C-O at 286.69±0.02 eV.

The absence of SiOx signal in the Si2p spectral surveys supports the argument that the O1s signal arises from physical bonding of hydrocarbons. Moreover, lack of F1s signal (which appears at 686.01±0.02 eV binding energy), confirmed that the NH4F(aq)-etched Si surface was not terminated by Si-F species.

Figure 6 shows the "(C-Si/Si2p)alkyl/(C-Si/Si2p)max.methyl" ratio as a function of alkylation time. As shown in the figure, the longer molecular chain requires a longer time for alkylation. For example, the time required to achieve 92±3 % of the saturation level in the adsorption curve, Γsat, for C1 functionality, stands at 20±2 min. Increasing the chain length to C6 increases Γsat to 120±10 min, while the maximum coverage that could be achieved is merely 56±10 % of C1- SiNWs. Longer alkylation times, up to 24 h, changed the total coverage by 5±3 %. Alkylation times of 24 to 72 hours did not further change the coverage on SiNWs and planar Si(100).

Table 1 summarizes the Γsat values for many of the molecules presented in our former stud‐ ies, along with the related (C-Si/Si2p)alkyl/(C-Si/Si2p)max.methyl ratios. As observed in the table, C2-SiNWs show a "(C-Si/Si2p)ethyl/(C-Si/Si2p)max.methyl" value of 68±3 %, indicating that C2 groups can be packed at a relatively high density without major steric effects. C3, C4, C5, C6, C8 and C10 produced 56±5 %, 49±5 %, 50±10 %, 56±10 %, 56 ±10 %, and 80±10 % cover‐ age, respectively, much higher than those of the same molecules on 2D Si (100) substrates (30±20 %,). Furthermore, the time required to achieve maximum coverage of the molecules on SiNWs is 4-30 times shorter than that of the 2D substrates. Using simple geometrical con‐ siderations, the gain in the steric hindrance on surfaces of 50±10 nm (in diameter) SiNWs is marginal. This cannot explain the higher molecular coverage on SiNWs, compared to 2D Si(100). Indeed, gain in steric hindrance between the adsorbed organic molecules starts to be important in NWs with <10-20 nm in diameter. This conclusion claims that the observed dif‐ ferences in coverage are related to different kinetics and higher available reactive sites on SiNWs, compared to 2D Si(100). The observation of 0.11±0.02 eV higher binding energy for a specific alkyl-terminated SiNWs, compared with equivalent 2D Si(100) surfaces, could fur‐ ther support the higher reactivity of SiNW atop sites. The broadening of Si2p fwhm in the SiNWs case suggests, compared to 2D Si(100), that wider distribution of bond strengths exist on the surface, thus leading to a wider distribution of reactive sites [30].

**Figure 6.** Coverage of alkyl molecules of SiNW atop sites as a function of alkylation time. Γsat stands for the time re‐ quired to achieve92± 3 % of the saturation level of the adsorption curve.


**Table 1.** Summary of the XPS results for C1-C6 alkyl chains bonded to SiNWs and 2D Si(100) surfaces via Si-C bond

#### **4.1. Early stages of oxide growth in H-SiNWs**

Γsat, for C1 functionality, stands at 20±2 min. Increasing the chain length to C6 increases Γsat to 120±10 min, while the maximum coverage that could be achieved is merely 56±10 % of C1- SiNWs. Longer alkylation times, up to 24 h, changed the total coverage by 5±3 %. Alkylation times of 24 to 72 hours did not further change the coverage on SiNWs and planar Si(100).

184 Nanowires - Recent Advances

Table 1 summarizes the Γsat values for many of the molecules presented in our former stud‐ ies, along with the related (C-Si/Si2p)alkyl/(C-Si/Si2p)max.methyl ratios. As observed in the table, C2-SiNWs show a "(C-Si/Si2p)ethyl/(C-Si/Si2p)max.methyl" value of 68±3 %, indicating that C2 groups can be packed at a relatively high density without major steric effects. C3, C4, C5, C6, C8 and C10 produced 56±5 %, 49±5 %, 50±10 %, 56±10 %, 56 ±10 %, and 80±10 % cover‐ age, respectively, much higher than those of the same molecules on 2D Si (100) substrates (30±20 %,). Furthermore, the time required to achieve maximum coverage of the molecules on SiNWs is 4-30 times shorter than that of the 2D substrates. Using simple geometrical con‐ siderations, the gain in the steric hindrance on surfaces of 50±10 nm (in diameter) SiNWs is marginal. This cannot explain the higher molecular coverage on SiNWs, compared to 2D Si(100). Indeed, gain in steric hindrance between the adsorbed organic molecules starts to be important in NWs with <10-20 nm in diameter. This conclusion claims that the observed dif‐ ferences in coverage are related to different kinetics and higher available reactive sites on SiNWs, compared to 2D Si(100). The observation of 0.11±0.02 eV higher binding energy for a specific alkyl-terminated SiNWs, compared with equivalent 2D Si(100) surfaces, could fur‐ ther support the higher reactivity of SiNW atop sites. The broadening of Si2p fwhm in the SiNWs case suggests, compared to 2D Si(100), that wider distribution of bond strengths exist

**Figure 6.** Coverage of alkyl molecules of SiNW atop sites as a function of alkylation time. Γsat stands for the time re‐

quired to achieve92± 3 % of the saturation level of the adsorption curve.

on the surface, thus leading to a wider distribution of reactive sites [30].

To examine the thermal stability of H-SiNWs, the native oxide shell on SiNWs was etched away by hydrofluoric acid and ammonium fluoride. Subsequently, freshly etched SiNWs were an‐ nealed in ambient conditions (with average humidity of 30 %) in seven distinct temperatures of 50 ˚C, 75 ˚C, 150 ˚C, 200 ˚C, 300 ˚C, 400 ˚C and 500 ˚C, each for five different time-spans: 5 min, 10 min, 20 min, 30 min and 60 min. After each heating stage, the specimens were examined by XPS. To identify various sub-oxide states in the Si2p spectrum, spectral decomposition was per‐ formed by a curve-fitting procedure, using Gaussian-Lorentzian functions after a Shirley back‐ ground subtraction as shown as example in Figure 7. The sub-oxides states and their respective chemical shifts (Δ) relative to the Si2p3/2 (at 99.60± 0.02 eV) are as follows: Si2p1/2 (Δ=0.60 eV), Si2O (Δ=0.97 eV), SiO (Δ=1.77 eV), Si2O3 (Δ=2.50 eV), and SiO2 (Δ=3.87 eV). The shift in binding energy is dependent on temperature and time (dashed line in Figure 7a). Generally speaking, upon the formation of sub-oxides and SiO2, the Si2p peak moves to higher binding energies to compensate the band bending and surface Fermi level shift as the oxide layer grows (not shown). The intensity of a sub-oxide (*Iox*) is expressed by the ratio of the integrated area under its peak to the Si2p peaks overall area (Si2p3/2+Si2p1/2) i.e. the total intensity of unoxidized sili‐ con atoms. Sub-oxide intensity can be expressed in terms of the number of monolayers formed corresponding to the intensity of 0.21 per monolayer [31].

Figure 8 depicts the variation of the average number of monolayers over time for individual sub-oxides (*Iox*) and the overall oxide intensity (*Itot=ISi2O+ ISiO+ISi2O3+ISiO2*) for low-temperature (T < 200 ˚C, Figure 8a) and high-temperature (T ≥200 ˚C, Figure 8b) oxidation of H-SiNWs. Comparison of the sub-oxides growth in Figure 8 illustrates their temperature and time de‐ pendence. In both temperature ranges, Si2O is present even in the etched surfaces with in‐ tensity of *Iox*=0.02±0.01 i.e. ~0.2 monolayer and its intensity plateaus at the value of *Iox*=0.06±0.01 (~0.25 to 0.3 monolayer) at the very early stages exhibiting negligible growth over the whole time range. SiO and Si2O3 exhibit approximately identical behavior and val‐ ues in both ranges despite the plateau region occurs 15 minutes earlier in the high tempera‐ ture range. They exhibit no peak at t=0 and an intensity of 30-60 % of the maximum intensity after 5 min and 20 min for high and low temperature regime, respectively. However, at T ≥ 200 ˚C, SiO and Si2O3 have 0.2 monolayer coverage that is twice as large as coverage in T < 200 ˚C (~0.1 monolayer). The behavior of the transient sub-oxides versus oxidation time is similar to that of SiNWs exposed to molecular oxygen wherein the intensities of the transi‐ ent sub-oxide almost remain constantly below one monolayer. Conversely, SiO<sup>2</sup> intensity was found to strongly rely on the temperature and time and two regimes of fast and slow growth were clearly recognized.

**Figure 7.** (a) An example of Si2p XPS spectra of the H-SiNW surfaces annealed at 500 ºC for different times; the in‐ clined dashed line represents the blue shift upon formation of the oxide layer. (b) Decomposed Si2p sample spectrum of oxidized SiNWs illustrating four different suboxides including Si2O, SiO, Si2O3 plus the stoichiomertic SiO2 at 100.6 eV, 101.7 eV 102.5 eV and 103.4 eV, respectively.

**Figure 8.** Number of monolayers over time for the sub-oxides Si2O (green), SiO (gray), Si2O3 (blue), SiO2 (red) and their sum as total oxide (black) in the two low-temperature (a) and high-temperature (b) regimes. The vertical dashed lines represent Γsat, while the horizontal one in (b) defines the onset for ordered oxide layer formation.

At higher temperatures (T≥200 °C), the time required to achieve 70±10 % of the final in‐ tensity (defined as Γsat) for SiO2 is 10±2min, while at lower temperatures (T<200 °C), Γsat shifts to 18±2min. On the other hand, Γsat for the total oxide amount lies approximately on the same times as SiO2 demonstrating its sizable weight in the total oxide amount. Therefore, temperature dependence of oxide growth can be well observed in Figure 9a that depicts the total oxide vs. time for all tested temperatures. According to the TEM measurements the extension of the annealing period at 500 °C from 1 hour to 24 hours results in an increase of oxide layer thickness from 4.5nm to 6nm. Thus, 23 hours of fur‐ ther annealing leads to less than 35 % increase in oxide thickness that is noticeably smaller than the growth observed at the starting times and the graph after Γsat can be assumed as a plateau like sub-oxides plots.

ture range. They exhibit no peak at t=0 and an intensity of 30-60 % of the maximum intensity after 5 min and 20 min for high and low temperature regime, respectively. However, at T ≥ 200 ˚C, SiO and Si2O3 have 0.2 monolayer coverage that is twice as large as coverage in T < 200 ˚C (~0.1 monolayer). The behavior of the transient sub-oxides versus oxidation time is similar to that of SiNWs exposed to molecular oxygen wherein the intensities of the transi‐ ent sub-oxide almost remain constantly below one monolayer. Conversely, SiO<sup>2</sup> intensity was found to strongly rely on the temperature and time and two regimes of fast and slow

**Figure 7.** (a) An example of Si2p XPS spectra of the H-SiNW surfaces annealed at 500 ºC for different times; the in‐ clined dashed line represents the blue shift upon formation of the oxide layer. (b) Decomposed Si2p sample spectrum of oxidized SiNWs illustrating four different suboxides including Si2O, SiO, Si2O3 plus the stoichiomertic SiO2 at 100.6

**Figure 8.** Number of monolayers over time for the sub-oxides Si2O (green), SiO (gray), Si2O3 (blue), SiO2 (red) and their sum as total oxide (black) in the two low-temperature (a) and high-temperature (b) regimes. The vertical dashed lines

represent Γsat, while the horizontal one in (b) defines the onset for ordered oxide layer formation.

growth were clearly recognized.

186 Nanowires - Recent Advances

eV, 101.7 eV 102.5 eV and 103.4 eV, respectively.

Early stages of high temperature oxidation in H-SiNWs can be modeled by the modified Deal-Grove model taking into account partly or completely sub-oxides as growth sites with varying concentrations over time and temperature [32, 33]. Nonetheless, above Γsat a linear Deal-Grove regime cannot be accommodated with the approximately flat tail of the SiO2 plot and a limiting mechanism is required to interpret such a behavior [34]. Between Γsat and maximum oxidation (1 hour) at high temperatures, a significant fraction of the superficial Si atoms are surrounded by fourfold occupied oxygen sites with high local strains that act as a barrier for further oxidation. Under this conditions, oxidation hindrance can be caused by two phenomena: self-limiting oxide film formation (seen in 2D Si) [32] and self-retarding ox‐ idation known for SiNWs and their particular geometry [35]. However, self-retarded oxida‐ tion becomes more significant on curved surfaces like SiNWs at diameters below 35 to 44 nm due to the extremely small amount of viscous flow in the oxide layer that could relax the applied stresses [36]. On the other hand, repetition of the same measurements for Si2D re‐ sulted in the same flat end for the Iox-t plot at T ≥ 200 °C (Figure 9a and 9b). Therefore, the inhibited oxide growth in the high temperature region is better understood as a result of self-limited oxidation rather than geometrical compressive that can be neglected for the cur‐ rent NWs. Non-retarded growth of oxide at 500 ˚C (Figure 9a) can be associated with partial melting at NW surfaces causing accelerated oxygen diffusion and NWs fragmentation (6 times length decrease) which led to deeper oxygen penetration and larger interfacial areas (Figure 9b and c).

At low temperatures Γsat of SiO2 is shifted to longer times (18±2min) and transient suboxides are the main sources for the total oxide intensity at times between 0 and Γsat (Figure 8b). Based on our earlier work, heating above 450 ºC can remove all the Si-H bonds on flat Si wafers [37]. At lower temperatures and initial times (i.e. before <Γsat), the oxidation occurs in the presence of functionalizing H atoms and OH groups (result‐ ed from water vapor interaction with the surfaces). Therefore, oxidation proceeds thor‐ ough Si-Si backbond oxidation forming sequentially peroxidic and disiloxanic bonds in the first few atomic layers. Nevertheless, heating in the atmosphere of 30 % average hu‐ midity, water assistance to early oxide growth and corresponding oxidation-hydroxyla‐ tion cannot be neglected [38, 39]. To understand the relation between the ratedetermining step and the surface bonds strength, we studied methyl-termination of SiNWs, which naturally provides the same coverage as hydrogen functionalities but with Si-C bonds, instead [19, 30]. Interestingly, when exposed to air at room tempera‐ ture for 44 days, SiO2 formation was relatively suppressed, while Si2O intensity was equal to the one for H-SiNW [8, 40]. After the 44-day period, methyl-terminated SiNWs showed ISiO2 intensity of 0.11; it almost equals that of H-SiNWs after 1 hour. This can be explained by the lower susceptibility of Si-Si underlying bonds in Si-C covered surfaces compared to Si-H covered ones [30]. For further oxidation at lower temperatures howev‐ er, the electron or hole tunneling mechanism proposed for Si(100) room temperature ox‐ idation encounters difficulties to be proven considering the fact that neither individual sub-oxides nor total oxide amount reaches the level of a single monolayer [41]. Alterna‐ tively, rupture of the surface bonds may act as the second oxidation source after local development of the backbond Si-O. In this respect, in methyl-terminated SiNWs mono‐ tonic growth of sub-oxides and SiO2 has been observed after a dramatic decline in the amount of Si-C concentration. Similarly, bond propagation can be conceived in H-SiNWs with a relatively lower resistance to the oxidizing species [30].

Figure 9d shows the temperature dependence of the total oxide formation rate in the 2D Si and H-SiNWs specimens before Γsat for all the temperatures investigated in this study. High and low temperature data points are fitted in two different lines for each specimen. Near-linear behavior of oxidation rate logarithm vs. reciprocal of temperature at the early stages of oxidation allows us to calculate respective activation energies (EA ox) based on the Arrhenius equation. In the high temperature region 2D Si and SiNWs exhibit EA ox of 46.35 m eV and 48.22 m eV, respectively. Such a minor difference in the activation energy magnitudes denotes again that geometrically-induced stresses affect negligibly the rate-determining step. Large divergence from values of activation energies for the linear Deal-Grove model (1.83 to 2.0 eV) approves that the rate is rath‐ er determined by the growth sites development than by surface oxidation reaction. In addition to sub-oxide islands, near-interfacial remnant silicon atoms may contribute to the oxidation process as growth sites [33, 42]. Moreover, higher absolute intensities and oxidation rate values for SiNWs at 500 ˚C beyond the first 10 minutes compared to 2D Si necessitate the existence of a growth-site-enhancing mechanism as partial melting of SiNWs. This mechanism improves the oxygen permeability to deeper layers and increas‐ es the interfaces due to NW fragment formation (Figure 9c).

For the low temperature regime EA ox values of 35.20 m eV for 2D Si and 23.31 m eV for SiNWs were also calculated. Lower activation energies for SiNWs may correspond to the strain induced by dangling bonds rebonding at the (110) and (111) adjoining surfaces as well as higher ratio of ambient exposure area to the oxide/silicon interface area [43]. Stretching weakens the very first layers of silicon backbonds and results in their higher susceptibility to bond rupture and suboxide formation. Conversely, bending stresses are able to increase ro‐ bustness of hydride terminations (di- and trihydrides) on the surface and lower the SiNW oxidation rate at higher temperatures where surface bond propagation determines the over‐ all rate (for 75 °C and 150 °C where rates are lower than 2D Si).

All the above descriptions suggest the following mechanism (schematically shown in Figure 10): (i) At low temperatures, Si-Si backbond oxidation and later Si-H bond prop‐ agation are the rate-determining steps. (ii) At higher temperatures, the oxygen diffusion is considered as the rate-determining step initially through controlling the growth sites concentration (conserving some Si-H interfacial bonds) where disorder monolayer still dominates the oxide intensity. Beyond Γsat, the reaction shifts to almost zero-order which denotes a self-limiting process because initially formed oxide shells prevent a further ox‐ idation. A fraction of the Si-H bonds converted to Si-OH according to H2O molecules present in the surrounding atmosphere.

SiNWs, which naturally provides the same coverage as hydrogen functionalities but with Si-C bonds, instead [19, 30]. Interestingly, when exposed to air at room tempera‐ ture for 44 days, SiO2 formation was relatively suppressed, while Si2O intensity was equal to the one for H-SiNW [8, 40]. After the 44-day period, methyl-terminated SiNWs showed ISiO2 intensity of 0.11; it almost equals that of H-SiNWs after 1 hour. This can be explained by the lower susceptibility of Si-Si underlying bonds in Si-C covered surfaces compared to Si-H covered ones [30]. For further oxidation at lower temperatures howev‐ er, the electron or hole tunneling mechanism proposed for Si(100) room temperature ox‐ idation encounters difficulties to be proven considering the fact that neither individual sub-oxides nor total oxide amount reaches the level of a single monolayer [41]. Alterna‐ tively, rupture of the surface bonds may act as the second oxidation source after local development of the backbond Si-O. In this respect, in methyl-terminated SiNWs mono‐ tonic growth of sub-oxides and SiO2 has been observed after a dramatic decline in the amount of Si-C concentration. Similarly, bond propagation can be conceived in H-

Figure 9d shows the temperature dependence of the total oxide formation rate in the 2D Si and H-SiNWs specimens before Γsat for all the temperatures investigated in this study. High and low temperature data points are fitted in two different lines for each specimen. Near-linear behavior of oxidation rate logarithm vs. reciprocal of temperature at the early stages of oxidation allows us to calculate respective activation energies

ox) based on the Arrhenius equation. In the high temperature region 2D Si and

in the activation energy magnitudes denotes again that geometrically-induced stresses affect negligibly the rate-determining step. Large divergence from values of activation energies for the linear Deal-Grove model (1.83 to 2.0 eV) approves that the rate is rath‐ er determined by the growth sites development than by surface oxidation reaction. In addition to sub-oxide islands, near-interfacial remnant silicon atoms may contribute to the oxidation process as growth sites [33, 42]. Moreover, higher absolute intensities and oxidation rate values for SiNWs at 500 ˚C beyond the first 10 minutes compared to 2D Si necessitate the existence of a growth-site-enhancing mechanism as partial melting of SiNWs. This mechanism improves the oxygen permeability to deeper layers and increas‐

SiNWs were also calculated. Lower activation energies for SiNWs may correspond to the strain induced by dangling bonds rebonding at the (110) and (111) adjoining surfaces as well as higher ratio of ambient exposure area to the oxide/silicon interface area [43]. Stretching weakens the very first layers of silicon backbonds and results in their higher susceptibility to bond rupture and suboxide formation. Conversely, bending stresses are able to increase ro‐ bustness of hydride terminations (di- and trihydrides) on the surface and lower the SiNW oxidation rate at higher temperatures where surface bond propagation determines the over‐

ox of 46.35 m eV and 48.22 m eV, respectively. Such a minor difference

ox values of 35.20 m eV for 2D Si and 23.31 m eV for

SiNWs with a relatively lower resistance to the oxidizing species [30].

es the interfaces due to NW fragment formation (Figure 9c).

all rate (for 75 °C and 150 °C where rates are lower than 2D Si).

For the low temperature regime EA

(EA

SiNWs exhibit EA

188 Nanowires - Recent Advances

**Figure 9.** Oxide monolayer growth over time for H-SiNW surfaces (a) and 2D H-Si (b) oxidized at different tempera‐ tures for time spans ranging from 5 to 60 min. (c) Comparison between pristine (right) and melted (left) SiNWs. (d) Temperature dependence of oxide formation in H-SiNWs and 2D Si in air between 50˚C and 500˚C. For all tempera‐ tures, the annealing time is below Γsat.

**Figure 10.** Schematic graph summarizing the inferred oxidation mechanism for H-SiNWs. The blue spots represent the sub-oxide islands. Interfaces at different stages of low-temperature oxidation as well as initial H-terminated one sare depicted in more detail in circles.

#### **4.2. Stability of molecules on SiNWs in long time**

To investigate the stability of the alkylated SiNWs, freshly prepared SiNWs having maxi‐ mum coverage of alkyl functionalities (mostly obtained after alkylation time of 24 h) were exposed to ambient air for several hundreds of hours (Table 2). Oxide intensity in SiNWs functionalized by methyl and ethyl equals zero up to 48 hours. After 336 h of exposure, ox‐ ide intensity reached 0.03 (∼0.14 monolayer) for C1-SiNWs and 0.05 (∼0.23monolayer of ox‐ ide) for C2-SiNWs. In contrary, after 24 hours of exposure, oxide intensity in SiNWs terminated with C3 to C6 rose to 0.01 (∼0.05 monolayer of oxide), 0.02 (∼0.10 monolayer of oxide), 0.02 (∼0.10 monolayer of oxide), and 0.01 (∼0.05monolayer of oxide), respectively. This implies that SiNWs terminated with C3-C6 alkyl molecules exhibit less stability against oxidation compared to C1- and C2-SiNWs. Nevertheless, after 336 hours, oxide intensity of C3- to C6-terminated SiNWs have approximately the same magnitudes (∼ 0.13). From the data in Tables 1 and 2, with a wider perspective, it can be deduced that the shorter the alkyl chain, the lower the SiOx/Si2p ratio and the less liable to oxidation the terminated SiNWs.

These observations can be explained by the fact that the shorter the molecular chains, the lower the coverage of the alkyl molecules and the higher the probability of interaction be‐ tween oxidizing agents (O2 and H2O) and molecule-free sites (or pinholes).The oxidation de‐ gree of the SiNWs was compared to planar Si(100) surfaces alkylated with the same series of molecules. C1-SiNWs exposed to air over a period of 336 h illustrated ca.4-fold lower oxida‐ tion intensity than the equivalent Si(100) surfaces, despite alike initial coverage levels. Simi‐ larly, C2-SiNWs exhibited ca. 3-folds lower oxide intensity than C2-Si(100). In contrast, (C3- C6)-SiNWs oxide amount was comparable to those of planar (C3-C6)-Si(100) after 336 h. These observations could be attributed to stronger Si-C bond on SiNWs surfaces. This can be supported by the shift in Si-C energy bond of C1 terminations from SiNW (284.22±0.02 eV) to planar Si(100) (284.11±0.02 eV). Stronger bonding of surface functionalities has also been observed in H-terminated SiNWs. Respective studies suggest that the robustness of the SiNW's surface hydride (or alkyls) is a consequence of bending stresses being developed at the edge of two adjoining facets of a SiNW.


**Table 2.** Summary of the Oxidation intensity (SiOx/Si2P Ratio) for Alkylated SiNWs and 2D Si(100) Surfaces, Having Maximum Coverage (Mostly, Obtained after 24 hrs of Alkylation Time), at Different Exposure Times to Ambient Air.

#### **4.3. Stability of Si-C bond on SiNWs**

**Figure 10.** Schematic graph summarizing the inferred oxidation mechanism for H-SiNWs. The blue spots represent the sub-oxide islands. Interfaces at different stages of low-temperature oxidation as well as initial H-terminated one sare

To investigate the stability of the alkylated SiNWs, freshly prepared SiNWs having maxi‐ mum coverage of alkyl functionalities (mostly obtained after alkylation time of 24 h) were exposed to ambient air for several hundreds of hours (Table 2). Oxide intensity in SiNWs functionalized by methyl and ethyl equals zero up to 48 hours. After 336 h of exposure, ox‐ ide intensity reached 0.03 (∼0.14 monolayer) for C1-SiNWs and 0.05 (∼0.23monolayer of ox‐ ide) for C2-SiNWs. In contrary, after 24 hours of exposure, oxide intensity in SiNWs terminated with C3 to C6 rose to 0.01 (∼0.05 monolayer of oxide), 0.02 (∼0.10 monolayer of oxide), 0.02 (∼0.10 monolayer of oxide), and 0.01 (∼0.05monolayer of oxide), respectively. This implies that SiNWs terminated with C3-C6 alkyl molecules exhibit less stability against oxidation compared to C1- and C2-SiNWs. Nevertheless, after 336 hours, oxide intensity of C3- to C6-terminated SiNWs have approximately the same magnitudes (∼ 0.13). From the data in Tables 1 and 2, with a wider perspective, it can be deduced that the shorter the alkyl chain, the lower the SiOx/Si2p ratio and the less liable to oxidation the terminated SiNWs.

These observations can be explained by the fact that the shorter the molecular chains, the lower the coverage of the alkyl molecules and the higher the probability of interaction be‐ tween oxidizing agents (O2 and H2O) and molecule-free sites (or pinholes).The oxidation de‐ gree of the SiNWs was compared to planar Si(100) surfaces alkylated with the same series of molecules. C1-SiNWs exposed to air over a period of 336 h illustrated ca.4-fold lower oxida‐ tion intensity than the equivalent Si(100) surfaces, despite alike initial coverage levels. Simi‐ larly, C2-SiNWs exhibited ca. 3-folds lower oxide intensity than C2-Si(100). In contrast, (C3- C6)-SiNWs oxide amount was comparable to those of planar (C3-C6)-Si(100) after 336 h. These observations could be attributed to stronger Si-C bond on SiNWs surfaces. This can be

depicted in more detail in circles.

190 Nanowires - Recent Advances

**4.2. Stability of molecules on SiNWs in long time**

SiNW surfaces were terminated by CH3 and compared with equivalent planar CH3–Si surfa‐ ces having same surface coverage with Si–C bonds. Figure 11 shows the coverage of methyl functionalities as a function of alkylation time. Comparing the alkylation kinetics on SiNWs with 2D Si(100) showed two main notable differences: (I) 92±3 % of the maximum coverage on 2D Si(100) was achieved after 1000±50 min alkylation; (II) the reaction kinetics on 2D Si(100) at short and long alkylation times are rather different than on SiNWs. These differen‐ ces in the kinetic behavior could be attributed mainly to differences in the surface energy. The cohesive energy becomes higher upon decreasing the dimensions of a given structure. Based on earlier work it can be seen that 50±10 nm (in diameter) SiNW surfaces has higher cohesive energy (~4.4 eV) than 2D Si(100) surfaces (~3.4 eV)[30, 44]. Furthermore, the broad‐ ening of Si2p fwhms in the case of SiNWs (data not shown), as compared to 2D Si(100) and 2D Si(111), suggests that wider distribution of Si–C bond strengths and/or ligands angle var‐ iations exists on the surface. Fig. 12 depicts variation of the SiO2 intensity and molecular coverage over exposure time for both NWs and planar specimens of silicon. Since determin‐ ing surface coverage requires accounting for escape length of photoelectrons, an ''impartial'' indicator was required to examine the robustness of the data presented in Fig. 12. With this in mind, ToF-SIMS experiments, by which the methyl coverage is proportional to the SiCH3 + (or CH3 - ) fragments, can be carried out. ToF-SIMS results showed similar concentrations (within 5 % experimental error) of characteristic fragments in both CH3–SiNWs and 2D CH3–Si(100) substrates, marking the reliability of the XPS data presented in Fig. 11 and 12.

**Figure 11.** Coverage of methyl functionalities on 2D and SiNW as a function of alkylation time.

As shown in Fig. 12a, CH3–SiNWs exposed to air over one month showed ca. 3-fold less oxi‐ dation than the equivalent 2D CH3–Si(100) surfaces, albeit similar molecular coverage's in both structures. To attain a wider perspective, we made a comparison between 2D CH3– Si(111) surfaces (naturally of 15–20 % coverage)and 2D CH3–Si(100).

**Figure 12.** (a) Ratio of the oxidized Si2p peak area to the bulk Si2p peak area for the methyl modification of SiNWs, 2D Si(100), and 2D Si(111), exposed to air over extended time periods. (b) Si–C decay as a function of exposure time for SiNW and 2D Si(100). The measurements were done three times for each sample and averages have been taken.

Interestingly, in spite of notably lower coverage, 2D CH3–Si(111) surfaces exhibit higher sta‐ bility against oxidation than the equivalent 2D Si(100) surfaces, while they showed ca. 2-fold higher oxide intensity than the CH3–SiNWs. This can also be referred to surface bond ener‐ gy. The observation of ~0.11 ±0.02 eV higher binding energy for a CH3–SiNW, compared to equivalent 2D CH3–Si(100) surfaces, could further be ascribed to the higher reactivity of atop sites. Indeed, the CH3–SiNW's C1s spectra showed 22 % decrease in the Si–C bond signal throughout 20 days, at the time which 2D CH3–Si(100) specimen showed ~34 % decrease of the same signal (Fig. 12b). Surface bond strength dependency on structural geometry can be backed by the reports of Lee and co-workers on stronger H-termination of SiNWs relative to 2D Si surfaces suggested that the robustness of the SiNWs hydride is a consequence of bend‐ ing stresses, where re-bonding of dangling bonds at the edge of two adjoining facets of a SiNW takes place [45].

(or CH3 -

192 Nanowires - Recent Advances

) fragments, can be carried out. ToF-SIMS results showed similar concentrations

(within 5 % experimental error) of characteristic fragments in both CH3–SiNWs and 2D CH3–Si(100) substrates, marking the reliability of the XPS data presented in Fig. 11 and 12.

**Figure 11.** Coverage of methyl functionalities on 2D and SiNW as a function of alkylation time.

Si(111) surfaces (naturally of 15–20 % coverage)and 2D CH3–Si(100).

As shown in Fig. 12a, CH3–SiNWs exposed to air over one month showed ca. 3-fold less oxi‐ dation than the equivalent 2D CH3–Si(100) surfaces, albeit similar molecular coverage's in both structures. To attain a wider perspective, we made a comparison between 2D CH3–

**Figure 12.** (a) Ratio of the oxidized Si2p peak area to the bulk Si2p peak area for the methyl modification of SiNWs, 2D Si(100), and 2D Si(111), exposed to air over extended time periods. (b) Si–C decay as a function of exposure time for SiNW and 2D Si(100). The measurements were done three times for each sample and averages have been taken.


**Table 3.** SiCxHy + peak intensities of ToF-SIMS spectra of SiNWs before and after 4 days exposure to an oxidizing atmosphere

To understand the relation between strength of the Si–C bond and the oxidation resistance, ToF-SIMS fragments of CH3–SiNWs and 2D CH3–Si(100) surfaces were collected after differ‐ ent exposure periods to the oxidizing agents. Table 3presents the selected ToF-SIMS spec‐ trum peak intensities of CH3–SiNW before and after 4-day exposure to pure O2(20 %)–N2(80 %) environment of 10–15 % relative humidity (RH). After 4 days of oxidation, SixO signal (where x = 1,2) of SiNW samples increased by 70–95±10 % while the concentration of Si–C and Si–CH3 bonds decreased by 36±2 % and 15± 2 %, respectively. Higher changes in the SixO signal suggest higher molecule-free Si–Si backbonds susceptibility to bond rupture upon interaction with oxidant molecules, than the Si–C bonds. And higher changes in the Si–C (36±2 %) signal than Si–CH3(15±2 %) indicates that Si-Si bonds adjacent to the CH3 groups are less prone to oxidation than Si-C surface bonds. Moreover, the results point out the early breakage of Si–Si back-bonds to form Si–O–Si back-bonds. After oxidation of the entire Si–Si back-bonds, oxidizing species start attacking the Si–C bonds. For 2D CH3–Si(100) surfaces, the SixO signal increased by 86–120±10 % while the concentration of Si–C and Si– CH3 bonds decreased by 47±2 % and 21±2 %, respectively, after 4 days of exposure to the oxidizing atmosphere. Evaluating the relative changes obtained for 2D CH3–Si(100) surfaces and CH3–SiNWs calls for two main conclusions: (1) CH3–SiNWs are more stable than 2D CH3–Si(100) surfaces, in consistency with the XPS observation, (2) The oxidation mechanism of 50±10 nm (in diameter) CH3–SiNWs is similar to that of 2D CH3–Si(100) surfaces.

#### **4.4. Effect of chain length of alkyl molecules**

According to theoretical simulations, steric effects (van der Waals diameter) can hinder for‐ mation of dense packing of alkyls longer than C1. Therefore, increasing length of the alkyl chains increases the van der Waals diameter from 2.5 Å (in the case of C1) to more than 4.5-5.0 Å for longer alkyl chains. The latter van der Waals diameters are much larger than the internulecular distance between adjacent Si atoms (3.8 Å), a matter that decreases their adsorption rate and limits their coverage to maximum 50-55 % of a monolayer of Si surface sites. With these findings in mind, we found the chain length dependence in Figure 13 to be consistent with the similar decay of the coverage reported by others on flat Si(111) surfaces for longer chains than in the present work, with, no increase of coverage above 7 carbon atoms [46]. Based on these inconsistencies it can be inferred that the higher coverage ob‐ tained for >C6 chain lengths is artificial and/or not significant owing to the problems in fit‐ ting of the XPS data. In contrast, few experimental observations suggest that the coverage behavior of chains longer than C6 on SiNWs might not be artificial, as justified by the fol‐ lowing explanations: (1) ToF SIMS experiments (which eliminate potential artificial observa‐ tions) have shown higher absolute coverage long chain lengths, in good consistency with XPS observations. (2) In the case of 2D Si(100) surface, the results of ToF-SIMS have shown (more or less) the expected coverage versus chain length behavior; at the time, similar be‐ havior was not observed in the case of SiNWs. (3) The experimental error and/or accuracy of the peak fitting are relatively small (4-11 %).

**Figure 13.** Max-alkyl versus alkyl chain length on SiNWs and, for comparison, on 2D Si(100) surfaces.

It is reasonable to consider passivation of SiNWs by alkyl molecules as a function of two main factors: (I) molecule-molecule lateral interaction and (II) molecule-substrate vertical in‐ teraction. For short alkyl chains (C1-C5), which exhibit liquid-like behavior and thermal fluctuations [47] the determining factor is the vertical interaction. Increasing the chain length to C6-C10 forms a solid-like phase, where the lateral interactions become more dominant in the consequent coverage of the SiNW surfaces. It is worth pointing out that the lateral inter‐ actions between long alkyl chains might be formed during the physisorption stage, before the covalent bindings between the carbon and silicon atoms (or chemisorption) arise [48].

## **5. Integration of hybrid SiNWs into technological devices**

Hybrid SiNWs modified by covalent Si-C bonds showed excellent atmospheric stability, high tansconductance values, low surface-defect densities, and allowing formation of techni‐ cally-feasible air-stable SiNW field-effect transistors (FETs) and SiNW-photovoltaics [9, 19]. Here we demonstrate the electrical characteristics of such SiNW-based devices molecularly modified by the two-step chlorination/alkylation method. Analyses suggest that extrinsic ef‐ fects (i.e. surface states and surface charge) accompanied by the adsorption of molecular lay‐ ers significantly affect the FET and solar cell characteristics.

#### **5.1. Fabrication of the FET**

**4.4. Effect of chain length of alkyl molecules**

194 Nanowires - Recent Advances

the peak fitting are relatively small (4-11 %).

**Figure 13.** Max-alkyl versus alkyl chain length on SiNWs and, for comparison, on 2D Si(100) surfaces.

It is reasonable to consider passivation of SiNWs by alkyl molecules as a function of two main factors: (I) molecule-molecule lateral interaction and (II) molecule-substrate vertical in‐ teraction. For short alkyl chains (C1-C5), which exhibit liquid-like behavior and thermal fluctuations [47] the determining factor is the vertical interaction. Increasing the chain length

According to theoretical simulations, steric effects (van der Waals diameter) can hinder for‐ mation of dense packing of alkyls longer than C1. Therefore, increasing length of the alkyl chains increases the van der Waals diameter from 2.5 Å (in the case of C1) to more than 4.5-5.0 Å for longer alkyl chains. The latter van der Waals diameters are much larger than the internulecular distance between adjacent Si atoms (3.8 Å), a matter that decreases their adsorption rate and limits their coverage to maximum 50-55 % of a monolayer of Si surface sites. With these findings in mind, we found the chain length dependence in Figure 13 to be consistent with the similar decay of the coverage reported by others on flat Si(111) surfaces for longer chains than in the present work, with, no increase of coverage above 7 carbon atoms [46]. Based on these inconsistencies it can be inferred that the higher coverage ob‐ tained for >C6 chain lengths is artificial and/or not significant owing to the problems in fit‐ ting of the XPS data. In contrast, few experimental observations suggest that the coverage behavior of chains longer than C6 on SiNWs might not be artificial, as justified by the fol‐ lowing explanations: (1) ToF SIMS experiments (which eliminate potential artificial observa‐ tions) have shown higher absolute coverage long chain lengths, in good consistency with XPS observations. (2) In the case of 2D Si(100) surface, the results of ToF-SIMS have shown (more or less) the expected coverage versus chain length behavior; at the time, similar be‐ havior was not observed in the case of SiNWs. (3) The experimental error and/or accuracy of

SiNWs were manufactured through a top-down approach based on electron beam lithogra‐ phy rather than the bottom-up VLS growth. It initiates from a bonded SOI wafer, with a thin (100 nm) top Si (100) layer (resistivity of 10–20 Ωcm) insulated from the (100) silicon sub‐ strate (resistivity of 0.8–1.2 Ωcm) by a silicon dioxide layer (380 nm). Surface was covered by a bilayer PMMA and mask definition was then performed by high-resolution e-beam lithog‐ raphy. The resist was then developed in a solution of MiBK:IPA (1:3). The pattern was trans‐ ferred from the PMMA to the top SiO2 layer by buffered HF etching. The central region, where the silicon was defined, was linked through small connections to the device leads. Subsequently, a 35 wt % KOH solution, saturated with isopropyl alcohol (IPA), was used to etch a single NW in the central region. Due to the anisotropy in etching, NW exhibits a tra‐ pezoidal cross-section. After this step, the aluminum contacts were defined by means of a lift-off process. The SiNWs, after the silicon etching, had a length (L) of 1mm, a top width (w) of 70 nm, a height (t) of 100 nm, and trapezoid base (W) ( of 200 nm, as illustrated in figure 14. The diagonal is based on (111) planes, whereas atop side is made up of a (100) plane. The available starting material had a resistivity of 10–20 Ωcm. At the end of the fabri‐ cation process, the Al electrodes were coated with PMMA to protect them from etching agents (e.g., buffer HF, NH4F, etc.) during the alkylation process. To assess the electrical characteristics of the various SiNW FETs, voltage-dependent back-gate measurements were performed in a probe station (Cascade Microtech Femtoguard, Summit/S300-11741B-6, USA) under the ambient conditions. In these measurements, back-gate voltages (Vg) of -20 V to +20 V, in steps of 1 V, were applied to the silicon substrate, of which the doping level is high enough such that the band bending in the Si substrate is negligible under operating condi‐ tions. In the present device geometry (figure 14), over 80 % of the applied gate potential is estimated to fall across the gate oxide, justifying the use of a large gate-voltage increment. For each gate voltage, the I–V characteristics were measured between the drain (d) and source (s) electrodes, contacted by a 0.5μm radius coax probe (73Ct-CMIA/05), at a bias range of between -6 V and +6 V, in 100-mV steps and/or from -0.2 to +0.2V in 4-mV steps, under ambient conditions.

**Figure 14.** In the right schematic the trapezoidal SiNW insulated from the silicon substrate by a buried SiO2 layer is shown. The SiNWs used in this study have a length (L) of 1µm, a top width (w) of 70 nm, and a height (t) of 100 nm, The diagonal is based on (111) planes, whereas the small top is based on (100) plane. W (trapezoid base) =200 nm. The left image is a SEM image of the SiNW device, which is coated with PMMA protection layer.

#### *5.1.1. Electrical effects of organic functionalities*

Figure 15 shows the current–voltage characteristics of SiO2-covered SiNW FETs and butylterminated SiNW FETs (as a representative example of the molecule-terminated SiNW FETs) at different gate voltages. The source–drain current (Isd) of both samples increased with increasing gate voltage, indicating that the transport through the semiconducting SiNW is dominated by negative carriers (electrons), as it applies to the n-type SiNW. Figure 15 depicts tansconductance (g) versus back-gate voltage (Vg) in the SiO2-SiNW-FET, butyl-SiNW-FET, 1,3-dioxan-2-ethyl-SiNW-FET, and propyl-alcohol-SiNW-FET. The data present‐ ed in figure 15 contain three distinctive features that seem to be at odds with one another and need to be explained. The first distinctive feature is an explicit dependence of the SiNW FET characteristics on the adsorbed molecules.

The zero-gate voltage (Vg=0), small-bias (Vsd=0–0.2V) tansconductance depends on the type of absorbed molecule. Particularly, one notices the following trend in the work function observed for these molecules (Table 4).: propyl-alcohol-SiNWs (0.5±0.1 μS)< SiO2-SiNWs (2.8±0.4 μS) < allyl-SiNWs (7.3±0.8 μS)< butyl-SiNWs (7.6±0.9 μS) < 1,3-diox‐ an-2-ethyl-SiNWs (10.1±0.9 μS). Furthermore, the threshold voltage (Vth), which is roughly defined as the minimum gate voltage for the low-bias conductance to reach the saturation level (figure16) correlates well with the work function in freshly prepared 1,3-dioxan-2-ethyl, butyl and propyl-alcohol functionalities altering the Vth values to -4V,-2V, and+2V,respectively. These correlations seem to suggest that the influence of the molecules on the operation of the SiNW FETs plays its most crucial role through a shift in the threshold gate voltage, as the effective work-function of the gate metal, or, equivalently, the electron affinity of the SiNW has been modified by introduction of the molecular layer. However, such a suggestion would be difficult to rationalize because the present molecular layer does not reside between the gate and the SiNW, but is lo‐ cated on the surface of the SiNW. Thus the molecular layer is not expected to directly shift the effective gate voltage [9]. Additionally, sign of the observed conductance shift is opposite to that of the expected work-function influence on the SiNW channel ex‐ pected. For example, allyl, butyl, and 1,3-dioxan-2-ethyl functionalities are expected to impede the conduction through the Si channel because of the reduced Si electron affin‐ ity (see Table 4) but indeed an enhanced small-bias conductance was observed. In con‐ trary to the distinctive feature for small gate-voltage characteristics, there are two other features concerning changes brought about by the adsorption of all the molecules. The first distinctive feature is the sizeable enhancement of the source–drain conductance at large, positive, gate voltages (Vg > +3V) from the one for SiO2-SiNWs upon introduc‐ tion of the molecules. However, this effect is almost independent of either the magni‐ tude or the sign of work-function, essentially leading to the same curve as in figure 16. The second feature is that the tansconductance of the molecularly-terminated SiNW-FET devices are significantly reduced from that of the SiO2-SiNWs at large negative gate bias (Vg < -5V). For SiO2-SiNW FETs, the tansconductance decreases from 4.54 μS at Vg=+20 V only slightly to 1.47 μS at Vg=-20 V, indicating absence of on/off character‐ istics in this Vg region [9].

range of between -6 V and +6 V, in 100-mV steps and/or from -0.2 to +0.2V in 4-mV steps,

**Figure 14.** In the right schematic the trapezoidal SiNW insulated from the silicon substrate by a buried SiO2 layer is shown. The SiNWs used in this study have a length (L) of 1µm, a top width (w) of 70 nm, and a height (t) of 100 nm, The diagonal is based on (111) planes, whereas the small top is based on (100) plane. W (trapezoid base) =200 nm.

Figure 15 shows the current–voltage characteristics of SiO2-covered SiNW FETs and butylterminated SiNW FETs (as a representative example of the molecule-terminated SiNW FETs) at different gate voltages. The source–drain current (Isd) of both samples increased with increasing gate voltage, indicating that the transport through the semiconducting SiNW is dominated by negative carriers (electrons), as it applies to the n-type SiNW. Figure 15 depicts tansconductance (g) versus back-gate voltage (Vg) in the SiO2-SiNW-FET, butyl-SiNW-FET, 1,3-dioxan-2-ethyl-SiNW-FET, and propyl-alcohol-SiNW-FET. The data present‐ ed in figure 15 contain three distinctive features that seem to be at odds with one another and need to be explained. The first distinctive feature is an explicit dependence of the SiNW

The zero-gate voltage (Vg=0), small-bias (Vsd=0–0.2V) tansconductance depends on the type of absorbed molecule. Particularly, one notices the following trend in the work function observed for these molecules (Table 4).: propyl-alcohol-SiNWs (0.5±0.1 μS)< SiO2-SiNWs (2.8±0.4 μS) < allyl-SiNWs (7.3±0.8 μS)< butyl-SiNWs (7.6±0.9 μS) < 1,3-diox‐ an-2-ethyl-SiNWs (10.1±0.9 μS). Furthermore, the threshold voltage (Vth), which is roughly defined as the minimum gate voltage for the low-bias conductance to reach the saturation level (figure16) correlates well with the work function in freshly prepared 1,3-dioxan-2-ethyl, butyl and propyl-alcohol functionalities altering the Vth values to -4V,-2V, and+2V,respectively. These correlations seem to suggest that the influence of

The left image is a SEM image of the SiNW device, which is coated with PMMA protection layer.

*5.1.1. Electrical effects of organic functionalities*

FET characteristics on the adsorbed molecules.

under ambient conditions.

196 Nanowires - Recent Advances

**Figure 15.** I–Vds curves of (a) SiO2-SiNW FET (b) butyl-SiNW FET at different back-gate voltages (Vg)

With adsorption of molecules, the channel current is more effectively (by orders of magnitude) turned off by the negative gate voltage, resulting in a significant improvement in the on/off ra‐ tio of the FET device. Particularly, on/off ratios of 9x103 , 4x103 , and 1x102 are observed for 1,3-di‐ oxan-2-ethyl-, butyl-, and propyl-alcohol-terminated SiNW-FETs, respectively. Nevertheless, it is worth to notice that models based on shifts in the threshold voltage from molecular dipolar layers are unable to explain the second and third characteristic observations.


**Table 4.** Change in work function due to molecule adsorption on Si(111) samples, as compared with SiO2-Si(111). The measured work function of SiO2-Si(111) samples was 4050 meV. The band banding was measured and found to be constant (80±5 meV) in all studied samples.

**Figure 16.** Tansconductance (g) versus back-gate voltage (Vg) in SiO2-SiNW FET, butyl-SiNW FET, 1,3-dioxan-2-ethyl-SiNW FET, and propyl alcohol- SiNW FET. The tansconductance values were obtained at Vsd=0.2 V.

While the transistor characteristics for SiO2-SiNW FETs have relative stability in air, proper‐ ties of molecule-covered SiNW-FETs were found to evolve upon air exposure. The time de‐ pendence of the average tansconductance at Vg=0 V of various SiNWs versus time in air is illustrated in Fig.17, by a normalized parameter, namely γ (=gt /gt=0). For SiO2-SiNWs, g is constant over the entire period of investigation. For both 1,3-dioxan-2-ethyl-SiNWs and bu‐ tyl-SiNWs, g decreased with time to 30–40 % of its initial values over a period of one month (see figure 17). For SiNWs covered with propyl-alcohol, the observed decrease in g was more gradual.

**Figure 17.** Normalized average conductance at Vg=0 of SiO2-SiNW FET, butyl-SiNW FET, dioxan -propyl -SiNW FET, and propyl alcohol SiNW FET, versus air exposure time.

#### **5.2. Photovoltaic cell based hybrid SiNWs**

it is worth to notice that models based on shifts in the threshold voltage from molecular dipolar

**Table 4.** Change in work function due to molecule adsorption on Si(111) samples, as compared with SiO2-Si(111). The measured work function of SiO2-Si(111) samples was 4050 meV. The band banding was measured and found to be

**Figure 16.** Tansconductance (g) versus back-gate voltage (Vg) in SiO2-SiNW FET, butyl-SiNW FET, 1,3-dioxan-2-ethyl-

While the transistor characteristics for SiO2-SiNW FETs have relative stability in air, proper‐ ties of molecule-covered SiNW-FETs were found to evolve upon air exposure. The time de‐ pendence of the average tansconductance at Vg=0 V of various SiNWs versus time in air is illustrated in Fig.17, by a normalized parameter, namely γ (=gt /gt=0). For SiO2-SiNWs, g is constant over the entire period of investigation. For both 1,3-dioxan-2-ethyl-SiNWs and bu‐ tyl-SiNWs, g decreased with time to 30–40 % of its initial values over a period of one month (see figure 17). For SiNWs covered with propyl-alcohol, the observed decrease in g was

SiNW FET, and propyl alcohol- SiNW FET. The tansconductance values were obtained at Vsd=0.2 V.

layers are unable to explain the second and third characteristic observations.

Propyl-alcohol Si(111) +500±50 SiO2-Si(111) 0±50 Allyl-Si(111) -150±50 Butyl-Si(111) -300±50 1,3-dioxan-2-ethyl-Si(111) -400±50

constant (80±5 meV) in all studied samples.

198 Nanowires - Recent Advances

more gradual.

Sample work-function [meV]

We compare three different surface terminations of SiNWs: (i) samples for which the oxide shell was removed by a standard HF dip, named H-SiNWs, (ii) samples with single monolayer of oxide grown after HF dip (14 days), shown by SiO2-SiNWs; (iii) samples further processed by methyl functionalization immediately after the HF dip, marked with CH3-SiNWs.

**Figure 18.** XP spectra of the Si2p region of SiO2-SiNW, H-SiNW and CH3-SiNW. The vertical solid line indicates the posi‐ tion of the chemically unshifted Si2p3/2 component in each spectrum.

The observed binding energy of the Si2p3/2 bulk signal allows determining the difference be‐ tween the Fermi level and the energy of the valance-band maximum at the surface. We de‐ rive the binding energy of the Si2p signal with respect to the valance band maximum from the comparison between the Si 2p core level and valance band spectra of the three samples and obtain the value of Ev-ESi2p3/2 = 98.72±0.02 eV. This is similar to previous studies done by Himpsel et al [49] and Hunger et al. i.e. 98.74 eV [50]. Correspondingly, the data for EF-EV yields: 0.83 eV, 0.98 eV and 1.05 eV for CH3-SiNW, H-SiNW and SiO2-SiNW, respectively. The values yield slight downward surface band bending (B.B.) for the H-SiNW and SiO2- SiNW and upward B.B for the CH3-SiNW as will be discussed later.

#### *5.2.1. Work function and band bending of the heterojunction*

Work function of all the samples measured independently by the Kelvin probe method as ΦSiO2-SiNW= 4.32 eV, ΦH-SiNW=4.26 eV, ΦCH3-SiNW=4.22 eV. Combining these data with the Fermi level position relative to the band edges (EF-EV), yields electron affinity values (χ) of 4.29 eV, 4.12 eV and 3.93 eV for SiO2-SiNW, H-SiNW and CH3-SiNW, respectively, according to χ= Φ-Eg +(EF-EV). Summarizing these data, the band diagram depicted in figure 19 can be obtained. The bulk Fermi level position was obtained from the specific resistance (1-5 Ωcm) of the n-type samples and gave EF-EV=0.88±0.02 eV. We get a slight upward band bending of 0.05 eV which is consistent with flat band conditions for the CH3-SiNW sample within our error margin. Identical results were found in photoemis‐ sion studies on two dimensional methyl terminated Si(111) by Hunger et al. We get downward band bending corresponding to electron accumulation at the surface of the H-SiNW and SiO2-SiNW. This uncommon aspect clearly indicates a yet unidentified sur‐ face doping process supporting the n-type character. As a tentative explanation, theory of Dittrich et al. can be referred that holds specific water based functional groups re‐ sponsible for this effect [51]. For the SiO2-SiNW, we add the band edges of the SiO2 top layer and obtain almost a symmetric heterojunction with identical valance and conduc‐ tion band discontinuities. With Eg=1.12 eV, estimating the electron affinity of the bulk silicon as 4.05 eV, the surface dipole, δss, caused by the different monolayer surfaces were calculated to be +0.24 eV, +0.07 eV and -0.12 eV for SiO2-SiNW, H-SiNW and CH3- SiNW respectively. The surface dipoles are attributed to different bond charge distribu‐ tion between the three samples.

#### *5.2.2. Photoelectron Yield Spectroscopy (PYS)*

The degree of surface band bending provides a measure for the quality of the electronic pas‐ sivation on a semiconductor surface. Flat band conditions can usually be achieved only when surface defect states like dangling bonds are essentially passivated making them elec‐ tronically inactive.

The improved electronic properties of the interface also influence directly the photoemission of electrons from the valance band edge and from surface states inside the band gap. These photoemission processes are studied by photoelectron yield spectroscopy (PYS).

Hybrid Silicon Nanowires: From Basic Research to Applied Nanotechnology http://dx.doi.org/10.5772/54383 201

The observed binding energy of the Si2p3/2 bulk signal allows determining the difference be‐ tween the Fermi level and the energy of the valance-band maximum at the surface. We de‐ rive the binding energy of the Si2p signal with respect to the valance band maximum from the comparison between the Si 2p core level and valance band spectra of the three samples and obtain the value of Ev-ESi2p3/2 = 98.72±0.02 eV. This is similar to previous studies done by Himpsel et al [49] and Hunger et al. i.e. 98.74 eV [50]. Correspondingly, the data for EF-EV yields: 0.83 eV, 0.98 eV and 1.05 eV for CH3-SiNW, H-SiNW and SiO2-SiNW, respectively. The values yield slight downward surface band bending (B.B.) for the H-SiNW and SiO2-

Work function of all the samples measured independently by the Kelvin probe method as ΦSiO2-SiNW= 4.32 eV, ΦH-SiNW=4.26 eV, ΦCH3-SiNW=4.22 eV. Combining these data with the Fermi level position relative to the band edges (EF-EV), yields electron affinity values (χ) of 4.29 eV, 4.12 eV and 3.93 eV for SiO2-SiNW, H-SiNW and CH3-SiNW, respectively, according to χ= Φ-Eg +(EF-EV). Summarizing these data, the band diagram depicted in figure 19 can be obtained. The bulk Fermi level position was obtained from the specific resistance (1-5 Ωcm) of the n-type samples and gave EF-EV=0.88±0.02 eV. We get a slight upward band bending of 0.05 eV which is consistent with flat band conditions for the CH3-SiNW sample within our error margin. Identical results were found in photoemis‐ sion studies on two dimensional methyl terminated Si(111) by Hunger et al. We get downward band bending corresponding to electron accumulation at the surface of the H-SiNW and SiO2-SiNW. This uncommon aspect clearly indicates a yet unidentified sur‐ face doping process supporting the n-type character. As a tentative explanation, theory of Dittrich et al. can be referred that holds specific water based functional groups re‐ sponsible for this effect [51]. For the SiO2-SiNW, we add the band edges of the SiO2 top layer and obtain almost a symmetric heterojunction with identical valance and conduc‐ tion band discontinuities. With Eg=1.12 eV, estimating the electron affinity of the bulk silicon as 4.05 eV, the surface dipole, δss, caused by the different monolayer surfaces were calculated to be +0.24 eV, +0.07 eV and -0.12 eV for SiO2-SiNW, H-SiNW and CH3- SiNW respectively. The surface dipoles are attributed to different bond charge distribu‐

The degree of surface band bending provides a measure for the quality of the electronic pas‐ sivation on a semiconductor surface. Flat band conditions can usually be achieved only when surface defect states like dangling bonds are essentially passivated making them elec‐

The improved electronic properties of the interface also influence directly the photoemission of electrons from the valance band edge and from surface states inside the band gap. These

photoemission processes are studied by photoelectron yield spectroscopy (PYS).

SiNW and upward B.B for the CH3-SiNW as will be discussed later.

*5.2.1. Work function and band bending of the heterojunction*

tion between the three samples.

tronically inactive.

200 Nanowires - Recent Advances

*5.2.2. Photoelectron Yield Spectroscopy (PYS)*

**Figure 19.** Surface energy band diagram of CH3-SiNW, H-SiNW and SiO2-SiNW surfaces. All the numbers are in eV units.

The photoelectron yield spectra of the three different NW samples investigated in this work are shown in figure 20. Each PY spectrum shows two thresholds near 5.0±0.2 eV and 4.2±0.2 eV. The higher energy band corresponds to the valence band density of states of the silicon while the lower energy band associates with defects in the band gap. These spectra essen‐ tially represent the occupied density of states. Since the reference energy for yield spectro‐ scopy is the vacuum level (EVac), we independently used the work function data from the Kelvin measurements (EVac-EF) to mark the spectral position of the Fermi level by the dotted vertical lines in figure 20.

The photoelectron intensity at lower photon energies compared to this level is due to ther‐ mal occupation of states beyond the Fermi level. The band edges of the two emissions were plotted by dashed lines as can be seen in figure 20. To this end, we can distinguish between the two emissions, for example, the CH3 show higher emission of 102 at 4.7 eV (note the log‐ arithmic scale). Obviously the ratio between defect states and valence band states is signifi‐ cantly different for the three samples. For a systematic comparison we have redrawn the yield spectra in figure 21. The top energy axes in those plots are now related to the valence band maximum instead of the vacuum level by using the electron affinities of the three sam‐ ples (see figure 21) and the band gap energy of silicon.

The spectra are further normalized at the energy of 0.76 eV below the valence band maxi‐ mum where they should be strongly dominated by the valence band emission only and thus, the PYS should be identical for all the three samples at this energy point (See figure 21). By this procedure, the relative defect contributions are scaled for all three samples to the Si valence band density of states and are better comparable. The highest defect density is seen for the SiO2-SiNWs and in this case the defect and the valence band emissions are hard‐ ly resolved as two different bands (see also figure 21a).

In the case of the H-SiNW, the defect band is already lower by a factor of about 2, and it can be well distinguished from the valence band emission. Finally, the lowest defect density is observed for the CH3-SiNWs. For this sample, however, a third band be‐ tween 4.7 and 5.3 eV is clearly visible and can be distinguished by a significantly small‐ er slope of the spectrum in this energy region compared to the spectra of the other two samples.

**Figure 20.** Photoemission yield spectrum of (a) SiO2-SiNW, (b) H-SiNW, (c) CH3-SiNW.

This additional band is even more obvious in the spectroscopic density of states that is a de‐ rivative of the yield spectra, as shown in figure 21b. Microscopic origin of this additional photoemission band which is presumably related to the methyl functionalization is not yet clear and is subject of further investigations. It is noteworthy to mention that the bandgap energies for SiO2-SiNWs and CH3-SiNWs are ~8.9 eV and 7.3 eV, respectively [52]. These val‐ ues notably differ and thus elucidate the importance of the surface termination with mole‐ cules for any electronic application of NWs.

**Figure 21.** Photoelectron yield Y(hν) spectra and spectroscopic density of states of SiO2-SiNW, H-SiNW and CH3-SiNW.

### *5.2.3. I-V curves of solar cells processed with functionalized SiNWs*

In the case of the H-SiNW, the defect band is already lower by a factor of about 2, and it can be well distinguished from the valence band emission. Finally, the lowest defect density is observed for the CH3-SiNWs. For this sample, however, a third band be‐ tween 4.7 and 5.3 eV is clearly visible and can be distinguished by a significantly small‐ er slope of the spectrum in this energy region compared to the spectra of the other two

**Figure 20.** Photoemission yield spectrum of (a) SiO2-SiNW, (b) H-SiNW, (c) CH3-SiNW.

cules for any electronic application of NWs.

This additional band is even more obvious in the spectroscopic density of states that is a de‐ rivative of the yield spectra, as shown in figure 21b. Microscopic origin of this additional photoemission band which is presumably related to the methyl functionalization is not yet clear and is subject of further investigations. It is noteworthy to mention that the bandgap energies for SiO2-SiNWs and CH3-SiNWs are ~8.9 eV and 7.3 eV, respectively [52]. These val‐ ues notably differ and thus elucidate the importance of the surface termination with mole‐

samples.

202 Nanowires - Recent Advances

Here we demonstrate a hybrid organic / inorganic solar cell based on a combination of a pol‐ ymer (PEDOT:PSS) constituting the hole conductor and SiNWs play the role of light absorb‐ er and electron conductor. NWs employed to assemble the photovoltaic cells are synthesized on the basis of metal-assisted etching rather than VLS growth. Solar cell per‐ formance differs between cells of SiNWs with various functionalities [53].

Surfaces of those SiNWs were terminated by H-, SiO2- and CH3- and integrated into a solar cell through identical processes. In all the cells the photogenerated electron-hole pairs are separat‐ ed at a heterojunction which forms at the SiNW/polymer interface (see figure 22a). The H-SiNWs can just be excluded from further solar cell studies due to the fact that such surfaces indicate hydrophobic behavior and the spun polymer does not wet the SiNW properly.

This hybrid solar cell concept follows the principle proposed by Lewis et al. [54] who made use of the advantageous separation in orthogonal directions of light absorption in the SiNWs and charge carrier separation at the wrap around heterojunction. This SiNW-based cell concept has the following advantages: (i) efficient absorption of light in SiNWs [55] (ii) short diffusion dis‐ tance for carriers to the wrapped around heterojunction; (iii) use of the air-stable and robust polymer, PEDOT:PSS, as an efficient hole conductor; (iv) the SiNW/ polymer solar cell design utilizes only 1 % of Si used in thin film cells of identical thicknesses. However, for photovoltaic application full coverage of SiNWs by PEDOT:PSS is desirable but not yet achieved (approxi‐ mately 320nm was covered on top SiNW edge as can be seen in figure 22a.

SiNWs serve mainly as light trapping structures; reduce diffusion distance to the nanowires length while uncovered surfaces represent recombination centers for charge carriers. Further studies on wetting agents added to PEDOT:PSS may result in full coverage and higher effi‐ ciencies. Figure 22b shows the current density-voltage (*J-V*) characteristics of CH3-SiNW/ PEDOT:PSS and SiO2-SiNW/PEDOT:PSS solar cells under AM1.5 illumination. SiO2-SiNW/ PEDOT:PSS samples bear a short circuit current (*Jsc*) of 1.6mA/cm2 , an open circuit voltage (*Voc*) of 320 mV, a fill factor (*FF*) of 0.53 and a conversion efficiency (*η*) of 0.28 %. CH3-SiNW/ PEDOT:PSS devices exhibit improved performance with *Jsc*, *Voc*, *FF* and *μ* magnitudes of 7.0 mA/cm2 , 399 mV, 0.44 and 1.2 %, respectively. The low values of FF and *Jsc* in both types of samples are attributed to high contact resistances (Rs 300Ω).

It should be noted that by decreasing the resistance we can improve the efficiency of the so‐ lar cell as shown by different groups [56-58]. In spite of notably low efficiencies relative to the values nowadays reported for efficient solar cells, their comparative increase (of about a factor of 4) upon methylation that proves a very promising prospective for this kind of sur‐ face functionalization.

The improved performance of the CH3-SiNWs is attributed to the removal of the tunnelling oxide and the reduction of defect states related PY which is caused by a reduced defect den‐ sity at the heterojunction interface. The advantage of methyl functionalization for the solar cell application has more aspects. The negative surface dipole polarity (-0.12 eV) of the CH3- SiNW leads to a favorable barrier formation between Si and PEDOT:PSS. Lewis et al. report that charge transfer rates for CH3-Si do not differ from those of H-Si so that CH3-monolayers represent no appreciable tunneling barrier for charge transfer at room temperature [59].

Moreover, using one monolayer of CH3 at the interface, allow a more efficient charge cou‐ pling between silicon and polymer. By reducing the defect density at CH3-SiNWs (see PYS section), the surface recombination is decreased and cause an improvement in *Voc*. Accord‐ ing to the Shockley diode equation *Voc* = *kBT/q*. ln(*Jsc*/*J0*), where *J0* is the saturation current and the other variables have their usual meanings. Since increase of *Jsc* influences *Voc* as well, it has to be mentioned that the *Voc* gain observed cannot rely on this effect alone.

Assuming a similar *J0*, which correspond to the removal of the tunneling oxide without changing the interface properties, the increase of *Jsc* would lead to a *Voc* gain of ∆*Voc* = kT/ q. ln(*Jsc,methyl*/*Jsc,oxide*) = *kBT/q*. ln(7.0/1.6) = 0.037 V. Regarding the observed gain of ∆*Voc* = 0.079 V, only a reduced surface recombination (as measured by PY) and/or a favorable barrier forma‐ tion (surface dipole) can establish consistency with the experimental data.

**Figure 22.** (a) Tilted view of the heterojunctionSiNW/PEDOT:PSS. (b) J-V characteristic under AM1.5 illuminations of the radial heterojunction solar cells from CH3-SiNW and SiO2-SiNW. Inset: schematic view of solar cell device structure.

A potential explanation of the increase in the PY of CH3-SiNW and an increased efficiency of processed solar cells could be that CH3-SiNWs simply absorb the solar light better than the other SiNW. This can experimentally falsified by the equal total absorbance (A), reflection (R) and transmission (T) of all three types of samples (see figure 23.

(*Voc*) of 320 mV, a fill factor (*FF*) of 0.53 and a conversion efficiency (*η*) of 0.28 %. CH3-SiNW/ PEDOT:PSS devices exhibit improved performance with *Jsc*, *Voc*, *FF* and *μ* magnitudes of 7.0

It should be noted that by decreasing the resistance we can improve the efficiency of the so‐ lar cell as shown by different groups [56-58]. In spite of notably low efficiencies relative to the values nowadays reported for efficient solar cells, their comparative increase (of about a factor of 4) upon methylation that proves a very promising prospective for this kind of sur‐

The improved performance of the CH3-SiNWs is attributed to the removal of the tunnelling oxide and the reduction of defect states related PY which is caused by a reduced defect den‐ sity at the heterojunction interface. The advantage of methyl functionalization for the solar cell application has more aspects. The negative surface dipole polarity (-0.12 eV) of the CH3- SiNW leads to a favorable barrier formation between Si and PEDOT:PSS. Lewis et al. report that charge transfer rates for CH3-Si do not differ from those of H-Si so that CH3-monolayers represent no appreciable tunneling barrier for charge transfer at room temperature [59].

Moreover, using one monolayer of CH3 at the interface, allow a more efficient charge cou‐ pling between silicon and polymer. By reducing the defect density at CH3-SiNWs (see PYS section), the surface recombination is decreased and cause an improvement in *Voc*. Accord‐

the other variables have their usual meanings. Since increase of *Jsc* influences *Voc* as well, it

Assuming a similar *J0*, which correspond to the removal of the tunneling oxide without changing the interface properties, the increase of *Jsc* would lead to a *Voc* gain of ∆*Voc* = kT/

only a reduced surface recombination (as measured by PY) and/or a favorable barrier forma‐

**Figure 22.** (a) Tilted view of the heterojunctionSiNW/PEDOT:PSS. (b) J-V characteristic under AM1.5 illuminations of the radial heterojunction solar cells from CH3-SiNW and SiO2-SiNW. Inset: schematic view of solar cell device structure.

ln(7.0/1.6) = 0.037 V. Regarding the observed gain of ∆*Voc* = 0.079 V,

has to be mentioned that the *Voc* gain observed cannot rely on this effect alone.

tion (surface dipole) can establish consistency with the experimental data.

ln(*Jsc*/*J0*), where *J0* is the saturation current and

samples are attributed to high contact resistances (Rs 300Ω).

ing to the Shockley diode equation *Voc* = *kBT/q*.

, 399 mV, 0.44 and 1.2 %, respectively. The low values of FF and *Jsc* in both types of

mA/cm2

204 Nanowires - Recent Advances

q.

ln(*Jsc,methyl*/*Jsc,oxide*) = *kBT/q*.

face functionalization.

All the three spectra (SiO2-SiNW, CH3-SiNW, H-SiNW) show the same basic structure and the same A, R, T in short wavelengths (hν > Eg). In wavelengths smaller than ~1000 nm, the three samples behave the same: reflection is around 10 %, the transmission is roughly 0 % and thus the absorption around 90 % for the whole visible spectrum. This tendency is ob‐ served even in multi monolayers of oxide. However, for wavelengths larger than ~1200 nm (attribute to surface state regime) the transmission is around 50 % for all three samples and the reflection is also nearly 50 %, except for the SiO2-SiNW sample which shows 40 %. The lower reflection in SiO2-SiNW is attributed to light adsorption of the surface state.

**Figure 23.** Spectra of SiO2-SiNW, H-SiNW and CH3-SiNW: (A) Reflectance (B) Transmission (C) absorbance. The dash line indicates the wavelength that corresponds to the SiNW band gap energy.

In summary, we terminated the SiNWs with organic molecules. The stability of the organic found to be depended on their coverage and surface. CH3 shows the highest oxide resistivi‐ ty, since it gives 100 % of coverage. To utilize the advantage of these molecules, we integrate them in FETs and solar cells. We showed that we can tune the properties of the FETs by at‐ taching molecules on the surface of the channel i.e. SiNW. The solar cells of hybrid SiNW shows higher efficiency than native oxide. This can be attributed to the low defects in the band gap as had been shown by photoelectron yield spectra.

## **Author details**

Muhammad Y. Bashouti1 , Matthias Pietsch1 , Kasra Sardashti1 , Gerald Brönstrup1 , Sebastian W. Schmitt1 , Sanjay K. Srivastava1 , Jürgen Ristein3 , Jordi Arbiol4 , Hossam Haick2 and Silke Christiansen1,5

1 Max-Planck-Institute for the Science of Light, Erlangen, Germany

2 The Department of Chemical Engineering and Russell Berrie Nanotechnology Institute, Technion – Israel Institute of Technology, Haifa, Israel

3 Institue for Technical Physics, University of Erlangen-Nuremberg, Erlangen, Germany

4 Institució Catalana de Recerca i EstudisAvançats (ICREA) and Institut de Ciència de Mate‐ rials de Barcelona,ICMAB-CSIC, Bellaterra, Spain

5 Institute of Photonic Technology, Jena, Germany

## **References**


[7] Bashouti, M.Y., et al., *Chemical Passivation of Silicon Nanowires with C(1)-C(6) Alkyl Chains through Covalent Si-C Bonds.* Journal of Physical Chemistry C, 2008. 112(49): p. 19168-19172.

taching molecules on the surface of the channel i.e. SiNW. The solar cells of hybrid SiNW shows higher efficiency than native oxide. This can be attributed to the low defects in the

, Kasra Sardashti1

, Jürgen Ristein3

2 The Department of Chemical Engineering and Russell Berrie Nanotechnology Institute,

4 Institució Catalana de Recerca i EstudisAvançats (ICREA) and Institut de Ciència de Mate‐

[1] Bandaru, P.R. and P. Pichanusakorn, *An outline of the synthesis and properties of silicon*

[2] Assad, O., et al., *Stable Scaffolds for Reacting Si Nanowires with Further Organic Func‐ tionalities while Preserving Si-C Passivation of Surface Sites.* Journal of the American

[3] Barth, S., et al., *Synthesis and applications of one-dimensional semiconductors.* Progress in

[4] Bronstrup, G., et al., *Statistical model on the optical properties of silicon nanowire mats.*

[5] Paska, Y., et al., *Enhanced Sensing of Nonpolar Volatile Organic Compounds by Silicon*

[6] Paska, Y. and H. Haick, *Interactive Effect of Hysteresis and Surface Chemistry on Gated Silicon Nanowire Gas Sensors.* Acs Applied Materials & Interfaces, 2012. 4(5): p.

*Nanowire Field Effect Transistors.* Acs Nano, 2011. 5(7): p. 5620-5626.

*nanowires.* Semiconductor Science and Technology, 2010. 25(2).

3 Institue for Technical Physics, University of Erlangen-Nuremberg, Erlangen, Germany

, Gerald Brönstrup1

,

, Jordi Arbiol4

,

band gap as had been shown by photoelectron yield spectra.

, Matthias Pietsch1

, Sanjay K. Srivastava1

1 Max-Planck-Institute for the Science of Light, Erlangen, Germany

and Silke Christiansen1,5

Technion – Israel Institute of Technology, Haifa, Israel

rials de Barcelona,ICMAB-CSIC, Bellaterra, Spain

5 Institute of Photonic Technology, Jena, Germany

Chemical Society, 2008. 130(52): p. 17670-+.

Materials Science, 2010. 55(6): p. 563-627.

Physical Review B, 2011. 84(12).

2604-2617.

**Author details**

206 Nanowires - Recent Advances

Muhammad Y. Bashouti1

Sebastian W. Schmitt1

Hossam Haick2

**References**


[37] Miyazaki, S., et al., *Surface Fermi level position of hydrogen passivated Si(111) surfaces.* Applied Physics Letters, 1996. 68(9): p. 1247-1249.

[22] Cohen, R., et al., *Molecular electronic tuning of Si surfaces.* Chemical Physics Letters,

[23] Ishii, H., et al., *Energy level alignment and interfacial electronic structures at organic/metal and organic/organic interfaces (vol 11, pg 605, 1999).* Advanced Materials, 1999. 11(12):

[24] Anagaw, A.Y., R.A. Wolkow, and G.A. DiLabio, *Theoretical study of work function mod‐ ification by organic molecule-derived linear nanostructure on H-silicon(100)-2 x 1.* Journal

[25] Hiremath, R.K., et al., *Molecularly controlled metal-semiconductor junctions on silicon sur‐*

[26] Aradi, B., et al., *Theoretical study of the chemical gap tuning in silicon nanowires.* Physical

[27] Leu, P.W., B. Shan, and K.J. Cho, *Surface chemical control of the electronic structure of silicon nanowires: Density functional calculations.* Physical Review B, 2006. 73(19).

[28] Wagner, R.S., et al., *Study of Filamentary Growth of Silicon Crystals from Vapor.* Journal

[29] Sivakov, V.A., et al., *Silicon nanowire oxidation: the influence of sidewall structure and*

[30] Bashouti, M.Y., et al., *Silicon nanowires terminated with methyl functionalities exhibit stronger Si-C bonds than equivalent 2D surfaces.* Physical Chemistry Chemical Physics,

[31] Haber, J.A. and N.S. Lewis, *Infrared and X-ray photoelectron spectroscopic studies of the reactions of hydrogen-terminated crystalline Si(111) and Si(100) surfaces with Br-2, I-2, and ferrocenium in alcohol solvents.* Journal of Physical Chemistry B, 2002. 106(14): p.

[32] Thanikasalam, P., T.K. Whidden, and D.K. Ferry, *Oxidation of silicon (100): Experimen‐ tal data versus a unified chemical model.* Journal of Vacuum Science & Technology B,

[33] Dimitrijev, S. and H.B. Harrison, *Modeling the growth of thin silicon oxide films on sili‐*

[34] Deal, B.E. and A.S. Grove, *General Relationship for Thermal Oxidation of Silicon.* Journal

[35] Liu, B.Z., et al., *Oxidation of silicon nanowires for top-gated field effect transistors.* Journal

[36] Buttner, C.C. and M. Zacharias, *Retarded oxidation of Si nanowires.* Applied Physics

*con.* Journal of Applied Physics, 1996. 80(4): p. 2467-2470.

of Vacuum Science & Technology A, 2008. 26(3): p. 370-374.

of Applied Physics, 1965. 36(12): p. 3770-&.

of Physical Chemistry C, 2008. 112(10): p. 3780-3784.

of Applied Physics, 1964. 35(10): p. 2993-&.

*gold distribution.* Nanotechnology, 2009. 20(40).

*face: A dipole effect.* Langmuir, 2008. 24(19): p. 11300-11306.

1997. 279(5-6): p. 270-274.

Review B, 2007. 76(3).

2009. 11(20): p. 3845-3848.

1996. 14(4): p. 2840-2844.

Letters, 2006. 89(26).

3639-3656.

p. 972-972.

208 Nanowires - Recent Advances


## **Wet – Chemically Etched Silicon Nanowire Solar Cells: Fabrication and Advanced Characterization**

Björn Hoffmann, Vladimir Sivakov, Sebastian W. Schmitt, Muhammad Y. Bashouti, Michael Latzel, Jiří Dluhoš, Jaroslav Jiruse and Silke Christiansen

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52585

## **1. Introduction**

[53] Sivakov, V., et al., *Silicon Nanowire-Based Solar Cells on Glass: Synthesis, Optical Proper‐*

[54] Boettcher, S.W., et al., *Energy-Conversion Properties of Vapor-Liquid-Solid-Grown Silicon*

[55] Bronstrup, G., et al., *A precise optical determination of nanoscale diameters of semiconduc‐*

[56] Ozdemir, B., et al., *Silicon nanowire - poly(3,4-ethylenedioxythiophene)-poly(styrenesulfo‐*

[57] Yuan, C.Z., et al., *Hierarchically structured carbon-based composites: Design, synthesis and their application in electrochemical capacitors.* Nanoscale, 2011. 3(2): p. 529-545.

[58] Shiu, S.C., et al., *Morphology Dependence of Silicon Nanowire/Poly(3,4-ethylenedioxythio‐ phene):Poly(styrenesulfonate) Heterojunction Solar Cells.* Chemistry of Materials, 2010.

[59] Maldonado, S. and N.S. Lewis, *Behavior of Electrodeposited Cd and Pb Schottky Junctions on CH3-Terminated n-Si(111) Surfaces.* Journal of the Electrochemical Society, 2009.

*ties, and Cell Parameters.* Nano Letters, 2009. 9(4): p. 1549-1554.

*Wire-Array Photocathodes.* Science, 2010. 327(5962): p. 185-187.

*nate) heterojunction solar cells.* Applied Physics Letters, 2011. 99(11).

*tor nanowires.* Nanotechnology, 2011. 22(38).

22(10): p. 3108-3113.

210 Nanowires - Recent Advances

156(2): p. H123-H128.

In the past decade the usage of solar energy production has increased rapidly and has be‐ come a reasonable alternative to fossil energy sources. Solar energy conversion is a chal‐ lenge of the millennium and solar cells are one eagerly thought way of converting solar energy into electrical energy. Solar cells are divided into three generations according to Mar‐ tin Green [1]: The 1st generation (1st gen) is till to date dominating production and installa‐ tion and makes use of bulk, essentially silicon wafers with planar cell geometry. Today, a typical silicon based cell uses 200μm thick Si wafers as substrates. The 2nd generation (2nd gen) uses thin-films rather that bulk wafers and materials can be wide spread. The idea is to use only very thin layers of photovoltaically active material on a passive mechanically sup‐ porting material e.g. glass. Such cells offer the option to use cheap substrates and even flexi‐ ble substrates [2] so that cost saving potentials as well as new applications can be envisaged. Until today, thin-film cells (2nd gen) do not live up to their full potential due to the fact that they are 50%-60% less efficient than wafer cells (1st gen). Solar cell concepts termed 3rd gen aim at the combination of low production costs and high efficiencies. In order to reach this goal, unique optical and physical properties of complex nanostructures as well as novel ma‐ terials and compounds are exploited. Silicon nanowires (SiNWs) are a very promising candi‐ date for next generation PV. The basic material is abundant and cheap and the semiconductor industry gained intense experience with its handling. The nanostructuring and cell preparation can also be performed in cheap and large scale processes. Thus it could easily be integrated into existing PV lines.

© 2012 Hoffmann et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Hoffmann et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### **1.1. Reasons for nanowires in PV concepts**

The conversion efficiency of solar cells is affected by many factors. The internal quantum ef‐ ficiency is already quite good for a long time and thus the industry concentrated more on optimizing the contacts for efficiency enhancement or reducing the overall costs. Whereas one efficiency record follows another in research [3], industry scale solar cells, however, are only improving slowly [4]. While optimizing factors like contact resistance or quantum effi‐ ciency only permit a moderate overall efficiency enhancement, the optical absorption is a property that can encourage large improvements of the SC efficiency. By using silicon nano‐ structures, the absorption of sunlight can be improved from typically 60% to over 90% in the wavelength range between 400 and 1100 nm [5–9]. This increased absorption is very encour‐ aging and makes the nanostructures so interesting for SC research as well as for future in‐ dustrial production.

One of many ways in order to reach this increased absorption of sunlight is the use of nano‐ wires constituting the SC absorbers. The nanowires act as light trapping specimens due to its strong scattering capacities [10] and for the same reason as antireflective coatings [11–13]. A lot of different methods exist that produce such highly absorbing nanostructures which are divided into bottom-up growth mechanisms [14–16] and top-down etching processes [17–20]. This work will concentrate on a metal assisted wet-chemical etching (MA-WCE) process that does not need any vacuum equipment and can be performed even on large sur‐ faces in a few minutes.

The application of micro- and nanowires in SCs is quite a new field of PV research. Having used the "Web of Knowledge" search engine, graphs were generated which are presented in Figure 11 . From the first publication in 2002 [21], the topic has rapidly increased in populari‐ ty and it has reached nearly 250 publications in 2011. The number of citations also shows a growing increase of nearly 5000 citations in 2011.

**Figure 1.** Publications and citations on nanowire-based solar cells in the last 10 years.

<sup>1</sup> http://www.webofknowledge.com/ Search properties: title includes (nanowire OR nanorod OR microwire OR micro‐ rod) AND (solar)

## **2. Experimental part**

**1.1. Reasons for nanowires in PV concepts**

dustrial production.

212 Nanowires - Recent Advances

faces in a few minutes.

growing increase of nearly 5000 citations in 2011.

**Figure 1.** Publications and citations on nanowire-based solar cells in the last 10 years.

Figure 11

rod) AND (solar)

The conversion efficiency of solar cells is affected by many factors. The internal quantum ef‐ ficiency is already quite good for a long time and thus the industry concentrated more on optimizing the contacts for efficiency enhancement or reducing the overall costs. Whereas one efficiency record follows another in research [3], industry scale solar cells, however, are only improving slowly [4]. While optimizing factors like contact resistance or quantum effi‐ ciency only permit a moderate overall efficiency enhancement, the optical absorption is a property that can encourage large improvements of the SC efficiency. By using silicon nano‐ structures, the absorption of sunlight can be improved from typically 60% to over 90% in the wavelength range between 400 and 1100 nm [5–9]. This increased absorption is very encour‐ aging and makes the nanostructures so interesting for SC research as well as for future in‐

One of many ways in order to reach this increased absorption of sunlight is the use of nano‐ wires constituting the SC absorbers. The nanowires act as light trapping specimens due to its strong scattering capacities [10] and for the same reason as antireflective coatings [11–13]. A lot of different methods exist that produce such highly absorbing nanostructures which are divided into bottom-up growth mechanisms [14–16] and top-down etching processes [17–20]. This work will concentrate on a metal assisted wet-chemical etching (MA-WCE) process that does not need any vacuum equipment and can be performed even on large sur‐

The application of micro- and nanowires in SCs is quite a new field of PV research. Having used the "Web of Knowledge" search engine, graphs were generated which are presented in

ty and it has reached nearly 250 publications in 2011. The number of citations also shows a

1 http://www.webofknowledge.com/ Search properties: title includes (nanowire OR nanorod OR microwire OR micro‐

. From the first publication in 2002 [21], the topic has rapidly increased in populari‐

### **2.1. Top-down fabrication of silicon**

SiNWs can be produced by many different methods, normally divided into bottom-up growth mechanisms and top-down etching processes. While growing of SiNWs normally is conducted by using the vapor-liquid-solid (VLS) mechanism [14], which needs high temper‐ atures, toxic gases and expensive vacuum equipment, top-down etching can be performed in easier ways. One differentiates between physical (dry) and chemical (wet) etching. Physi‐ cal or dry etching is usually conducted in reactive gases offered in a reactive ion etching (RIE) setup [22], while chemical etching is usually based on wet chemistries, preferably on the basis of hydrofluoric acid for the case of silicon. Both methods have assets and draw‐ backs, but when it comes to the question of fast and cost-effective fabrication with the per‐ spective of mass production, wet-chemical (WC) etching is intriguing.

Since details of WC etching processes were already discussed in a previous Intech book chapter [20], this work will only give a short summary and point out the latest optimization processes. The SiNW substrates that constitute the SC absorber are fabricated in a two-step MA-WCE procedure. Therefore, the substrate material, which is typically a lowly doped (1-5 Ohmcm) n-type Si wafer, passes four different etching solutions. In the beginning, the native oxide is removed with 2% hydrofluoric acid (HF, solution I). In the following step (solution II), silver (Ag) nanoparticles are formed on the surface of the wafer in a mixture of 0.02 mo‐ lar (M) AgNO3 and 5M HF in a ratio of 1:10. The duration of this step influences the density of particles and thus also the forming SiNW geometry. This procedure needs to be less than 30 seconds. Otherwise, the Si is completely coated by a continuous Ag layer and no access of the etching solutions to the Si surface can be achieved. Controlled SiNW properties can be achieved when using the following etching steps. Solution III consists of a mixture of 5M HF and 30% H2O2 in a ratio of 1:1. In this step the Si is oxidized by the interplay of H2O2 and the nanoscale Ag particles that form the thermodynamically unstable intermediate compound AgO(OH) which is a very efficient oxidant that oxidizes the silicon beneath the Ag nanopar‐ ticle. The resulting SiO2 is dissolved by the HF in the solution. Thus, the Si is removed below the Ag particles which dig into the Si wafer and thus move away from the wafer surface in‐ ducing the SiNW formation. The direction of this Ag nanoparticle movement is anisotropic following certain crystal orientations and depends on the temperature as well [20]. Under controlled conditions, the SiNWs form perpendicularly to the surface of a Si(100) wafer and are oblique to the surface of a Si(111) wafer following <100>-directions [18]. In the last step (solution IV), concentrated nitric acid is used to remove the remaining Ag nanoparticles. Figure 2 illustrates the etching steps process.

During the SiNW formation, an increasing absorption can be observed after a few seconds. The shiny Si wafer surface turns matt finished and changes to a brown color which be‐ comes deep black after one minute of etching. After three minutes the SiNWs reach a length of 1μm and elongate with a rate of 1 μm/min after that.

**Figure 2.** The WCE process is based on a sequence of four solutions starting with oxide removal (solution I: dilute HF, not shown), followed by solution based Ag nanoparticle deposition (solution II), Ag nanoparticle assisted catalytic etching of Si (solution III) and Ag nanoparticle removal (solution IV).

#### **2.2. Fabrication of Semiconductor-Insulator-Semiconductor (SIS) solar cells**

SiNWs enable several concepts of efficient solar cell geometries. The charge carrier dividing junction (p-n or Schottky type) can be applied in a radial [23,24] or axial [25,26] concept in‐ side the NWs and a possible solution is also the wrapping of NWs with a second material [27]. In this work, we used a semiconductor-insulator-semiconductor (SIS) heterojunction solar cell concept that has already been developed in the 1970s [28–30]. Therefore, the com‐ plex SiNW structure is uniformly coated with a very thin tunneling barrier and a thicker lay‐ er of aluminum (Al) doped zinc oxide (ZnO) as a second, transparent wide band gap, degenerately doped semiconductor. In order to achieve this conformal coating, atomic layer deposition (ALD, Oxford Opal) is used. This technique can deposit many materials in a selflimiting process with a thickness control at the Angstrom level even on highly complex sur‐ faces [31,32]. Two different classes of precursors are needed for the ALD deposition of metal oxides like Al2O3, TiO2 or ZnO: metal containing molecules and an oxidizing agent. The two precursors are pulsed into the ALD vacuum chamber one after the other by using very short pulse durations of typically <100ms, while longer inert gas purging steps are used between the precursor pulses to remove residual gas from the chamber. The special advantage of an ALD process is that the precursors are forming a self-limiting monolayer that does not allow further deposition of that precursor. The oxidizing agent can be water or an oxygen plasma which removes the functional groups of the previously deposited metal precursor molecules and forms the metal oxide [33,34]. This step is also self-limiting so that a completed cycle forms a stoichiometric monolayer of the desired material even on 3D surfaces. In the case of Al2O3, trimethylaluminum and oxygen plasma were used for its formation and for ZnO dep‐ osition, diethylzinc and water pulses are applied. The doping of ZnO with aluminum was realized by substituting every 20th cycle of ZnO by one cycle of Al2O3 [35,36].

Two different insulating layer materials were tested to form the barrier layer in the SIS cell: TiO2 and Al2O3. Depending on the material, 10 to 30 Å of the barrier layer were deposited on‐ to the SiNWs. Subsequently, the second semiconductor, Al doped ZnO (AZO), which acts as transparent front contact at the same time, is deposited in the same ALD reactor. The thick‐ ness of this layer was typically around 450 nm which resulted in a complete filling of the gaps in between the SiNWs and also created a continuous AZO layer on top of the SiNWs for effi‐ cient charge carrier conduction. Figure 3 shows the three steps of the ALD process.

Wet – Chemically Etched Silicon Nanowire Solar Cells: Fabrication and Advanced Characterization http://dx.doi.org/10.5772/52585 215

**Figure 3.** Schematic of the layout of an SIS solar cell concept based on SiNWs: etched SiNWs (left) are coated with a thin tunnel barrier (middle) and subsequently with a transparent front contact – Al doped ZnO (right).

**Figure 2.** The WCE process is based on a sequence of four solutions starting with oxide removal (solution I: dilute HF, not shown), followed by solution based Ag nanoparticle deposition (solution II), Ag nanoparticle assisted catalytic

SiNWs enable several concepts of efficient solar cell geometries. The charge carrier dividing junction (p-n or Schottky type) can be applied in a radial [23,24] or axial [25,26] concept in‐ side the NWs and a possible solution is also the wrapping of NWs with a second material [27]. In this work, we used a semiconductor-insulator-semiconductor (SIS) heterojunction solar cell concept that has already been developed in the 1970s [28–30]. Therefore, the com‐ plex SiNW structure is uniformly coated with a very thin tunneling barrier and a thicker lay‐ er of aluminum (Al) doped zinc oxide (ZnO) as a second, transparent wide band gap, degenerately doped semiconductor. In order to achieve this conformal coating, atomic layer deposition (ALD, Oxford Opal) is used. This technique can deposit many materials in a selflimiting process with a thickness control at the Angstrom level even on highly complex sur‐ faces [31,32]. Two different classes of precursors are needed for the ALD deposition of metal oxides like Al2O3, TiO2 or ZnO: metal containing molecules and an oxidizing agent. The two precursors are pulsed into the ALD vacuum chamber one after the other by using very short pulse durations of typically <100ms, while longer inert gas purging steps are used between the precursor pulses to remove residual gas from the chamber. The special advantage of an ALD process is that the precursors are forming a self-limiting monolayer that does not allow further deposition of that precursor. The oxidizing agent can be water or an oxygen plasma which removes the functional groups of the previously deposited metal precursor molecules and forms the metal oxide [33,34]. This step is also self-limiting so that a completed cycle forms a stoichiometric monolayer of the desired material even on 3D surfaces. In the case of Al2O3, trimethylaluminum and oxygen plasma were used for its formation and for ZnO dep‐ osition, diethylzinc and water pulses are applied. The doping of ZnO with aluminum was

**2.2. Fabrication of Semiconductor-Insulator-Semiconductor (SIS) solar cells**

realized by substituting every 20th cycle of ZnO by one cycle of Al2O3 [35,36].

cient charge carrier conduction. Figure 3 shows the three steps of the ALD process.

Two different insulating layer materials were tested to form the barrier layer in the SIS cell: TiO2 and Al2O3. Depending on the material, 10 to 30 Å of the barrier layer were deposited on‐ to the SiNWs. Subsequently, the second semiconductor, Al doped ZnO (AZO), which acts as transparent front contact at the same time, is deposited in the same ALD reactor. The thick‐ ness of this layer was typically around 450 nm which resulted in a complete filling of the gaps in between the SiNWs and also created a continuous AZO layer on top of the SiNWs for effi‐

etching of Si (solution III) and Ag nanoparticle removal (solution IV).

214 Nanowires - Recent Advances

The structural analysis of the completed SIS solar cells was performed by scanning electron microscopy (SEM) techniques including 3D analysis supported by a sequence of focused ion beam (FIB) etching steps. Thereby, the gap filling between SiNWs and SiNW topologies were investigated. Figure 4 shows two colored cross section micrographs of as etched SiNWs and of a completed SIS solar cell that contains wrapping layer stacks on the SiNWs of barrier layer and AZO.

Additionally, the complete SCs were analyzed by glow discharge optical emission spectrom‐ etry (GD-OES, Horiba Jobin Yvon JY 5000 RF) to obtain a depth profile of the compositions of the material [37,38]. In a GD plasma source, cathodic sputtering is used to remove materi‐ al layer-by-layer from a sample surface with an accuracy of a few nanometers [39]. The sam‐ ple atoms are transported into a plasma plume, in which they are excited and ionized through collisions with electrons, ions and the metastable carrier gas (here: Ar).

**Figure 4.** Left: colored SEM micrograph of as etched SiNWs. Right: colored SEM micrograph of a FIB polished cross section of a completed SIS solar cell. The SiNWs (blue) are embedded in barrier layer and Al:ZnO (AZO) (orange).

The analyte isotopes of interest can then be detected via downstream spectroscopic techni‐ ques. In the presented study, OES is performed by analyzing the intensities of characteristic emission lines of excited analytes with a spectrometer (typically a Paschen–Runge mono‐ chromator or Czerny–Turner polychromator) with A GD-OES Such an analysis was per‐ formed on a SiNW SC with a TiO2 tunnel barrier in the optimization of the solar cell production process. In the SEM profile in Figure 5, the SC can be divided into three regions: region 1 shows the front contact which solely consists of aluminum doped zinc oxide (AZO). The second region is composed of the SiNW array with a wrapping TiO2 layer which is filled with AZO. In the third region, the bulk Si wafer is reached. From the measured spectra several interesting information can be extracted [40]. In region 1 the Al content is found to be 2.5 mass% (m%) leading to an ideal conductivity of the AZO layers. In region 2, the silicon content increases only slowly in depth which leads to the assumption that the wires are tapered. This can be interesting for further theoretical investigations regarding op‐ tical simulations of light trapping effects. After having checked the residual silver, it is re‐ vealed that the cleaning process of the nitric acid did not completely remove the metal. According to these findings, the cleaning process was improved to eliminate residual silver contaminations.

The GD-OES measurements enable another interesting analysis of the silicon nanostructure: Knowing the total mass and thickness of the tunnel barrier layer, the surface of the nano‐ structures can be calculated. Thereby, the surface enhancement of the etched nanostructures as compared to a flat wafer can be derived. In the case of the TiO2 barrier SiNW SC (Figure 5), an enhancement factor of 6.2 was calculated. This factor is strongly dependent on the etching conditions and can easily reach values of 40 and more. For a solar cell, a higher sur‐ face amount is a clear disadvantage because higher junction areas result in higher surface recombination of charge carriers. On the other hand a higher surface increases the absorp‐ tion of light and leads to an overall higher charge carrier collection at the interface. Thus, an optimum has to be found between the counteracting mechanisms.

**Figure 5.** Top: SiNW SC cross section and scheme, showing three different regions inside the cell. Bottom: GD-OES spectra illustrating the depth dependent mass% of the elements Si, O, Zn, Al, Ti and Ag. Adapted from [40].

During the research state, typical SCs had only a size of a few mm² and thus did not need metal front contacts. In the later course of cell development, first demonstrators that had the size to power electrical devices were needed. Therefore, two different prototypes with gold front contact grids and sizes of 36cm² and 1.4cm² were fabricated (see Figure 6). To demon‐ strate the functionality of the SCs, the small demonstrator was contacted to a motor with a rotating wheel and was illuminated by a battery powered flashlight. Even though the flash‐ light delivered only a fraction of the sunlight's power, the SC produced enough energy to drive the motor.

**Figure 6.** Left: SiNW SC prototypes with gold front contacts. The large cell has a size of 36 cm², the small cell measures 1.4 cm². For comparison: a 10 €-cent coin. Right: The 1.4 cm² cell under a battery-driven flashlight produces enough power to rotate a small toy device.

## **3. Characterization**

region 1 shows the front contact which solely consists of aluminum doped zinc oxide (AZO). The second region is composed of the SiNW array with a wrapping TiO2 layer which is filled with AZO. In the third region, the bulk Si wafer is reached. From the measured spectra several interesting information can be extracted [40]. In region 1 the Al content is found to be 2.5 mass% (m%) leading to an ideal conductivity of the AZO layers. In region 2, the silicon content increases only slowly in depth which leads to the assumption that the wires are tapered. This can be interesting for further theoretical investigations regarding op‐ tical simulations of light trapping effects. After having checked the residual silver, it is re‐ vealed that the cleaning process of the nitric acid did not completely remove the metal. According to these findings, the cleaning process was improved to eliminate residual silver

The GD-OES measurements enable another interesting analysis of the silicon nanostructure: Knowing the total mass and thickness of the tunnel barrier layer, the surface of the nano‐ structures can be calculated. Thereby, the surface enhancement of the etched nanostructures as compared to a flat wafer can be derived. In the case of the TiO2 barrier SiNW SC (Figure 5), an enhancement factor of 6.2 was calculated. This factor is strongly dependent on the etching conditions and can easily reach values of 40 and more. For a solar cell, a higher sur‐ face amount is a clear disadvantage because higher junction areas result in higher surface recombination of charge carriers. On the other hand a higher surface increases the absorp‐ tion of light and leads to an overall higher charge carrier collection at the interface. Thus, an

**Figure 5.** Top: SiNW SC cross section and scheme, showing three different regions inside the cell. Bottom: GD-OES

spectra illustrating the depth dependent mass% of the elements Si, O, Zn, Al, Ti and Ag. Adapted from [40].

optimum has to be found between the counteracting mechanisms.

contaminations.

216 Nanowires - Recent Advances

The characterization of our fabricated SCs can be divided into investigation of structural and electro-optical properties. Understandably, the conversion efficiency is the most important property. It is determined in a current-voltage (I-V) measurement under a solar simulator that reproduces the certified AM 1.5 solar spectrum. An optical analysis was performed with an integrating sphere in order to investigate the connection between light absorption and efficiency. Figure 7 shows the measured photocurrent under illumination at -2V of 12 different solar cells and the respective optical absorption (mean value between 400 – 1100 nm). In most cases a connection between these two values is clearly visible. Therefore, it is crucial to optimize the absorption by tailoring the nanostructured material composite. Fur‐ thermore it can be seen that even a small reduction in absorption from 90% to 88% (wafer 1 and 4) can cause a serious drop in photocurrent from 35 mA to 15 mA. Although other ef‐ fects may play a role, this effect cannot be neglected.

After these findings the best solar cells were analyzed in detail regarding energy conversion and quantum efficiency as well as local conversion properties in terms of light and electron beam induced current mappings (LBIC/EBIC).

Over 120 solar cells with three different types of Si wafers and 7 different etching conditions were fabricated. For the record cell a n-type Si(111) wafer with a doping of 1-9 Ohm\*cm was used that was etched for 30s/6min in the two etching solutions. The Al2O3 tunneling barrier had a thickness of 12 Å. From this cell a small piece of 8mm² size was analyzed in the solar simulator. The SC shows a clearly visible diode-like behavior in the dark with a rectification factor of 6600 at +/-2V. Under AM1.5 illumination the record cell reaches an open circuit voltage VOC of 453 mV, a short circuit current density JSC of 35 mA/cm² and a fill factor (FF) of 57%. This results in a conversion efficiency of 9.1% which is an impressive result for a nanowire based solar cell with as little optimization as we did, e.g. in terms of surface passi‐ vation [41,42] or size and shape of the SiNWs [22,43,44]. The corresponding dark and illumi‐ nated J-V curves are shown in Figure 8.

**Figure 7.** Measured photocurrent (at -2V) and absorption of 12 different solar cells. A correlation is clearly visible.

**Figure 8.** Dark and illuminated current density – voltage (J-V) curves of our record SiNW-SC out of 120 test samples. Very high JSC combined with average VOC and fill factor yield in a conversion efficiency of 9.1%.

#### **3.1. Local electro-optical properties**

used that was etched for 30s/6min in the two etching solutions. The Al2O3 tunneling barrier had a thickness of 12 Å. From this cell a small piece of 8mm² size was analyzed in the solar simulator. The SC shows a clearly visible diode-like behavior in the dark with a rectification factor of 6600 at +/-2V. Under AM1.5 illumination the record cell reaches an open circuit voltage VOC of 453 mV, a short circuit current density JSC of 35 mA/cm² and a fill factor (FF) of 57%. This results in a conversion efficiency of 9.1% which is an impressive result for a nanowire based solar cell with as little optimization as we did, e.g. in terms of surface passi‐ vation [41,42] or size and shape of the SiNWs [22,43,44]. The corresponding dark and illumi‐

**Figure 7.** Measured photocurrent (at -2V) and absorption of 12 different solar cells. A correlation is clearly visible.

**Figure 8.** Dark and illuminated current density – voltage (J-V) curves of our record SiNW-SC out of 120 test samples.

Very high JSC combined with average VOC and fill factor yield in a conversion efficiency of 9.1%.

nated J-V curves are shown in Figure 8.

218 Nanowires - Recent Advances

To determine the influence of structural inhomogeneity on the electrical properties, it is necessary to generate localized charge carriers inside the solar cell and perform a map‐ ping across the cell while monitoring the generated external current between front and back contacts. Two different methods enable such an analysis: light and electron beam induced current (LBIC [45] & EBIC [25, 46]) measurements. In LBIC, a laser is scanned over the solar cell surface and the generated current is measured for each spot. In this way an efficiency mapping can be obtained. By using lasers with different wavelengths the response of the solar cell to distinct excitation energies can be investigated. Further‐ more, an external quantum efficiency (EQE) mapping can be calculated from such a measurement. The disadvantage of LBIC is its large laser spot focus which inhibits the analysis of μm-sized defects. Here EBIC becomes suitable. The electron beam of an SEM can be focused on a few nm and with a low acceleration voltage the expansion of excit‐ ed charge carriers can be kept at a minimum of <100nm. Every incident electron gener‐ ates thousands of charge carrier pairs that can be separated by the nearby SIS junction. This enables the visualization of the internal charge carrier separation at the SiNW front contact interface.

The LBIC and EBIC analyses were performed on the 1.4 cm² prototypes with metal front grid shown in Figure 6. For the LBIC measurements, a custom built system with four differ‐ ent lasers with wavelengths of 405, 544, 633 and 1064 nm was applied. Figure 9 shows the mapping results and an EBIC mapping of the same cell as well in order to be able to draw a comparison. The LBIC mappings are normalized to show the dependence of the power out‐ put on the light wavelength. Nearly no current is generated at 405 nm. At 544 nm, several areas are visible that have better or worse efficiency. These areas arise from an inhomogene‐ ous etching process which generates areas with slightly higher or lower absorption. At 633 nm two small areas reach a generated current of 25 μA. Visible for the naked eye, these areas appear darker than the remaining surface. At 1064 nm, the light is not absorbed in the SiNWs themselves anymore but in the bulk wafer below. Therefore the signal is very homo‐ geneous as the nanowires do not influence the current generation anymore. The EBIC map‐ ping can be compared best with the 633 nm LBIC mapping. Due to the higher resolution of the EBIC, this measurement shows much more details.

In order to explain the different current amounts in the LBIC measurements in a better way, the EQE of the SC was measured. The results are shown in Figure 10. The wave‐ length-dependent behavior is in accordance with the determined LBIC mappings. At 405 nm, the EQE is at nearly zero and 633 nm exhibits the highest EQE from all four laser wavelengths. Although the EQE curve is similar in shape to typical commercial solar cell EQE results, only lower values are reached. The reasons are still under investigation. Furthermore, Figure 10 shows the optical absorption of the SC which was calculated by 1 minus the measured total reflectance. In Figure 11, a high resolution large scale EBIC scan of a SC demonstrator is shown. The simultaneously gathered SEM and EBIC signals are coded in brown and green color schemes. The EBIC signal clearly shows regions of lower or higher current due to etching variations and also many circular defects which seem to be the result of pinholes in the tunnel barrier layers. In several circular defects, a dirt particle in the middle seems to be the cause of the short circuit.

**Figure 9.** LBIC mapping at four different laser wavelengths and the corresponding EBIC measurement at 30 kV accel‐ erating voltage. The inactive region in the top-right of each LBIC measurement is caused by shading from the mount‐ ing.

**Figure 10.** Absorption (1-Rtot) and EQE of a SiNW-based solar cell prototype.

The EBIC measurement is a powerful tool not only for observing the surface of a solar cell but also for looking inside the cell when combining sequences of EBIC analyses and FIB sec‐ tioning and polishing. Combined with a focused ion beam (FIB), information on sub-surface structures can be obtained. In a tomographic approach, a section into the SiNW based SC is carried out and a sub-surface slice of a few μm in depth and length is polished. The pre‐ pared area can subsequently be investigated by EBIC mapping and typical SEM SE imaging. Further slicing and polishing and repeated EBIC analysis permit a 3D reconstruction of EBIC as well as SE data resulting in tomographic information. Figure 12 represents the con‐ cept of a cross sectional EBIC measurement: The electron beam scans over the ion beam pol‐ ished cross section of a SiNW based SIS SC and the current that flows between the front and back contact is amplified and mapped by using the electron beam position. A resulting im‐ age is also shown.

seem to be the result of pinholes in the tunnel barrier layers. In several circular defects,

**Figure 9.** LBIC mapping at four different laser wavelengths and the corresponding EBIC measurement at 30 kV accel‐ erating voltage. The inactive region in the top-right of each LBIC measurement is caused by shading from the mount‐

The EBIC measurement is a powerful tool not only for observing the surface of a solar cell but also for looking inside the cell when combining sequences of EBIC analyses and FIB sec‐ tioning and polishing. Combined with a focused ion beam (FIB), information on sub-surface structures can be obtained. In a tomographic approach, a section into the SiNW based SC is

**Figure 10.** Absorption (1-Rtot) and EQE of a SiNW-based solar cell prototype.

ing.

220 Nanowires - Recent Advances

a dirt particle in the middle seems to be the cause of the short circuit.

**Figure 11.** A large scale combined SEM/EBIC image of a solar cell surface. Point defects appear as darker dots and larger sized contrast differences indicate structural variations due to inhomogeneous etching.

**Figure 12.** Left: Scheme of a cross sectional EBIC measurement of a SiNW based SIS SC. Right: colored dual-signal mi‐ crograph that shows the SEM image in red and the EBIC in green. A pronounced EBIC current of an individual SiNW is clearly visible in green.

The 3D-EBIC is devoted to solve an important question in the field of SiNW solar cells: Are the nanowires active photovoltaic components or do they only act as passive anti-reflecting building blocks? It is generally accepted that SiNWs absorb light by generating charge carri‐ er pairs [10]. However, whether it is possible to separate electrons and holes prior to their recombination is a question to be answered for each and every material and cell concept again. To answer these interesting and important questions a dual beam SEM/FIB system (Tescan Lyra 3) with an integrated quantitative EBIC detector (Point electronic and Kamm‐ rath & Weiss) is used. The gallium ion beam of the FIB cuts a volume of 10x10x10 μm³ of the solar cell perpendicular to the surface slice by slice. The electron beam observes each slice from an angle of 55° to the perpendicular and creates a secondary electron image as well as an EBIC image. Afterwards, these images can be combined in order to create a 3D recon‐ struction of the solar cell. In Figure 13, three slices of a 3D reconstruction are displayed. The assembled dataset offers a multitude of visualizations which can be shown best in a video. Therefore, the microscopy manufacturer Tescan provides a video of the analysis of our SiNW-SIS-SC which can be viewed via this internet link: http://www.youtube.com/watch? v=X9O6pwcN4Tg. It shows the possibilities in signal combination as well as the visualiza‐ tion of complex 3D surfaces.

**Figure 13.** Three single slices out of 400 that show the SE-signal (top) as well as the EBIC signal (bottom) at the same position.

The 3D determination of electrical properties becomes very important in times of emerging nanostructured solar cells. Our results which are shown in Figure 12 and 13, prove that nanowires can be active photovoltaic components. This can be seen in the second slice where a single nanowire is visible in the EBIC image. Otherwise, many NWs are visible in the SEM image that cannot be seen in the EBIC signal. The difference between active and passive PV component seems to lie in the diameter of the wire. NWs, which are smaller than 50nm, exhibit no EBIC signal. That can be explained by the high surface to volume ratio, which leads to higher recombination, or a lesser absorption and thus no charge carrier gen‐ eration occurs. These information are of great value when it comes to further optimization of the initial etching process. The nanowire dimensions should be in an area where high ab‐ sorption is still achieved and charge carrier separation can take place inside the wires.

## **4. Outlook on nanowire research**

again. To answer these interesting and important questions a dual beam SEM/FIB system (Tescan Lyra 3) with an integrated quantitative EBIC detector (Point electronic and Kamm‐ rath & Weiss) is used. The gallium ion beam of the FIB cuts a volume of 10x10x10 μm³ of the solar cell perpendicular to the surface slice by slice. The electron beam observes each slice from an angle of 55° to the perpendicular and creates a secondary electron image as well as an EBIC image. Afterwards, these images can be combined in order to create a 3D recon‐ struction of the solar cell. In Figure 13, three slices of a 3D reconstruction are displayed. The assembled dataset offers a multitude of visualizations which can be shown best in a video. Therefore, the microscopy manufacturer Tescan provides a video of the analysis of our SiNW-SIS-SC which can be viewed via this internet link: http://www.youtube.com/watch? v=X9O6pwcN4Tg. It shows the possibilities in signal combination as well as the visualiza‐

**Figure 13.** Three single slices out of 400 that show the SE-signal (top) as well as the EBIC signal (bottom) at the same

The 3D determination of electrical properties becomes very important in times of emerging nanostructured solar cells. Our results which are shown in Figure 12 and 13, prove that nanowires can be active photovoltaic components. This can be seen in the second slice where a single nanowire is visible in the EBIC image. Otherwise, many NWs are visible in the SEM image that cannot be seen in the EBIC signal. The difference between active and passive PV component seems to lie in the diameter of the wire. NWs, which are smaller than 50nm, exhibit no EBIC signal. That can be explained by the high surface to volume ratio, which leads to higher recombination, or a lesser absorption and thus no charge carrier gen‐ eration occurs. These information are of great value when it comes to further optimization of the initial etching process. The nanowire dimensions should be in an area where high ab‐ sorption is still achieved and charge carrier separation can take place inside the wires.

tion of complex 3D surfaces.

222 Nanowires - Recent Advances

position.

The field of nanowire based solar cell research develops very fast and thus new methods and materials are coming up continuously. Our group is driving this type of research for‐ ward very actively and we are even constantly improving our own records. This outlook will give the reader some ideas about the progress of NW cell research in our lab. By further improving nanowire based cell concepts, we are aiming to achieve the goal of seeing these cells on top of a roof one day.

#### **4.1. Silicon nanowire arrays formed by reactive ion etching**

Reactive ion etching (RIE) in combination with nanosphere lithography (NSL) represents a new way of forming hexagonal arrays of SiNWs with controllable geometries in wafers [22] or even multicrystalline silicon (mc-Si) thin films on glass. [44] These structures offer several advantages regarding further systematic research on SiNW based optoelectronic and solar cell devices. Since device constituting properties like spectral absorption, reflection and the area and distribution of charge separating surfaces (p/n-junctions, heterojunctions etc.) are dependent on SiNW array geometries, a control of these geometries creates the opportunity for systematic device development. Figure 14 shows an example of the potential of this highly flexible and scalable process which results in large areas of nanostructured surfaces. The nanospheres arrange in self-aligned hexagonal colloidal 2D crystals on the wafer sur‐ face. The initial sphere diameter determines the lattice constant of the structure, whereas oxygen-plasma reduced spheres pre-determine the diameter of the SiNWs.

**Figure 14.** Left: RIE of a NSL patterned wafer surface. The nanospheres act as a shadow mask for the RIE process which results in aligned arrays of SiNWs. Inset: The lattice constant of the hexagonal arrays is pre-determined by the initial sphere diameter di , the diameter of the structures dr can be controlled by the reduction of the initial spheres. Right: With the described method, the structuring of large surfaces is possible as it can be seen in the SEM micro‐ graph.

RIE etched SiNWs may not offer the high throughput and cheap production of WCE NWs, but the predefined non-random structure offers many advantages in the control of optical and electrical properties in NW SCs compared to the random structures of chemically etch‐ ed NWs and thus permit tailoring and optimizing SC efficiencies. Moreover, the etched SiNWs usually have a smaller surface to volume ratio, thus permitting smaller surfaces that need passivation [41].

#### **4.2. Novel electrodes based on silver nanowires or graphene**

One-dimensional (1D) metal nanostructures, especially single crystalline nanowires, have at‐ tracted intensive interest because of their great potential for practical applications in nano‐ scale electronics, optoelectronics, and molecular sensing devices. However, one of the most significant challenges still facing device fabrication and processing is the large scale repro‐ ducibility of the nanowires into highly controlled arrays with high uniformity. Herein, we demonstrate an extremely simple method for large-scale and highly ordered silver nanowire (AgNW) monolayer synthesis via a thermal assembling process. These AgNWs offer new possibilities in the formation of transparent and highly conductive contacts.

The AgNW were synthesized by wet chemistry. For a typical synthesis, 5 ml of ethylene-gly‐ col (EG) was heated to 175±10 °C for 1 h under magnetic stirring. Subsequently, Cu-additive solution (40 μL, 4 mM CuCl2) was injected together with the polymer PVP (1.5 ml with a concentration of 0.147 M) into the heated EG. The volume, systematically influences can de‐ sign the final shape and size of the AgNW. Finally, 1.5 ml of a 0.094 M AgNO3 was injected slowly (for 10min). After 1 to 1.5 h, the reaction was stopped and cooled to room tempera‐ ture by a water bath. Products were then washed with ethanol three times.

SEM images with low and high magnification of Ag nanowires (AgNW) are presented in Figure 15. The AgNW had a typical length of 30±10 μm, with diameter of 50±10 nm.

**Figure 15.** SEM micrographs of AgNW prepared by thermal assembly in the presence of PVP polymer. (a) A low-mag‐ nification SEM image reveals the highly uniform shape of AgNWs, (b) A higher magnification SEM image shows the high aspect ratio of AgNW. Inset is a SEM image that illustrates the pentagonal cross-section of the AgNWs.

Graphene with its outstanding optical and electrical properties offers many advantages when applied as contact material to SiNW SCs. The high in-plane conductivity as well as the high optical transparency recommends graphene as a new type of TCMs (transparent conductive materials) and it might substitute or support the typical TCO in our SiNW SC concept. It can be easily transferred to desired surfaces and its flexibility offers advantag‐ es for flexible SCs (e.g. on plastic substrates or metal foils). Figure 16 shows a transferred layer of graphene on top of RIE fabricated SiNWs. Preliminary investigations strongly promise excellent contact properties and thus research on this material will be forced with a lot of effort.

**Figure 16.** A graphene monolayer was deposited on top of SiNWs that are created by RIE to act as front contact. In the upper part the contact needle of the EBIC module can be seen.

## **5. Conclusions and outlook**

ed NWs and thus permit tailoring and optimizing SC efficiencies. Moreover, the etched SiNWs usually have a smaller surface to volume ratio, thus permitting smaller surfaces that

One-dimensional (1D) metal nanostructures, especially single crystalline nanowires, have at‐ tracted intensive interest because of their great potential for practical applications in nano‐ scale electronics, optoelectronics, and molecular sensing devices. However, one of the most significant challenges still facing device fabrication and processing is the large scale repro‐ ducibility of the nanowires into highly controlled arrays with high uniformity. Herein, we demonstrate an extremely simple method for large-scale and highly ordered silver nanowire (AgNW) monolayer synthesis via a thermal assembling process. These AgNWs offer new

The AgNW were synthesized by wet chemistry. For a typical synthesis, 5 ml of ethylene-gly‐ col (EG) was heated to 175±10 °C for 1 h under magnetic stirring. Subsequently, Cu-additive solution (40 μL, 4 mM CuCl2) was injected together with the polymer PVP (1.5 ml with a concentration of 0.147 M) into the heated EG. The volume, systematically influences can de‐ sign the final shape and size of the AgNW. Finally, 1.5 ml of a 0.094 M AgNO3 was injected slowly (for 10min). After 1 to 1.5 h, the reaction was stopped and cooled to room tempera‐

SEM images with low and high magnification of Ag nanowires (AgNW) are presented in

**Figure 15.** SEM micrographs of AgNW prepared by thermal assembly in the presence of PVP polymer. (a) A low-mag‐ nification SEM image reveals the highly uniform shape of AgNWs, (b) A higher magnification SEM image shows the

Graphene with its outstanding optical and electrical properties offers many advantages when applied as contact material to SiNW SCs. The high in-plane conductivity as well as the high optical transparency recommends graphene as a new type of TCMs (transparent conductive materials) and it might substitute or support the typical TCO in our SiNW SC concept. It can be easily transferred to desired surfaces and its flexibility offers advantag‐

high aspect ratio of AgNW. Inset is a SEM image that illustrates the pentagonal cross-section of the AgNWs.

Figure 15. The AgNW had a typical length of 30±10 μm, with diameter of 50±10 nm.

**4.2. Novel electrodes based on silver nanowires or graphene**

possibilities in the formation of transparent and highly conductive contacts.

ture by a water bath. Products were then washed with ethanol three times.

need passivation [41].

224 Nanowires - Recent Advances

In the last years not only have we developed and realized a novel nanowire based solar cell concept, but we have also developed advanced analysis methods for such kind of nano‐ structured devices. By systematically using cheap and abundant materials and avoiding ex‐ pensive vacuum equipment for the nanostructuring process, we generated solar cells with over 9% conversion efficiency. As these values are the results of early stage research, further improvement up to an efficiency of 12-15% are expected. Furthermore, the concept will be transferred to thin films like multicrystalline silicon on glass or flexible substrates.

The development of 3D-EBIC will be most likely of great influence for the analysis of 3-dimen‐ sional nanostructured solar cells and thus it is also a great advance for the PV research field.

## **Acknowledgements**

The authors gratefully acknowledge Florian Talkenberg for the ALD processes, Arne Boch‐ mann for the LBIC measurements and Maria Hadjipanayi for the EQE investigations. Many thanks go to Gerald Brönstrup and Felix Voigt for fruitful discussions.

The authors gratefully acknowledge financial support by the European Commission projects within the framework of FP7, "ROD-SOL" and "Fiblys".

## **Author details**

Björn Hoffmann1,2, Vladimir Sivakov2 , Sebastian W. Schmitt1 , Muhammad Y. Bashouti1 , Michael Latzel1 , Jiří Dluhoš<sup>3</sup> , Jaroslav Jiruse3 and Silke Christiansen1,2

1 Max Planck Institute for the Science of Light, TDSU Photonic Nanostructures, Erlangen, Germany

2 Institute of Photonic Technology, Department Semiconductor Nanostructures, Jena, Germany

3 TESCAN, a.s., Brno, Czech Republic

## **References**


[9] Sivakov V, Andrä G, Gawlik A, Berger A, Plentz J, Falk F, et al. (2009) Silicon nano‐ wire-based solar cells on glass: synthesis, optical properties, and cell parameters. Nano Lett. 9(4),1549–54. http://dx.doi.org/10.1021/nl803641f

**Author details**

226 Nanowires - Recent Advances

Michael Latzel1

Germany

**References**

Björn Hoffmann1,2, Vladimir Sivakov2

3 TESCAN, a.s., Brno, Czech Republic

S092702481000214X

, Jiří Dluhoš<sup>3</sup>

, Sebastian W. Schmitt1

1 Max Planck Institute for the Science of Light, TDSU Photonic Nanostructures, Erlangen,

2 Institute of Photonic Technology, Department Semiconductor Nanostructures, Jena, Germany

[1] Green MA. (2001) Third generation photovoltaics: Ultra-high conversion efficiency at low cost. Prog Photovolt Res Appl. 9(2),123-35. http://doi.wiley.com/10.1002/pip.360

[2] Kaufmann CA, Neisser A, Klenk R, Scheer R. (2005) Transfer of Cu(In,Ga)Se2 thin film solar cells to flexible substrates using an in situ process control. Thin Solid Films.

[3] Green MA, Emery K, Hishikawa Y, Warta W, Dunlop ED. (2012) Solar cell efficiency tables (version 39). Prog Photovolt Res Appl. 20(1),12–20. http://onlineli‐

[4] Bagnall DM, Boreland M. (2008) Photovoltaic technologies. Energy Policy. 36(12),

[5] Du QG, Kam CH, Demir HV, Yu HY, Sun XW. (2011) Broadband absorption en‐ hancement in randomly positioned silicon nanowire arrays for solar cell applications.

[6] Tsakalakos L, Balch J, Fronheiser J, Shih M-Y, LeBoeuf SF, Pietrzykowski M, et al. (2007) Strong broadband optical absorption in silicon nanowire films. J Nanophoton‐

[7] Srivastava SK, Kumar D, Singh PK, Kar M, Kumar V, Husain M. (2010) Excellent an‐ tireflection properties of vertical silicon nanowire arrays. Sol. Energy Mater. Sol. Cells. 94(9),1506–11. http://linkinghub.elsevier.com/retrieve/pii/S0927024810000863

[8] Srivastava SKSK, Kumar D, Singh PKP, Kar M, Kumar V, Husain M. (2011) Fabrica‐ tion of silicon nanowire arrays based solar cell with improved performance. Sol. En‐ ergy Mater. Sol. Cells. 95(1),215–8. http://linkinghub.elsevier.com/retrieve/pii/

ics. 1(1),013552. http://link.aip.org/link/JNOACQ/v1/i1/p013552/s1&Agg=doi

4390–6. http://linkinghub.elsevier.com/retrieve/pii/S0301421508004552

Opt Lett. 36(10),1884–6. http://www.ncbi.nlm.nih.gov/pubmed/21593923

and Silke Christiansen1,2

, Jaroslav Jiruse3

480-481,515-9. http://dx.doi.org/10.1016/j.tsf.2004.11.067

brary.wiley.com/doi/10.1002/pip.2163/full

, Muhammad Y. Bashouti1

,


atomic layer deposition. Phys. Status Solidi RRL. 4(1-2),10–2. http://doi.wiley.com/ 10.1002/pssr.200903334

[34] Profijt HB, Potts SE, van de Sanden MCM, Kessels WMM. (2011) Plasma-Assisted Atomic Layer Deposition: Basics, Opportunities, and Challenges. J Vac Sci Technol A. 29(5),050801. http://link.aip.org/link/JVTAD6/v29/i5/p050801/s1&Agg=doi

[21] Huynh WU, Dittmer JJ, Alivisatos AP. (2002) Hybrid nanorod-polymer solar cells. Science (80- ). 295(5564),2425–7. http://www.sciencemag.org/content/

[22] Garnett E, Yang P. (2010) Light trapping in silicon nanowire solar cells. Nano Lett.

[23] Garnett EC, Yang P. (2008) Silicon nanowire radial p-n junction solar cells. J. Am. Chem. Soc. 130(29),9224–5. http://www.ncbi.nlm.nih.gov/pubmed/18576622

[24] Ke Y, Wang X, Weng XJ, Kendrick CE, Yu YA, Eichfeld SM, et al. (2011) Single wire radial junction photovoltaic devices fabricated using aluminum catalyzed silicon nanowires. Nanotechnology. 22(44),445401. http://www.ncbi.nlm.nih.gov/pubmed/

[25] Hoffmann S, Bauer J, Ronning C, Stelzner T, Michler J, Ballif C, et al. (2009) Axial p-n junctions realized in silicon nanowires by ion implantation. Nano Lett. 9(4),1341–4.

[26] Wen C-Y, Reuter MC, Bruley J, Tersoff J, Kodambaka S, Stach EA, et al. (2009) Forma‐ tion of compositionally abrupt axial heterojunctions in silicon-germanium nano‐ wires. Science (80- ). 326(5957),1247–50. http://www.ncbi.nlm.nih.gov/pubmed/

[27] Tsai S-H, Chang H-C, Wang H-H, Chen S-Y, Lin C-A, Chen S-A, et al. (2011) Signifi‐ cant Efficiency Enhancement of Hybrid Solar Cells Using Core-Shell Nanowire Ge‐ ometry for Energy Harvesting. ACS Nano., 5(12), 9501-10. http://

[28] Shewchun J, Dubow J, Myszkowski A, Singh R. (1977) The operation of the semicon‐ ductor-insulator-semiconductor (SIS) solar cell: Theory. J Appl Phys. 49(2),855–64.

[29] Shewchun J, Dubow J, Wilmsen C, Singh R, Burk D, Wager J. (1978) The operation of the semiconductor-insulator-semiconductor solar cell: Experiment. J Appl Phys.

[30] Genis AP, Smith PA, Emery K, Singh R, DuBow JB. (1980) Efficient indium tin oxide/ polycrystalline silicon semiconductor-insulator-semiconductor solar cells. Appl Phys

[31] Ritala M, Leskela M. (2001) Atomic layer deposition. Handbook of thin film materi‐ als. 1,103–59. http://cambridgenano.com/papers/atomic layer deposition ritala leske‐

[32] Leskelä M, Ritala M. (2002) Atomic layer deposition (ALD): from precursors to thin film structures. Thin Solid Films. 409(1),138–46. http://linkinghub.elsevier.com/

[33] Dingemans G, Seguin R, Engelhart P, Sanden MCMVD, Kessels WMM. (2010) Silicon surface passivation by ultrathin Al 2 O 3 films synthesized by thermal and plasma

50(4),2832–9. http://ieeexplore.ieee.org/xpls/abs\_all.jsp?arnumber=5105997

Lett. 37(1),77. http://link.aip.org/link/APPLAB/v37/i1/p77/s1&Agg=doi

http://ieeexplore.ieee.org/xpls/abs\_all.jsp?arnumber=5104464

10(3),1082–7. http://www.ncbi.nlm.nih.gov/pubmed/20108969

http://www.ncbi.nlm.nih.gov/pubmed/19256535

pubs.acs.org/doi/abs/10.1021/nn202485m

295/5564/2425.abstract

21983364

228 Nanowires - Recent Advances

19965471

la.pdf

retrieve/pii/S0040609002001177


## **Thermal Transport in Semiconductor Nanowires**

J. Anaya, J. Jiménez and T. Rodríguez

Additional information is available at the end of the chapter

http://dx.doi.org/10. 5772/52588

## **1. Introduction**

[45] Bajaj J, Tennant WE. (1990) Remote contact LBIC imaging of defects in semiconduc‐ tors. J Cryst Growth. 103(1-4),170–8. http://www.sciencedirect.com/science/

[46] Donolato C. (1983) Theory of beam induced current characterization of grain boun‐ daries in polycrystalline solar cells. J Appl Phys. 54(3),1314. http://link.aip.org/link/

article/pii/002202489090186O

230 Nanowires - Recent Advances

JAPIAU/v54/i3/p1314/s1&Agg=doi

Since the late 1990's semiconductor NWs research is in expansion because of the huge num‐ ber of potential applications of these one dimensional structures. This research effort is giv‐ en expression as a vast number of research articles. Semiconductor NWs are the building blocks of advanced nanodevices; e. g. field effect transistors, interconnects, heterostructured devices, as vertically oriented nanoscale laser diodes and HBTs, thermoelectric converters, photovoltaic devices, and sensors [1-5].

The thermal transport in NWs is receiving a great deal of attention, because it constitutes a critical issue for the performance and reliability of NWs based devices. Because of the size reduction a large power density needs to be dissipated, especially in devices operating at high current and power densities; which induces a non negligible Joule heating; therefore, the control of the thermal conductivity is crucial to the thermal extraction from the active parts of these devices [6]. On the other hand, bulk semiconductors are usually good thermal and electrical conductors, which limits their potential performance in thermoelectric devi‐ ces; however, it was demonstrated that surface engineered semiconductor NWs behave es‐ sentially as thermal insulators [2], being at the same time good electrical conductors, this means that semiconductor NWs can have a large themoelectric figure of merit [2,7-9]; which can make them very good thermoelectric materials for waste heat recovery and power gen‐ eration over a large temperature range.

The experimental measurement of the thermal conductivity at the nanoscale is very challeng‐ ing [11-13], with very few available contrasted measurements. In this context of lack of experi‐ mental results, the development of confident models for describing the thermal conductivity of semiconductor NWs is very important in order to design the thermal management of devices, or the thermoelectric power conversion. All this has created the need for a great modeling ef‐ fort permitting the design of nanodevices in which the thermal transport plays a capital role.

© 2012 Anaya et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Anaya et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We present herein a review of the thermal transport in semiconductor NWs. The main ex‐ perimental achievements are presented, both smooth and surface engineered NWs, also al‐ loyed NWs are considered; these data evidence the relevance of the thermal transport problem in these one dimensional structures. These experimental data, though scarce, per‐ mit a certain degree of contrast of the theoretical models. According to this we present a crit‐ ical overview of the main theoretical approaches to the problem of the thermal conductivity of semiconductor NWs following a historic sequence. Along this way, we review a large number of models, among others the classical approaches based on the Callaway formalism, atomistic models based on molecular dynamics simulations, or the hydrodynamic ap‐ proaches. One revises the contribution of the different resistive phonon scattering processes. Also, the role of normal scattering processes is discussed, as well as the recent consideration of the scattering by optic phonons. Special attention is paid to the boundary scattering which is the main cause for the thermal conductivity suppression in semiconductor NWs. The role of engineered surfaces is also discussed.

The strengths and weakness of the main models are discussed. Finally, we introduce a predic‐ tive approach extended to the full temperature range, based on the Callaway-Holland formal‐ ism, which permits an excellent fitting to the experimental data available. Using this approach empiric relations are deduced for the case of Si NWs, which permit to extract the thermal con‐ ductivity for a wide range of diameters and temperatures, which presents an unquestionable interest for practical purposes related to the nanodevice design and thermal managing.

## **2. Experimental approach to the thermal transport in semiconductor nanowires**

The measurement of the thermal conductivity at the nanoscale is a challenging experiment. The first measurements of thermal transport of semiconductor NWs were carried out by D. Li et al. [11,13], they measured the thermal conductivity of a few individual Si NWs grown by the vapour liquid solid (VLS) method. The experiments were performed in a microdevice specially designed for measuring the thermal properties of one-dimensional structures [10]. This device, called thermal bridge, consists of a nanowire suspended in between two plates of platinum, one of them acting as the heater, and the other as the thermometer. To guaran‐ tee accurate measurements with this device, the junction between the NW and the platinum plates must have an ideally null thermal resistance [10,12].

The results obtained by D. Li et al. [11] showed a strong suppression of the thermal conduc‐ tivity of the Si NWs as compared with bulk Si (Figure 1-A). Furthermore, a dependence of the thermal conductivity with the NWs diameter was revealed; the smaller the NW diame‐ ter the lower the thermal conductivity. On the other hand, the experimental data suggest a change in the temperature dependence of the thermal conductivity of Si NWs for diameters below 25 nm; the experimental data give a linear dependence with T for the thermal conduc‐ tivity of a NW with 22 nm diameter, in which phonon confinement cannot be neglected [15], with respect to the dependence observed for larger diameter NWs, which follow a behavior similar to the bulk(Figure 1-B).

We present herein a review of the thermal transport in semiconductor NWs. The main ex‐ perimental achievements are presented, both smooth and surface engineered NWs, also al‐ loyed NWs are considered; these data evidence the relevance of the thermal transport problem in these one dimensional structures. These experimental data, though scarce, per‐ mit a certain degree of contrast of the theoretical models. According to this we present a crit‐ ical overview of the main theoretical approaches to the problem of the thermal conductivity of semiconductor NWs following a historic sequence. Along this way, we review a large number of models, among others the classical approaches based on the Callaway formalism, atomistic models based on molecular dynamics simulations, or the hydrodynamic ap‐ proaches. One revises the contribution of the different resistive phonon scattering processes. Also, the role of normal scattering processes is discussed, as well as the recent consideration of the scattering by optic phonons. Special attention is paid to the boundary scattering which is the main cause for the thermal conductivity suppression in semiconductor NWs.

The strengths and weakness of the main models are discussed. Finally, we introduce a predic‐ tive approach extended to the full temperature range, based on the Callaway-Holland formal‐ ism, which permits an excellent fitting to the experimental data available. Using this approach empiric relations are deduced for the case of Si NWs, which permit to extract the thermal con‐ ductivity for a wide range of diameters and temperatures, which presents an unquestionable

interest for practical purposes related to the nanodevice design and thermal managing.

**2. Experimental approach to the thermal transport in semiconductor**

The measurement of the thermal conductivity at the nanoscale is a challenging experiment. The first measurements of thermal transport of semiconductor NWs were carried out by D. Li et al. [11,13], they measured the thermal conductivity of a few individual Si NWs grown by the vapour liquid solid (VLS) method. The experiments were performed in a microdevice specially designed for measuring the thermal properties of one-dimensional structures [10]. This device, called thermal bridge, consists of a nanowire suspended in between two plates of platinum, one of them acting as the heater, and the other as the thermometer. To guaran‐ tee accurate measurements with this device, the junction between the NW and the platinum

The results obtained by D. Li et al. [11] showed a strong suppression of the thermal conduc‐ tivity of the Si NWs as compared with bulk Si (Figure 1-A). Furthermore, a dependence of the thermal conductivity with the NWs diameter was revealed; the smaller the NW diame‐ ter the lower the thermal conductivity. On the other hand, the experimental data suggest a change in the temperature dependence of the thermal conductivity of Si NWs for diameters below 25 nm; the experimental data give a linear dependence with T for the thermal conduc‐ tivity of a NW with 22 nm diameter, in which phonon confinement cannot be neglected [15], with respect to the dependence observed for larger diameter NWs, which follow a behavior

The role of engineered surfaces is also discussed.

plates must have an ideally null thermal resistance [10,12].

similar to the bulk(Figure 1-B).

**nanowires**

232 Nanowires - Recent Advances

**Figure 1.** Experimental thermal conductivities as a function of temperature for bulk Si [14], and NWs with different diameters [11] (A), thermal conductivity of NWs in an expanded scale (B) data from [11].

Using the thermal bridge device developed by S. Li et al., the thermal conductivity of other semiconductor NWs has been measured [13,16-20]. All the NWs measured, independently of their nature display a dramatic reduction of the thermal conductivity with the diameter, clearly revealing that the thermal conductivity of NWs is limited by the NWs dimensions. M. C. Wingert et al. carried out thermal conductivity measurements on Ge NWs [20], show‐ ing that below a critical diameter of the NW, the thermal conductivity follows a linear de‐ pendence with temperature (Figure 2), in agreement with the behavior of the 22 nm Si NW measured in [11,13]. Therefore, this phenomenon can be considered as an intrinsic property of the NWs with reduced dimension, suggesting changes in the phonon transport when phonon confinement effects start to be relevant.

**Figure 2.** Thermal conductivity of two Ge NWs with diameters below the phonon confinement threshold, showing a linear dependence with temperature [20]

The thermal conductivity of NWs is not only suppressed by the reduced diameter, but, A. Hochbaum et al. showed that the surface corrugation of the NWs is a major issue for the re‐ duction of the thermal conductivity [2]. Using a similar microdevice to the one described in [10], they measured the thermal conductivity of Si NWs fabricated by an aqueous electroless etching method. These NWs present rough surfaces; in particular, the mean roughness height was estimated near 5 nm, with a lateral roughness correlation length of a few nano‐ meters [2]. The experimental thermal conductivities of two NWs of 115 nm diameter and comparable length, but different surface roughness are represented in Figure 3, showing a five times reduction in the thermal conductivity for the rough NW. Furthermore, in NWs with engineered surfaces, one can expect that the thermal conductivity could be reduced down to almost the amorphous limit [2].

**Figure 3.** Measured thermal conductivities of two Si NWs of 115 nm diameter, one of them smooth [11] and the other rough [2].

More recently, Y. Park et al. [21] carried out an experiment similar to that performed in [2], but using rough Si NW grown by an VLS method, by which they grew Si NWs with a mean roughness height close to 10 nm, even greater than the NWs used in [2]. However, the ther‐ mal conductivity was not so severely suppressed as it did for the electroless etched NWs prepared in [2]. This illustrates that the reduction of the thermal conductivity is not only propitiated by the roughness height, but as J. Lim et al. have recently shown it is the combi‐ nation of the roughness height and its lateral correlation length, defined as the mean dis‐ tance between surface hills/depressions, that limits the thermal conductivity in rough NWs [22]. They measured the thermal conductivity of several Si NWs with different surface mor‐ phologies, showing that the distance between surface hills/depressions plays a crucial role in the decrease of the thermal conductivity. The SEM images of the NWs measured in [21] compared to the SEM images of the NWs measured in [2], clearly show that the distance be‐ tween hills/depressions of the NWs of reference [21] are significantly longer than for the NWs of reference [2], which should account for the lower thermal conductivity reduction observed by Lim [22] with respect to the strong reduction reported by Hoschbaum, despite the higher value of the mean roughness height measured in [21].

H. Kim et al. showed the effect of composition on the thermal conductivity of semiconductor NWs. They measured the thermal conductivity of alloyed Si1-xGex NWs with a small Ge frac‐ tion [23]. Even for very diluted alloys the thermal conductivity of the NWs showed a dramatic reduction compared to pure Si NWs. Thereafter, H. Kim et al. grew rough Si1-xGex NWs and showed the combined effect of the alloy composition and the surface roughness [24], which permits additional reductions of the thermal conductivity; thus, opening almost unlimited possibilities for designing nanodevices by its thermal conductivity, which is very attractive for thermoelectric devices. These results are summarized in Figure 4, where the experimental val‐ ues for a smooth Si NW, and a smooth Si1-xGex NW of comparable diameter and a very low Ge fraction (x=0. 004) are shown. The effect of the roughness is summarized in two NWs of the same composition and diameter but differing only by the surface roughness.

**Figure 4.** Thermal conductivities of a smooth Si NW from [11] (circles), and alloyed SiGe NWs, smooth NWs data from [23] (triangles) and rough NWs data from [24] (diamonds).

In a recent work, Y. Choongho reported additional thermal conductivity data of Si1-xGex NWs for several compositions [18]; evidencing the relevant role played by the composition, but one cannot be conclusive about the composition bowing, because of the diameter dispersion.

## **3. Theoretical models**

The thermal conductivity of NWs is not only suppressed by the reduced diameter, but, A. Hochbaum et al. showed that the surface corrugation of the NWs is a major issue for the re‐ duction of the thermal conductivity [2]. Using a similar microdevice to the one described in [10], they measured the thermal conductivity of Si NWs fabricated by an aqueous electroless etching method. These NWs present rough surfaces; in particular, the mean roughness height was estimated near 5 nm, with a lateral roughness correlation length of a few nano‐ meters [2]. The experimental thermal conductivities of two NWs of 115 nm diameter and comparable length, but different surface roughness are represented in Figure 3, showing a five times reduction in the thermal conductivity for the rough NW. Furthermore, in NWs with engineered surfaces, one can expect that the thermal conductivity could be reduced

**Figure 3.** Measured thermal conductivities of two Si NWs of 115 nm diameter, one of them smooth [11] and the other

More recently, Y. Park et al. [21] carried out an experiment similar to that performed in [2], but using rough Si NW grown by an VLS method, by which they grew Si NWs with a mean roughness height close to 10 nm, even greater than the NWs used in [2]. However, the ther‐ mal conductivity was not so severely suppressed as it did for the electroless etched NWs prepared in [2]. This illustrates that the reduction of the thermal conductivity is not only propitiated by the roughness height, but as J. Lim et al. have recently shown it is the combi‐ nation of the roughness height and its lateral correlation length, defined as the mean dis‐ tance between surface hills/depressions, that limits the thermal conductivity in rough NWs [22]. They measured the thermal conductivity of several Si NWs with different surface mor‐ phologies, showing that the distance between surface hills/depressions plays a crucial role in the decrease of the thermal conductivity. The SEM images of the NWs measured in [21] compared to the SEM images of the NWs measured in [2], clearly show that the distance be‐ tween hills/depressions of the NWs of reference [21] are significantly longer than for the NWs of reference [2], which should account for the lower thermal conductivity reduction observed by Lim [22] with respect to the strong reduction reported by Hoschbaum, despite

the higher value of the mean roughness height measured in [21].

down to almost the amorphous limit [2].

234 Nanowires - Recent Advances

rough [2].

Before the first experimental measurement of the thermal conductivity in a semiconductor NW, a few theoretical approaches already pointed to a thermal conductivity decrease in low dimensional structures. Five years early with respect to the first experimental measurements of NW thermal conductivity, A. Balandin et al. in a pioneering work on the effect of low di‐ mensionality on the thermal transport, presented a theoretical analysis of the lattice thermal conductivity of a semiconductor quantum well [25]. This analysis was built up on the basis that the spatial confinement involves a reduction of the phonon group velocities, therefore, of the phonon relaxation rate. This thermal transport reduction makes evident the increased relevance of the resistive phonon processes in reduced dimension structures. They calculat‐ ed the thermal conductivity of Si quantum well [25] making use of a reduced Callaway for‐ malism [26], in which only considered the resistive processes given by the Umpklapp scattering, the boundary scattering, and the processes related to the impurity scattering. A detailed description of the Callaway model and the different scattering mechanisms is ad‐ dressed in paragraph 4. 2. The model proposed in [25] revealed that the thermal conductivi‐ ty of a Si nanostructure must be substantially reduced with respect to its bulk counterpart. Shortly after, Zou and Balandin applied the same formalism to the study of the thermal con‐ ductivity of Si NWs [27]. They modified the simple model described on [25] under the basis of a phonon redistribution by the NW boundaries, which implies a modification of the re‐ duced Callaway expression used in the previous work. On the other hand, the thermal con‐ ductivity dependence with the surface roughness of the NWs was also added. The scattering processes considered were a modification of the Umpklapp scattering, the electron-phonon scattering, and a modified boundary scattering taking account of the effect of roughness, ex‐ pressed in terms of the fraction of specularly to diffusively reflected phonons at the NW boundary [28]. With those modifications, the thermal conductivity of a Si NW with 20 nm diameter was calculated, showing a great dependence of the thermal conductivity reduction with the surface characteristics. Nevertheless, even in the case of the maximum surface con‐ tribution, the thermal conductivity calculated was higher, 14 W/mK at 300 K for 20 nm di‐ ameter, near twice, than the experimental value reported for a 22 nm diameter Si NW [11,13]. In a different approach, A. Voltz et al. made use of molecular dynamics simulations in thin Si NWs with diameters below 5 nm [29], reaching the same result of a dramatic re‐ duction of the thermal conductivity.

The first experimental data about the thermal conductivity of Si NWs opened the possibility to contrast the theoretical models [11,13], allowing the possibility of developing predictive tools regarding the thermal conductivity of NWs. As mentioned above, the experimental measurements evidenced a strong dependence with the NWs morphology, including diame‐ ter and surface topography; as a matter of fact, the construction of a model capable of pre‐ dicting the thermal conductivity of semiconductor NWS becomes a necessary tool for the design of nanodevices in which the thermal transport becomes a critical issue for perform‐ ance and reliability. N.Mingo build up a model based on a modification of the ballistic pho‐ non transport in a suspended NW [30]. The model takes account of the scattering events that undergo the phonons in a real semiconductor, and requires the calculation of the complete phonon dispersion relations in the NW. The phonon scattering considered by Mingo, as it was the case for the Balandin models [25,27], are the resistive processes given by Umpklapp, boundary, and impurity scattering mechanisms. At difference of the models reported on ref‐ erences [25,27], where exact expressions for the Umpklapp scattering were used, Mingo in‐ troduced two parameters which the values are obtained from the adjustment of their model to the thermal conductivity of bulk Si. [30] These parameters are fixed for the calculation of the thermal conductivity of the NWs; therefore, it can be considered as a free of parameters model. Unfortunately, this model underestimates the thermal conductivity value with re‐ spect to the experimental data, requiring of an additional adjustment parameter concerning the boundary scattering; this parameter is defined as an effective diameter, greater than the diameter of the NWS experimentally measured in the SEM [30]. Thereby, using effective di‐ ameters up to 30% higher than the experimentally measured ones, Mingo obtained a satis‐ factory fitting to the experimental values reported for NWs with diameters above the limit for phonon confinement. [30,31] On the other hand, Mingo showed that a reduced Callaway model as the one developed by Balandin et al. [25,27] can give a good fitting to the experi‐ mental values if one lowers the cut-off temperature with respect to the bulk [30].

relevance of the resistive phonon processes in reduced dimension structures. They calculat‐ ed the thermal conductivity of Si quantum well [25] making use of a reduced Callaway for‐ malism [26], in which only considered the resistive processes given by the Umpklapp scattering, the boundary scattering, and the processes related to the impurity scattering. A detailed description of the Callaway model and the different scattering mechanisms is ad‐ dressed in paragraph 4. 2. The model proposed in [25] revealed that the thermal conductivi‐ ty of a Si nanostructure must be substantially reduced with respect to its bulk counterpart. Shortly after, Zou and Balandin applied the same formalism to the study of the thermal con‐ ductivity of Si NWs [27]. They modified the simple model described on [25] under the basis of a phonon redistribution by the NW boundaries, which implies a modification of the re‐ duced Callaway expression used in the previous work. On the other hand, the thermal con‐ ductivity dependence with the surface roughness of the NWs was also added. The scattering processes considered were a modification of the Umpklapp scattering, the electron-phonon scattering, and a modified boundary scattering taking account of the effect of roughness, ex‐ pressed in terms of the fraction of specularly to diffusively reflected phonons at the NW boundary [28]. With those modifications, the thermal conductivity of a Si NW with 20 nm diameter was calculated, showing a great dependence of the thermal conductivity reduction with the surface characteristics. Nevertheless, even in the case of the maximum surface con‐ tribution, the thermal conductivity calculated was higher, 14 W/mK at 300 K for 20 nm di‐ ameter, near twice, than the experimental value reported for a 22 nm diameter Si NW [11,13]. In a different approach, A. Voltz et al. made use of molecular dynamics simulations in thin Si NWs with diameters below 5 nm [29], reaching the same result of a dramatic re‐

The first experimental data about the thermal conductivity of Si NWs opened the possibility to contrast the theoretical models [11,13], allowing the possibility of developing predictive tools regarding the thermal conductivity of NWs. As mentioned above, the experimental measurements evidenced a strong dependence with the NWs morphology, including diame‐ ter and surface topography; as a matter of fact, the construction of a model capable of pre‐ dicting the thermal conductivity of semiconductor NWS becomes a necessary tool for the design of nanodevices in which the thermal transport becomes a critical issue for perform‐ ance and reliability. N.Mingo build up a model based on a modification of the ballistic pho‐ non transport in a suspended NW [30]. The model takes account of the scattering events that undergo the phonons in a real semiconductor, and requires the calculation of the complete phonon dispersion relations in the NW. The phonon scattering considered by Mingo, as it was the case for the Balandin models [25,27], are the resistive processes given by Umpklapp, boundary, and impurity scattering mechanisms. At difference of the models reported on ref‐ erences [25,27], where exact expressions for the Umpklapp scattering were used, Mingo in‐ troduced two parameters which the values are obtained from the adjustment of their model to the thermal conductivity of bulk Si. [30] These parameters are fixed for the calculation of the thermal conductivity of the NWs; therefore, it can be considered as a free of parameters model. Unfortunately, this model underestimates the thermal conductivity value with re‐ spect to the experimental data, requiring of an additional adjustment parameter concerning the boundary scattering; this parameter is defined as an effective diameter, greater than the

duction of the thermal conductivity.

236 Nanowires - Recent Advances

Those models only deal with the resistive scattering processes; this means that the contribu‐ tion of the normal three-phonon events is neglected. However, it is well known that these processes contribute to the thermal transport resistance through the Herring mechanism [32]. Y. Chen developed a model based on Monte Carlo simulation considering normal and resistive processes [33]. Making use of a complex genetic algorithm and resorting to two free adjusting parameters, they showed a good fitting to the experimental data in the low tem‐ perature range (T<120K), but overestimated the thermal conductivity for higher tempera‐ tures. Once again the behavior exhibited by the 22 nm smooth Si NW was not reproduced by the model, which responded reasonably well above this limit.

Coming back to the first models developed by Zou et al. [27], M. Huang et al. carried out a more detailed treatment of the boundary scattering [34], presenting it as a rational function of the fraction of specularly reflected phonons at the boundary instead of the linear depend‐ ence originally used in [27]. Despite of being a model with a more realistic treatment of the boundary scattering the result was not satisfactory. On the other hand, they showed that not only the sound velocities decreased due the reduced diameter of the NWs, but the Debye temperatures had also be lower [34]; which, should account for the good agreement to the experimental data reached by Mingo when using cut-off temperatures lower than the bulk silicon cut-off temperatures in the Callaway model [30].

L. Liang et al. used a simple approach in which they established direct relations between the bulk semiconductor melting point and the nanowire melting point [35]; furthermore, they introduced an exponential expression for describing the boundary scattering due to surface roughness. This simple approach was able of predicting the thermal conductivity at 300 K for all the NWs measured in [11], even for the problematic 22 nm Silicon NW. The anoma‐ lous behavior of this thin NW, and thus of those NWs with diameter below the limit for phonon confinement,[11,20] was studied by P. Murphy et al. [36]. In contrast to previous works, they argued that in a strongly disordered surface not all the phonons are equally scattered, being these events dependent on the frequency of the incident phonons. Based on these assumptions, they developed a model splitting the behavior of the phonons in two dif‐ ferent contributions, determined by the magnitude of the component of their wave vectors in the direction perpendicular to the NW axis. The behavior is nearly ballistic for phonons with a small transverse wave vector component, while for those with a large transverse wavevector component the phonon scattering was assumed to be purely diffusive, with a mean free path of the order of the NW diameter. Under these assumptions, they showed that for the low temperature range, the calculated thermal conductivity approaches to a line‐ ar dependence with the temperature for decreasing NW diameters [36]; further showing that the change from the T3 law for bulk and thick NWs to the linear dependence for the thinner NWs occurs between 25 nm and 20 nm in diameter, as the experimental measure‐ ments demonstrate [11,20].

Z. Tian et al. suggested that the role of the optical phonons must be taken into account for nanostructures [37]. Making use of the phonon-phonon scattering description derived by A. Ward et al. [38], they carried out the calculation of the thermal conductivity contribution of each phonon branch, showing that the optical phonon contribution to the thermal conduc‐ tivity in NWs can represent near 20% of the total. Under the assumption of a diffusive re‐ gime for the boundary scattering, therefore in the Casimir limit [28], they calculated the thermal conductivity of smooth Si NWs in a good concordance with the experimental val‐ ues. In a previous work, M. Kazan et al. had already considered the optical phonon contri‐ bution [39]. They showed how the consideration of the optical phonon decay into acoustic phonons in the frame of the Callaway formalism can cause an improvement in the predic‐ tion of the thermal conductivity of bulk Ge. Subsequently M. Kazan et al. applied this meth‐ od to smooth Si NWs [40]. It should be noted, the Callaway formalism was conceived without the contribution of the optical modes [41-43]; although, it is implicitly taken into ac‐ count because of the increase of the cut-off temperatures necessary for fitting the experimen‐ tal data [44]. Therefore, in the modified Callaway model introduced in [40],which explicitly considers the optical contribution, the cut-off temperatures are lower than the cut-off tem‐ peratures used when applying the Callaway formalism without considering the optical pho‐ non contribution in a explicit way. Finally, working in the Casimir limit for boundary scattering and making use of a large number of adjustment parameters, they obtained a very good fitting to the thermal conductivity of Si NWs [40].

All those models, which give good fittings to the experimental values of the thermal conduc‐ tivity for Si NWs work in the diffusive limit for the boundary scattering; this means that one cannot further reduce the thermal conductivity due the boundary scattering when consider‐ ing rough surfaces. Thereby, when A. Hotchbaum et al. measured experimentally the ther‐ mal conductivity of rough Si NWs showing a great reduction in the thermal conductivity of rough NWs compared with the smooth Si NWs [2], none of those models could fit the exper‐ imental values.

F. Alvarez et al. showed that the thermal conductivity of semiconductor NWs can be descri‐ bed making use of an analogue hydrodynamic model for the phonon flux [45]. By solving the Guyer-Krumhansl equation for the phonon flux in a NW [46, 47], they reached a very simple expression for the thermal conductivity depending on the bulk thermal conductivity, the dimension of the NWs, and the fraction of specularly reflected phonons at the boundary [48]. In a further study, A.Sellito et al. showed that in a hydrodynamic framework, the roughness of the boundaries may cause a local inversion of the phononic flux, thus extra limiting the thermal conductivity of the NW [49]. This result is consistent with a previous work by A. Moore, who showed how the surface roughness of a NW may cause backscatter‐ ing events resulting in a more aggressive limitation of the thermal conductivity than the dif‐ fusive limit considered in the previous models [50]. On this basis, A. Sellito et al. [51] returned to the hydrodynamic model described in [48], which gave a simple expression for the thermal conductivity, and added the contribution of the backscattering events. With this modification, they attained a good fitting to the thermal conductivity of the smooth Silicon NWs below 150 K [51], besides, obtaining values for the thermal conductivity of the rough NWs close to those reported in reference [2] in the same temperature range.

thinner NWs occurs between 25 nm and 20 nm in diameter, as the experimental measure‐

Z. Tian et al. suggested that the role of the optical phonons must be taken into account for nanostructures [37]. Making use of the phonon-phonon scattering description derived by A. Ward et al. [38], they carried out the calculation of the thermal conductivity contribution of each phonon branch, showing that the optical phonon contribution to the thermal conduc‐ tivity in NWs can represent near 20% of the total. Under the assumption of a diffusive re‐ gime for the boundary scattering, therefore in the Casimir limit [28], they calculated the thermal conductivity of smooth Si NWs in a good concordance with the experimental val‐ ues. In a previous work, M. Kazan et al. had already considered the optical phonon contri‐ bution [39]. They showed how the consideration of the optical phonon decay into acoustic phonons in the frame of the Callaway formalism can cause an improvement in the predic‐ tion of the thermal conductivity of bulk Ge. Subsequently M. Kazan et al. applied this meth‐ od to smooth Si NWs [40]. It should be noted, the Callaway formalism was conceived without the contribution of the optical modes [41-43]; although, it is implicitly taken into ac‐ count because of the increase of the cut-off temperatures necessary for fitting the experimen‐ tal data [44]. Therefore, in the modified Callaway model introduced in [40],which explicitly considers the optical contribution, the cut-off temperatures are lower than the cut-off tem‐ peratures used when applying the Callaway formalism without considering the optical pho‐ non contribution in a explicit way. Finally, working in the Casimir limit for boundary scattering and making use of a large number of adjustment parameters, they obtained a very

All those models, which give good fittings to the experimental values of the thermal conduc‐ tivity for Si NWs work in the diffusive limit for the boundary scattering; this means that one cannot further reduce the thermal conductivity due the boundary scattering when consider‐ ing rough surfaces. Thereby, when A. Hotchbaum et al. measured experimentally the ther‐ mal conductivity of rough Si NWs showing a great reduction in the thermal conductivity of rough NWs compared with the smooth Si NWs [2], none of those models could fit the exper‐

F. Alvarez et al. showed that the thermal conductivity of semiconductor NWs can be descri‐ bed making use of an analogue hydrodynamic model for the phonon flux [45]. By solving the Guyer-Krumhansl equation for the phonon flux in a NW [46, 47], they reached a very simple expression for the thermal conductivity depending on the bulk thermal conductivity, the dimension of the NWs, and the fraction of specularly reflected phonons at the boundary [48]. In a further study, A.Sellito et al. showed that in a hydrodynamic framework, the roughness of the boundaries may cause a local inversion of the phononic flux, thus extra limiting the thermal conductivity of the NW [49]. This result is consistent with a previous work by A. Moore, who showed how the surface roughness of a NW may cause backscatter‐ ing events resulting in a more aggressive limitation of the thermal conductivity than the dif‐ fusive limit considered in the previous models [50]. On this basis, A. Sellito et al. [51] returned to the hydrodynamic model described in [48], which gave a simple expression for the thermal conductivity, and added the contribution of the backscattering events. With this

ments demonstrate [11,20].

238 Nanowires - Recent Advances

imental values.

good fitting to the thermal conductivity of Si NWs [40].

The effect of the surface roughness in the boundary scattering was studied by P. Martin et al. using perturbation theory [52]; the perturbation due to surface roughness was introduced in the Hamiltonian; also, the existence of a thin layer of native SiO2 layer surrounding the NW was considered. This model shows values close to the experimental measurements for smooth Si NW, but without a satisfactory fitting to the experimental data of reference [11]. More interesting is the capability shown by the model when one evaluates the thermal con‐ ductivity of rough NWs, insomuch as the values are close to those measured by Hochbaum et al. in [2]. On the other hand, they showed that the effect of the boundaries in the thermal conductivity reduction increases the more the smaller the diameter [52]; which suggests a dependence of the thermal conductivity with the diameter and the mean roughness height (*η*), following the relation *k* ∝(*D* / *η*)<sup>2</sup> .

Other approaches making use of molecular dynamics simulations intended to assess the ef‐ fect of the roughness in the phonon transport [53, 54], showing that the reduction of the thermal conductivity is strongly dependent on the morphology of the surface. On the other hand, D. Donadio using atomistic simulations [55], suggested that this reduction is due to the flattened phonon dispersion curves causing non-propagating phonon modes. After‐ wards, M. Luisier on the basis of the atomistic quantum transport corroborated these results [56]. Moreover, he suggested that there is no need of large mean roughness height to enter in the pure diffusive limit in which the mean free path (MFP) is comparable to the diameter of NW; thus, giving an explanation of why the models that work in this hypothesis for the smooth NWs measured in [11] were getting a good fitting.

J. Shadhu et al developed a model in which they introduced different frequency-dependent regimes for the scattering rates and a complex analysis of the roughness contribution [57]. They found that the correlation length plays a major role in the reduction of the thermal con‐ ductivity. Short correlation lengths leads to a larger number of phonons suffering diffusive surface scattering events; even more, a large roughness correlation length implies that only the phonons with long wavelengths, and thus phonons that do not significantly contribute to the thermal transport, are scattered by the surface. The correlation length also determines the incident angles for which the phonons suffer a scattering event. On the other hand, they indicate that the short wavelength modes are strongly attenuated in rough boundaries with short correlation lengths, resulting in a MFP for these phonons shorter than the NW diame‐ ter, which can be interpreted as a breakdown of the Casimir limit. With this complex model, they obtained a very good fitting to the experimental thermal conductivities for both rough and smooth Si NWs. Even more, the results achieved in this work are consistent with the recent work of J. Lin that demonstrated experimentally the importance of the correlation length together with the roughness mean height on the thermal conductivity reduction in semiconductor NWs [22].

The idea of a breakdown in the Casimir limit for rough NWs was previously pointed out, but without a physical explanation accounting for it. When C. Guthly et al. measured the thermal conductivity of rough gallium nitride NWs [17], they showed that a very good fit‐ ting to the experimental values can be achieved using the Callaway model with a modified boundary scattering. Taking into account that the diffusive limit, characterized by a phonon MFP equal to the diameter of the NW, cannot take account of the roughness effect, they used the MFP as an adjusting parameter. A very good fitting to the experimental values was achieved for values of MFP below the NW diameter, therefore, below the Casimir limit. The result of this paper, together with the confirmation of the violation of the Casimir limit for rough NWs [57], allow the use of simple models based on the Callaway formalism for calcu‐ lating the thermal conductivity of rough NWs, making use of a reduced effective diameter for the boundary scattering.

When considering alloyed NWs, e. g. Si1-xGex NWs, the thermal conductivity modeling must address the effect of the alloy scattering, besides all the above mentioned scattering terms. N. Yang et al. carried out a first approximation in NWs composed of a mixture of Si isotopes making use of a molecular dynamic simulation [58]. They showed a large reduction in ther‐ mal conductivity, due to the isotopic scattering, with a maximum reduction for a composi‐ tion of 50% between the constituent specimens. Afterwards, J. Chen et al. performed an equivalent molecular dynamics simulation on Si1-xGex NWs [59], showing a strong depend‐ ence of the thermal conductivity on the composition of the NW. Furthermore, they found that the thermal conductivity is severely reduced even for small fractions of Ge, what was later on experimentally demonstrated [23]. They estimated a reduction in the thermal con‐ ductivity of 50% with respect to pure Si NWs for a Ge fraction of only 5%; however, one ex‐ perimentally observes a much greater reduction, even for a smaller fraction of Ge in the NW [23]. In another attempt to modeling the thermal conductivity of Si1-xGex NWs, Z. Wang [60] suggested a model on the basis of the previous works made by N. Mingo for Si NWs [30]. They assumed that all the scattering relaxation times, except the impurity relaxation time, can be viewed as independent processes for Si and Ge; therefore, they were added according to the Mathisessen's rule weighted by the composition, while the impurity scattering related to the alloy, is assumed to depend of the composition by a quadratic bowing function. In this way, they obtained a simple model able to achieve a prediction of the thermal conduc‐ tivity of the order of magnitude of the experimental values obtained in [23]. Unfortunately, although they estimate thermal conductivity values of the order of magnitude of the experi‐ mental results, the fitting to the experimental results was not satisfactory, especially in the low temperature range.

Additional work is still needed to supply a good model accounting for the experimental thermal conductivity of Si1-xGex NWs, and in general of alloyed, or doped NWs. If one ob‐ serves the historical development of the models elaborated for describing the thermal con‐ ductivity of pure semiconductor NWs, one can see that the major advances were achieved once reliable experimental data were available. Very recently, new measurements of Si1-xGex NWs have been published by Y. Choongho in a wide range of compositions [18], suggesting that more new models concerning alloyed NWs will arrive in the next future.

## **4. A Predictive model for the thermal conductivity of NWs**

#### **4.1. Preliminary remarks**

thermal conductivity of rough gallium nitride NWs [17], they showed that a very good fit‐ ting to the experimental values can be achieved using the Callaway model with a modified boundary scattering. Taking into account that the diffusive limit, characterized by a phonon MFP equal to the diameter of the NW, cannot take account of the roughness effect, they used the MFP as an adjusting parameter. A very good fitting to the experimental values was achieved for values of MFP below the NW diameter, therefore, below the Casimir limit. The result of this paper, together with the confirmation of the violation of the Casimir limit for rough NWs [57], allow the use of simple models based on the Callaway formalism for calcu‐ lating the thermal conductivity of rough NWs, making use of a reduced effective diameter

When considering alloyed NWs, e. g. Si1-xGex NWs, the thermal conductivity modeling must address the effect of the alloy scattering, besides all the above mentioned scattering terms. N. Yang et al. carried out a first approximation in NWs composed of a mixture of Si isotopes making use of a molecular dynamic simulation [58]. They showed a large reduction in ther‐ mal conductivity, due to the isotopic scattering, with a maximum reduction for a composi‐ tion of 50% between the constituent specimens. Afterwards, J. Chen et al. performed an equivalent molecular dynamics simulation on Si1-xGex NWs [59], showing a strong depend‐ ence of the thermal conductivity on the composition of the NW. Furthermore, they found that the thermal conductivity is severely reduced even for small fractions of Ge, what was later on experimentally demonstrated [23]. They estimated a reduction in the thermal con‐ ductivity of 50% with respect to pure Si NWs for a Ge fraction of only 5%; however, one ex‐ perimentally observes a much greater reduction, even for a smaller fraction of Ge in the NW [23]. In another attempt to modeling the thermal conductivity of Si1-xGex NWs, Z. Wang [60] suggested a model on the basis of the previous works made by N. Mingo for Si NWs [30]. They assumed that all the scattering relaxation times, except the impurity relaxation time, can be viewed as independent processes for Si and Ge; therefore, they were added according to the Mathisessen's rule weighted by the composition, while the impurity scattering related to the alloy, is assumed to depend of the composition by a quadratic bowing function. In this way, they obtained a simple model able to achieve a prediction of the thermal conduc‐ tivity of the order of magnitude of the experimental values obtained in [23]. Unfortunately, although they estimate thermal conductivity values of the order of magnitude of the experi‐ mental results, the fitting to the experimental results was not satisfactory, especially in the

Additional work is still needed to supply a good model accounting for the experimental thermal conductivity of Si1-xGex NWs, and in general of alloyed, or doped NWs. If one ob‐ serves the historical development of the models elaborated for describing the thermal con‐ ductivity of pure semiconductor NWs, one can see that the major advances were achieved once reliable experimental data were available. Very recently, new measurements of Si1-xGex NWs have been published by Y. Choongho in a wide range of compositions [18], suggesting

that more new models concerning alloyed NWs will arrive in the next future.

for the boundary scattering.

240 Nanowires - Recent Advances

low temperature range.

As discussed in the previous paragraph different approaches have been attempted for the description of the thermal conductivity of semiconductor NWs. The measurement of the thermal conductivity at the nanometer scale is a rather complex experimental exercise; usu‐ ally, the experimental data are hindered by the incertitude introduced by the NW/ heat sink contact. Therefore, the experimental data providing the thermal conductivity of NWs are scarce, and only a few experiments have supplied reliable thermal conductivity data; these data have been used to check the validity of the thermal transport models allowing a signifi‐ cant advance of the modeling formalisms. Concerning the different approaches used for modeling the thermal conductivity of NWs we have already mentioned that, some of them attained a good fitting to the experimental data using complex calculations, some other re‐ quired of too many fitting parameters; while some other provided a good fitting in a re‐ duced temperature range. The lack of generalized experimental results is a serious problem for both the design of thermoelectric devices based on NWs, and the thermal management of NW based devices. Therefore, a comprehensive model giving a predictive estimation of the thermal conductivity with the factors that contribute to limit it, e. g. the NW diameter, and the surface roughness, would be of high technological interest. On the other hand, ex‐ perimental data are restricted to temperatures below room temperature, the data at higher temperatures are not available; however, this information is required for many applications that can operate at higher temperatures. We present herein a predictive model based in the well-known Callaway-Holland formalism that substantially simplifies the estimation of the thermal conductivity, provides an excellent fit to the experimental data, and can be used as a tool for all the technical situations in which the thermal conductivity of NWs is required for the design of NW based devices.

### **4.2. The Callaway-Holland formalism**

The Callaway-Holland formalism has been successfully used for the description of the ther‐ mal conductivity of bulk semiconductors [40,42,61]; therefore, it must be conveniently adapted to the nanoscale requirements. It is well known that one of the main consequences of the reduced dimension of the NWs is the appearance of quantum confinement effects; in particular, the phonon confinement is relevant for diameters below 20-25 nm [15]. Many of the applications with NWs are built up with NWs with diameters above this phonon con‐ finement limit; therefore, for these NWs one can reasonably assume a bulk like behavior.

First, we will introduce the main features of the Callaway-Holland formalism, initially de‐ veloped for the low temperature range [41,43]. Despite the original limitation in tempera‐ ture, the formalism was demonstrated to be a very useful tool in the calculation of thermal conductivity of semiconductor materials over the full range of temperatures, by the addition of some modifications.

The thermal conductivity is calculated by solving the Boltzmann equation

$$
\left(\frac{\partial \mathbf{N}}{\partial t}\right)\_\mathbf{c} - \mathbf{c} \nabla T \frac{dN}{dT} = 0\tag{1}
$$

where N is the phonon density, and *c* the group velocity. In the presence of a small tempera‐ ture gradient, ∇*T* , the deviation from equilibrium of the phonon mode k damps out expo‐ nentially, and the phonon population decay can be described by a specific relaxation time.

$$
\left(\frac{\partial \mathbf{N}}{\partial t}\right)\_{\mathbf{c}} = \frac{\mathbf{N}^0 - \mathbf{N}}{\tau(\mathbf{k})} \tag{2}
$$

*N* <sup>0</sup> is the Planck distribution function, and *τ*(k) is the relaxation time of the k mode. Callaway split out this expression in order to taking account of the contributions of both normal and resis‐ tive phonon processes, giving the following expression for the variation of the phonon density.

$$\left(\frac{\partial \mathbf{N}}{\partial t}\right)\_{\mathbf{c}} = \frac{\mathbf{N}(\mathbf{A}) - \mathbf{N}}{\tau\_N} + \frac{\mathbf{N}^0 - \mathbf{N}}{\tau\_R} \tag{3}$$

Where *N* (*λ*) is the shifted Planck distribution function towards which the Normal processes approach [62]; it can be expressed in a first approximation as:

$$N(\lambda) \sim N(0) + \lambda \left(\frac{\partial N}{\partial \lambda}\right)\_{\mathbf{0}} = N^0 + \frac{\lambda \mathbf{k}}{K\_B T} \frac{e^{\hbar o/K\_B T}}{\left(e^{\hbar o/K\_B T} - 1\right)^2} \tag{4}$$

On the other hand, because a small deviation from equilibrium was assumed one can ap‐ proximate *dN dT* ≊ *d N* <sup>0</sup> *dT* , therefore, one can simplify the Boltzmann equation as:

$$\left(-\frac{\hbar\alpha}{K\_B T} \mathbf{c} \nabla T + \frac{1}{\tau\_N} \frac{\lambda \mathbf{k}}{K\_B T}\right) \frac{e^{\hbar\alpha/K\_B T}}{\left(e^{\hbar\alpha/K\_B T} - 1\right)^2} - \left(\frac{1}{\tau\_N} + \frac{1}{\tau\_R}\right) \left(N - N^0\right) = 0\tag{5}$$

By introducing a global relaxation time, *α*(k) , the term *N* − *N* <sup>0</sup> of the equation 5 can be ex‐ pressed as:

$$N - N^0 = -\alpha \left(\mathbf{k}\right) \mathbf{c} \nabla T \left(\frac{\hbar \alpha \alpha}{K\_B T^2}\right) \frac{e^{\hbar \alpha / K\_B T}}{\left(e^{\hbar \alpha / K\_B T} - 1\right)^2} \tag{6}$$

allowing to transform the Boltzmann equation in the more tractable form:

<sup>0</sup> *N dN <sup>T</sup> t dT*

where N is the phonon density, and *c* the group velocity. In the presence of a small tempera‐ ture gradient, ∇*T* , the deviation from equilibrium of the phonon mode k damps out expo‐ nentially, and the phonon population decay can be described by a specific relaxation time.

> 0 (k)

t

<sup>0</sup> ( )

Where *N* (*λ*) is the shifted Planck distribution function towards which the Normal processes

0

On the other hand, because a small deviation from equilibrium was assumed one can ap‐

*dT* , therefore, one can simplify the Boltzmann equation as:

<sup>2</sup> /

*B*

<sup>æ</sup> ö æö ç- Ñ+ ÷ ç÷ -+ -= <sup>ç</sup>

h

w

*K T B NB N R*

w

h

*e*

è ø èø -

( )

a

*<sup>e</sup> NN T*

æ ö - =- Ñ ç ÷ ç ÷

**k c**

**λk**

*B*

*K T*

1

By introducing a global relaxation time, *α*(k) , the term *N* − *N* <sup>0</sup> of the equation 5 can be ex‐

h

w

*<sup>e</sup> <sup>T</sup> N N*

*N N NN N*

æ ö ¶ - ç ÷ = +

**λ** t

*N e NN N*

æ ö ¶ + =+ ç ÷

**λk λ λ λ**

*N R*

 t

*K T e*

( ) ( ) /

( )

*B*

2 2 / <sup>1</sup>

h

*K T <sup>B</sup>*

h

w

è ø -

*K T <sup>e</sup>*

/

w

*B*

(6)

*K T*

t t

<sup>1</sup> 1 1 <sup>0</sup>

è ø ¶ **<sup>0</sup>** -

( )

*K T B*

h

w

*B*

/

w

h

<sup>¬</sup> (4)

<sup>2</sup> /

*B*

*K T*

1

0

(5)

 is the Planck distribution function, and *τ*(k) is the relaxation time of the k mode. Callaway split out this expression in order to taking account of the contributions of both normal and resis‐ tive phonon processes, giving the following expression for the variation of the phonon density.

*N NN*

æ ö ¶ ç ÷ <sup>=</sup> è ø ¶ **<sup>c</sup>**

*t*

*t*

è ø ¶ **<sup>c</sup>**

approach [62]; it can be expressed in a first approximation as:

( ) (0)

*KT KT*

t

0

*N* <sup>0</sup>

242 Nanowires - Recent Advances

proximate *dN*

pressed as:

*dT* ≊

*d N* <sup>0</sup>

**c**

h

w

**c** (1)

(2)

(3)

ç ÷ -Ñ =

æ ö ¶

è ø ¶ **<sup>c</sup>**

$$\left(\frac{\alpha(\mathbf{k})}{\tau\_{\mathbb{C}}} - 1\right) \left(\frac{\hbar \omega}{T}\right) \mathbf{c} \nabla \ T + \frac{1}{\tau\_{\mathbb{N}}} \lambda \mathbf{k} = \mathbf{0} \tag{7}$$

Where *τ<sup>C</sup>* =(*τ<sup>N</sup>* <sup>−</sup><sup>1</sup> <sup>+</sup> *<sup>τ</sup><sup>R</sup>* −1 )−1 . On the other hand, assuming isotropic media, *λ* must be a constant vector in the direction of the temperature gradient. Thus, one can define a parameter, *β*, ex‐ pressed as *<sup>β</sup>* <sup>=</sup> <sup>−</sup> *<sup>λ</sup>* ℏ*v* <sup>2</sup> *T* <sup>∇</sup> *<sup>T</sup>* , where *v* is the speed of sound. Finally, in the acoustic approxima‐ tion k= *<sup>c</sup><sup>ω</sup> <sup>v</sup>* 2 and substituting this in the reduced Boltzmann equation, one obtains the following expression for the relaxation time:

$$\alpha = \tau\_{\mathcal{C}} \left( 1 + \left( \frac{\mathcal{P}}{\tau\_N} \right) \right) \tag{8}$$

Using this relaxation time, one can estimate the thermal conductivity through the following equation:

$$\kappa = \frac{1}{\left(2\pi\right)^3} \sum\_{i} \int\_0^{2\pi} \int\_0^{\pi} \left( \int v\_i^2 a\_i \left( k \right) \mathbb{C}\_{ph} \left( k \right)\_i \cos^2 \theta d^3 \mathbf{k} \right) d\theta d\phi \tag{9}$$

Where C*ph* is the phonon specific heat, and *θ* the angle formed between the phonon wave vector and the temperature gradient. Making use of the branch separation proposed by Hol‐ land [42], and considering that the phonon specific heat is expressed as:

$$\mathbf{C}\_{ph} = \frac{d}{dT} \left(\hbar o \mathbf{N}^0\right) = \frac{\left(\hbar o\right)^2}{K\_B T^2} \frac{e^{\hbar o/K\_B T}}{\left(e^{\hbar o/K\_B T} - 1\right)^2} \tag{10}$$

The thermal conductivity transforms into the expression

$$\kappa\_i = \sum\_i \left(\frac{\upsilon\_i^2}{8\pi^3}\right) \left\| \frac{k\_D^i}{K\_B T^2} \right\| \frac{1}{\tau\_C^i} \left(1 + \frac{\beta}{\tau\_N^i}\right) \frac{e^{\hbar o/K\_B T}}{\left(e^{\hbar o/K\_B T} - 1\right)^2} \cos^2\theta d^3k \tag{11}$$

where the subindex refers to the different acoustic branches, the longitudinal and the two transversal branches. The explicit form of the *β* parameter is obtained from the momentum conservation rule in the normal scattering processes; therefore:

$$\int \left(\frac{\partial N}{\partial t}\right)\_N \mathbf{k} d^3 k = \int \left(\frac{N\{\mathbf{A}\} - N}{\tau\_N^i}\right) \mathbf{k} d^3 k = 0 \tag{12}$$

Replacing the term *N* (*λ*)− *N* by their explicit form, previously derived in equation 6, one ar‐ rives to the equation:

$$\int \frac{e^{\hbar\omega/K\_BT}}{\left(e^{\hbar\omega/K\_BT} - 1\right)^2} \left(\frac{\hbar\omega}{K\_BT^2}\right) \mathbf{c} \nabla \, T \{\alpha(\kappa) - \beta\} \frac{\epsilon\omega}{\tau\_N^{i\_BT}} d^3 \mathbf{k} = \mathbf{0} \tag{13}$$

And making use of *c* **<sup>2</sup>** ∇*T* =*vi* 2 *cos* <sup>2</sup> *θ* and replacing *α*(k) by its value given in equation 8, one obtains:

$$\int \frac{e^{\hbar o/K\_{\bf B}T}}{\left(e^{\hbar o/K\_{\bf B}T} - 1\right)^{2}} \left(\frac{\hbar o}{K\_{\bf B}T^{2}}\right) \left(\pi\_{\rm C}^{i} + \left(1 - \frac{\pi\_{\rm C}^{i}}{\pi\_{N}^{i}}\right) \beta\right) \frac{o\nu}{\pi\_{N}\nu^{2}} \upsilon^{2} \cos^{2}\theta d^{3}\mathbf{k} = 0\tag{14}$$

From this expression and taking account of the independence between *β* and k one obtains

$$\beta = \frac{\int \frac{\tau\_C^i}{\tau\_N^i} \frac{e^{\hbar \alpha / K\_B T}}{\left(e^{\hbar \alpha / K\_B T} - 1\right)^2} \left(\frac{\hbar \alpha}{K\_B T^2}\right) \alpha \cos^2 \theta d^3 \mathbf{k}}{\int \left(1 - \frac{\tau\_C^i}{\tau\_N^i}\right) \frac{1}{\tau\_N^i} \frac{e^{\hbar \alpha / K\_B T}}{\left(e^{\hbar \alpha / K\_B T} - 1\right)^2} \left(\frac{\hbar \alpha}{K\_B T^2}\right) \alpha \cos^2 \theta d^3 \mathbf{k}} \tag{15}$$

Finally, taking into account that *d* <sup>3</sup> *<sup>k</sup>* <sup>=</sup> *<sup>ω</sup>* <sup>2</sup> *<sup>v</sup>* <sup>3</sup> *<sup>d</sup>ωdΩ<sup>k</sup>* <sup>=</sup> *<sup>ω</sup>* <sup>2</sup> *<sup>v</sup>* <sup>3</sup> *sinθdθdϕdω*, and the change of variable *x* =ℏ*ω* / *KBT* , the thermal conductivity can be expressed as:

( ) ( ) ( ) 4 3 <sup>2</sup> 4 2 3 3 00 0 2 2 4 2 00 0 2 2 4 2 00 0 2 8 1 1 <sup>1</sup> <sup>1</sup> 1 *i i i <sup>B</sup> <sup>i</sup> <sup>T</sup> <sup>C</sup> i i i x T C <sup>i</sup> <sup>x</sup> <sup>N</sup> i x T C i i <sup>x</sup> N N x x K T <sup>e</sup> x cos sin d d dx <sup>v</sup> <sup>e</sup> <sup>e</sup> x cos sin d d dx e <sup>e</sup> x cos sin d d dx e* <sup>ì</sup> ì ü æ öï ï ï <sup>=</sup> ç ÷í í <sup>ý</sup> <sup>+</sup> ç ÷ è øï ï - <sup>ï</sup> <sup>î</sup> î þ ì ü ï ï í ý ï ï - î þ ì ü æ ö ï ï í ý ç ÷ - ç ÷ ï ï è ø - <sup>î</sup> æ è þ + å òò ò òò ò òò ò h q p p q p p q p p k t q qqj p t q qqf t t q qqf t t ( ) 2 4 2 00 0 2 1 *<sup>i</sup> i x T C <sup>i</sup> <sup>x</sup> <sup>N</sup> <sup>e</sup> x cos sin d d dx e* ì ü ï ï í ý ï <sup>ö</sup> <sup>ü</sup> ç ÷ <sup>ï</sup> ï ï ï ý ï ï <sup>ï</sup> <sup>ç</sup> ï þ <sup>÷</sup> <sup>ï</sup> <sup>ø</sup> <sup>þ</sup> - <sup>î</sup> òò ò q p p t q qqf t (16)

Were the integral limits *θ<sup>i</sup>* are the characteristic temperature cut-offs of each branch.

#### **4.3. The phonon scattering terms**

( ) 3 3 <sup>0</sup> *<sup>i</sup> <sup>N</sup> <sup>N</sup>*

Replacing the term *N* (*λ*)− *N* by their explicit form, previously derived in equation 6, one ar‐

*KBT* <sup>2</sup> )*c*∇*<sup>T</sup>* (*α*(κ)−*β*) *<sup>c</sup><sup>ω</sup>*

*τN i <sup>v</sup>* <sup>2</sup> *<sup>d</sup>* <sup>3</sup>

1 k 0

 w

 t

*θ* and replacing *α*(k) by its value given in equation 8, one

2 23

2 3

( ) 2 4 2

*e*


1

<sup>÷</sup> <sup>ï</sup> <sup>ø</sup> <sup>þ</sup>

00 0 2

t

t

ï

q

òò ò

p p

*<sup>i</sup> i x T C <sup>i</sup> <sup>x</sup> <sup>N</sup>*

k

 q

(12)

k=0 (13)

(14)

*<sup>v</sup>* <sup>3</sup> *sinθdθdϕdω*, and the change of variable

*<sup>e</sup> x cos sin d d dx*

q qqf

ì ü ï ï í ý (15)

ï ï ï ý ï ï (16)

ï þ

*d k d k*

*N N N*

2 2 <sup>2</sup> /

*<sup>C</sup> <sup>i</sup> K T <sup>B</sup> N N*

( )

*<sup>k</sup>* <sup>=</sup> *<sup>ω</sup>* <sup>2</sup>

*i K T*

w

h h

w

*C*

t

t

ò

ò

b

Finally, taking into account that *d* <sup>3</sup>

( )

1

*i x*

*e*

è þ

2 4 2

*i x*

*e*

8 1

*<sup>B</sup> <sup>i</sup> <sup>T</sup> <sup>C</sup>*

p p

2 4 2

*i i <sup>x</sup> N N*

 t

t

00 0 2

t

t

*T C <sup>i</sup> <sup>x</sup> <sup>N</sup>*

00 0 2

<sup>1</sup> <sup>1</sup>

t

*T C*

*i*

q

òò ò

p p

òò ò

p p

*i i*

p

h

æ

k

+

*i*

q

å òò ò

( )

1

4 3 <sup>2</sup> 4 2 3 3 00 0 2

t

<sup>ì</sup> ì ü æ öï ï ï <sup>=</sup> ç ÷í í <sup>ý</sup> <sup>+</sup> ç ÷ è øï ï - <sup>ï</sup> <sup>î</sup> î þ ì ü ï ï í ý ï ï - î þ ì ü æ ö ï ï í ý ç ÷ - ç ÷ ï ï è ø - <sup>î</sup>

*i*

q *C*

t

t t

*x* =ℏ*ω* / *KBT* , the thermal conductivity can be expressed as:

( )

*<sup>e</sup> x cos sin d d dx*

q qqf

*<sup>e</sup> x cos sin d d dx*

q qqf

*x x K T <sup>e</sup> x cos sin d d dx <sup>v</sup> <sup>e</sup>*

( )

*i i K T N N <sup>B</sup>*

q qqj

<sup>ö</sup> <sup>ü</sup> ç ÷ <sup>ï</sup>

<sup>ï</sup> <sup>ç</sup>

h h

w

2 2 /

*B B*

<sup>1</sup> 1 k 1

*K T <sup>e</sup>*

*<sup>v</sup>* <sup>3</sup> *<sup>d</sup>ωdΩ<sup>k</sup>* <sup>=</sup> *<sup>ω</sup>* <sup>2</sup>

1

*K T <sup>e</sup>*

*B B*

*<sup>i</sup> K T N B i K T*

<sup>=</sup> æ ö æ ö ç ÷ - ç ÷ ç ÷ ç ÷ è ø - è ø

w

*K T <sup>v</sup> <sup>e</sup>*

*i C*

t tb

t

From this expression and taking account of the independence between *β* and k one obtains

/ 2 3

w w q

æ ö ç ÷ ç ÷ - è ø

*<sup>e</sup> cos d*

h

2 2

*<sup>e</sup> cos d*

h

w w q

*<sup>e</sup> v cos d*

æ öæ ö æ ö ç ÷ +- ç ÷ ç ÷ <sup>=</sup> ç ÷ç ÷ ç ÷ - è øè ø è ø <sup>ò</sup> <sup>h</sup>

*K T i*

h

w

æ ö ¶ æ ö - = = ç ÷ ç ÷ è ø ¶ è ø ò ò **<sup>λ</sup> k k** t

*t*

ℏ*ω*/*KBT*

∇*T* =*vi* 2 *cos* <sup>2</sup>

1

<sup>ℏ</sup>*ω*/*KBT* <sup>−</sup>1)<sup>2</sup> ( <sup>ℏ</sup>*<sup>ω</sup>*

*<sup>∫</sup> <sup>e</sup>*

(*e*

( )

h

w

/

w

*B B*

rives to the equation:

244 Nanowires - Recent Advances

And making use of *c* **<sup>2</sup>**

obtains:

The application of the Callaway-Holland formalism to a particular problem in an extended temperature range resumes in the appropriate characterization of the different scattering terms. Therefore, a detailed analysis of the different scattering processes is required.

#### *4.3.1. Three phonon scattering*

This kind of process is determined by the exclusive interaction between phonons and fol‐ lows two ways. In the first one, a phonon of wave vector q1 combines with a phonon of wave vector q2 given another phonon of wave vector q3. In the second process a phonon of wave vector q1 decays in two phonons of wave vectors q2 and q3 respectively.

$$\begin{aligned} q\_1 + q\_2 &= q\_3\\ q\_1 &= q\_2 + q\_3 \end{aligned} \tag{17}$$

In these processes the energy is conserved; but such conservation does not necessarily hold for the momentum. The processes in which the momentum is not conserved are called Umpkklap, in this process the initial and final moment differs by a non zero reciprocal lat‐ tice vector, and the resultant phonons are out of the Brillouin zone; the momentum selection rule for these processes reads as:

$$\begin{aligned} q\_1 + q\_2 &= q\_3 + G\\ q\_1 &= q\_2 + q\_3^\cdot + G \end{aligned} \tag{18}$$

Where q'3 is the "reflected" phonon of wave vector q'3, and G the reciprocal lattice vector [28]. The phonon wave vector is opposite to q1 and q2 phonons, resulting in a process resis‐ tive to the heat flow.

Peierls proposed for the Umpklapp processes a scattering rate of the form *τ<sup>U</sup>* <sup>−</sup>1∝*T <sup>a</sup> e <sup>θ</sup>*/*bT* [62], where the values of the parameters *a* and *b* have been determined in different ways [62-65]. Klemens showed that in the high temperature range the resistive contribution of the three phonon processes obeys to *τ<sup>U</sup>* <sup>−</sup><sup>1</sup> =*Cω* <sup>2</sup> *T* [64]. Using the Slack's variant [65] of the Peierls form, one can extend the temperature range, the Slack relaxation time for the U processes is given as *τ<sup>U</sup>* <sup>−</sup><sup>1</sup> = *Bω* <sup>2</sup> *T e <sup>θ</sup>*/3*<sup>T</sup>* . Using this form, Slack established that the thermal conductivities of different pure crystals can be modeled over an extended temperature range giving a good agreement with the experimental values [65]. Furthermore, when the temperature increases, it adopts the same expression as the Klemens relaxation time. Klemens evaluated the con‐ stant C as *<sup>C</sup>* <sup>=</sup> <sup>2</sup>*<sup>γ</sup>* <sup>2</sup> ℏ *M v* <sup>2</sup> *<sup>θ</sup>* [27,64]. Where *γ*is the Grüneisen parameter, M the average atomic mass in the crystal, and *v* the velocity of sound. Finally, we can introduce the Umpklapp relaxa‐ tion time for each phonon branch as:

$$\sigma \left( \tau\_{\mathcal{U}}^{-1} \right)^{i} = \frac{2 \gamma\_{i}^{2} \hbar}{M v\_{i}^{2} \theta\_{i}} \alpha^{2} T e^{\theta\_{i}/3T} \tag{19}$$

Regarding the Normal processes, it has been claimed that they do not contribute to the ther‐ mal resistance [25, 27, 30, 31, 33]. However, they cannot be neglected because the Callaway-Holland formalism requires of the normal process relaxation time [41-43]. Furthermore, the normal processes can contribute indirectly to the thermal resistance by redistributing the phonon population, which can further participate in resistive scattering processes. These can be explained by the Herring mechanism [32], which redistributes the phonon momentum among the different phonon branches. In this mechanism, the relaxation of transverse pho‐ nons consists of its interaction with a longitudinal acoustic phonon, resulting in another lon‐ gitudinal phonon. On the other hand, the relaxation of the longitudinal phonons occurs via two different processes; they decay in two transverse phonons; or a longitudinal phonon arises from two transverse phonons. Taking account of these processes, the relaxation time suggested by Herring for each normal process can be written as [32]:

$$\left(\left(\boldsymbol{\tau}\_{N}^{L}\right)^{-1} = \mathbb{C}\_{L}o\boldsymbol{\alpha}^{2}T^{3}\;\;\;\;\left(\left(\boldsymbol{\tau}\_{N}^{T}\right)^{-1} = \mathbb{C}\_{T}o\boldsymbol{\alpha}T^{4}\right) \tag{20}$$

Where the values of the constants *Ci* can be described in terms of the physical properties of the material, which according to the approximation made by Asen-Palmer et al. [66], take the form:

$$\mathbf{C}\_{L} \approx \frac{\mathbf{K}\_{B}{}^{3} \boldsymbol{\gamma}\_{L}^{2} \mathbf{V}}{\mathbf{M} \boldsymbol{\hbar}^{2} \boldsymbol{v}\_{L}^{5}} \; ; \mathbf{C}\_{T} \approx \frac{\mathbf{K}\_{B}{}^{4} \boldsymbol{\gamma}\_{T}^{2} \mathbf{V}}{\mathbf{M} \boldsymbol{\hbar}^{3} \boldsymbol{v}\_{T}^{5}} \tag{21}$$

where V is the unit cell volume.

#### *4.3.2. Isotope scattering*

The impurities are centers for phonon scattering. This mechanism is described as the isotope scattering and is related to the mass difference between the host atoms and the impurity atoms. Even in the case of pure crystals, without impurities, is necessary to consider the ex‐ istence of isotopes, which efficiently scatter high frequency phonons. The perturbation in the lattice due to a substitutional atom was evaluated by Klemens using perturbation theory [44,64], and the expression for the relaxation time due to isotope mixing can be expressed as

$$\left(\tau\_I^i\right)^{-1} = \frac{V\Gamma}{4\pi v\_i^3} \alpha^4 T^4 \tag{22}$$

Where *Γ* is the mass-fluctuation parameter defined for a single element as *Γ* =∑ *i f i* (1− *Mi <sup>M</sup>* ) 2 where *f <sup>i</sup>* is the fractional atomic abundance of the isotope (impurity) of atomic mass *Mi* [44].

#### *4.3.3. Boundary scattering*

( )

t


*U*

suggested by Herring for each normal process can be written as [32]:

 wt

*L T*

g

t

Where the values of the constants *Ci*

where V is the unit cell volume.

*4.3.2. Isotope scattering*

the form:

246 Nanowires - Recent Advances

( ) ( ) 1 1 2 3 4 ; *L T NL NT C T CT* - -

the material, which according to the approximation made by Asen-Palmer et al. [66], take

3 2 4 2 2 5 3 5 ; *B L B T*

*KV KV C C Mv Mv* » » h h

*L T*

The impurities are centers for phonon scattering. This mechanism is described as the isotope scattering and is related to the mass difference between the host atoms and the impurity atoms. Even in the case of pure crystals, without impurities, is necessary to consider the ex‐ istence of isotopes, which efficiently scatter high frequency phonons. The perturbation in the lattice due to a substitutional atom was evaluated by Klemens using perturbation theory [44,64], and the expression for the relaxation time due to isotope mixing can be expressed as

> ( ) <sup>1</sup> 4 4 <sup>3</sup> 4


*i <sup>V</sup> <sup>T</sup> v*

 w = p

*i I*

t

 g  w

= = (20)

can be described in terms of the physical properties of

(22)

(21)

2 1 2 /3 2 2 *<sup>i</sup> <sup>i</sup> <sup>i</sup> <sup>T</sup>*

*Mv*

gh

*i i*

q

 w

Regarding the Normal processes, it has been claimed that they do not contribute to the ther‐ mal resistance [25, 27, 30, 31, 33]. However, they cannot be neglected because the Callaway-Holland formalism requires of the normal process relaxation time [41-43]. Furthermore, the normal processes can contribute indirectly to the thermal resistance by redistributing the phonon population, which can further participate in resistive scattering processes. These can be explained by the Herring mechanism [32], which redistributes the phonon momentum among the different phonon branches. In this mechanism, the relaxation of transverse pho‐ nons consists of its interaction with a longitudinal acoustic phonon, resulting in another lon‐ gitudinal phonon. On the other hand, the relaxation of the longitudinal phonons occurs via two different processes; they decay in two transverse phonons; or a longitudinal phonon arises from two transverse phonons. Taking account of these processes, the relaxation time

*Te*

q

(19)

A real crystal has a finite size; therefore, the scattering due to the crystal boundaries cannot be neglected. The boundaries act as highly efficient scattering regions, confining the MFP of the phonons to the order of the dimension of the sample. This problem was discussed by Ziman on the basis of the Casimir theory [28], and for the nanoscaled samples is the domi‐ nant scattering mechanism accounting for the reduction of the thermal conductivity. This process can be phenomenologically explained by considering two types of phonon boun‐ dary scattering processes; one of them corresponding to the specular phonon scattering, which mainly happens in smooth surfaces; while the other one corresponds to the pure dif‐ fusive phonon scattering, which mainly occurs on rough surfaces. Taking account of both processes, which occur with a probability *p* for the pure specular scattering, and 1-*p* for purely diffusive scattering, Ziman derived an expression for the MFP of phonons in a finite sample as *<sup>Λ</sup>* <sup>=</sup> <sup>1</sup> <sup>+</sup> *<sup>p</sup>* <sup>1</sup> <sup>−</sup> *<sup>p</sup> Λ*0 where *Λ*0 is the MFP for a perfectly rough boundary [28], where only diffusive processes occur. It depends on the geometry of the sample; in the case of a cylin‐ der, this MFP is equal to the diameter of the cylinder, *Λ*<sup>0</sup> =*D*.

The key issue of this scattering lies in the form of the probability function, *p,* which has been addressed in different ways in the literature. Commonly, it has been taken equal to zero [25, 30, 33, 37] or as a constant independent of temperature and frequency [27, 34, 35], thereby simplifying the boundary scattering relaxation time. However, a more realistic approxima‐ tion can be achieved if one eliminates such oversimplifications. Ziman showed that the probability function *p* depends on the wavelength of the incident phonons [28], being the surface characterized by the mean value of the height deviation over a surface reference lev‐ el; this parameter is called the asperity parameter,*η*. In a first approximation, the probability function can be written as *<sup>p</sup>* <sup>=</sup>*exp* <sup>−</sup>16*<sup>π</sup>* <sup>3</sup> *η* 3 *<sup>λ</sup>* <sup>2</sup> , which has been used by some authors [50]. De‐ spite being a more accurate way of treating the boundary scattering with respect to the as‐ sumption of a constant value of *p*, the hypotheses assumed in the construction of this probability function were considered by Ziman asnon very realistic. Therefore, Ziman him‐ self developed an alternative treatment, which takes account of the existence of a distribu‐ tion function for the roughness instead of the average value given by the asperity parameter. In this way, the probability function takes the form *<sup>p</sup>*(*η*, *<sup>λ</sup>*)=*∫P*(*η*)*Exp* <sup>−</sup>16*<sup>π</sup>* <sup>3</sup> *η* 2 *<sup>λ</sup>* <sup>2</sup> *<sup>d</sup>η* where *P*(*η*) is the roughness distribution probability. This expression can be simplified if one considers that the reflection probability is a measure of the fraction of area whose average roughness is less than *<sup>λ</sup> <sup>π</sup>*<sup>4</sup> [28], reducing the probability function to *p*(*η*, *λ*)≈ *∫* 0 *λ* 4*π P*(*η*)*dη*. This is the form used by Mingo et al. for approaching the thermal conductivity in the low temperature range for rough Silicon NWs with a roughness distribution given as *P*(*η*)= <sup>1</sup> *<sup>η</sup> Exp* <sup>−</sup> *<sup>η</sup> η*0 [67].

Moreover, based on the dependence of the probability function *p* with frequency and rough‐ ness, Soffer addressed a different solution in the presence of a temperature gradient along the sample [68], according to which the probability function takes the form:

$$p\left(\alpha\nu,\eta\right) = \exp\left[-\left(2\eta\left(\bigvee\_{v\_i}\alpha\bigvee\_{v\_i}\right)\sin\theta\right)^2\right] \tag{23}$$

where *η* is the asperity parameter defined by Ziman, and *θ* the angle of incidence of pho‐ nons. This form given by Soffer is the one used in our simulations because we tested that it was the one providing the best results. Finally, adding the limitation in length to the boun‐ dary relaxation time, the final expression for the boundary relaxation time can be written as

$$\left(\tau\_B^i\right)^{-1} = \left(\frac{1-p}{1+p}\Lambda\_0^{-1} + L^{-1}\right)v\_i \tag{24}$$

#### **4.4. The thermal conductivity of NWs**

Using this relaxation time, the thermal conductivity can be calculated over the full tempera‐ ture range. The solution of the complex equation 16 was carried out by the Gauss-Kronrod local adaptive method, fixing a tolerance <10-9 for enhancing the accuracy. This method pro‐ vides calculation several times faster than other common numerical methods applied to this kind of problems. In order to test the model, first we carried out a prediction of the thermal conductivity of bulk silicon and germanium, obtaining an excellent fit to the experimental measurements over the full temperature range (Figure 5-A and 5-B).

The parameters used for the calculations are given in table 1. The values for *Γ* and *γ<sup>i</sup>* were taken from reference [61]; the Grüneisen parameter, *γ<sup>i</sup>* , was approximated to a constant val‐ ue. Actually, this is crude approach, since the Grüneisen parameter has a dependence with temperature, especially in the low temperature range [44]. The rough simplification of a con‐ stant value for this parameter arises from the dependence with the temperature and fre‐ quency of the three phonon scattering relaxation times, which is generically expressed as *τj* <sup>−</sup><sup>1</sup> <sup>=</sup>*Ctsj* <sup>×</sup>*Gj* (*ω*)× *Fj* (*T* ), namely the product of some function of *ω*, some function of the tem‐ perature T and a given constant value for each type of scattering, *Ctsj* . Therefore, the tem‐ perature dependence of each three-phonon scattering process can be completely included in the function *F* (*T* ); which encloses the temperature dependence of the Grüneisen parameter, while the constant parts of the Grüneisen parameters are encompassed in the constant group *Ctsj* . The bulk value for the sound velocities of each phonon branch was taken from [44]. Finally, the asperity parameter was chosen assuming a smooth surface, matching the values reported in [40] for smooth silicon NWs.

function to *p*(*η*, *λ*)≈ *∫*

248 Nanowires - Recent Advances

distribution given as *P*(*η*)= <sup>1</sup>

0

*<sup>η</sup> Exp* <sup>−</sup> *<sup>η</sup>*

( )

wh

*i*

measurements over the full temperature range (Figure 5-A and 5-B).

perature T and a given constant value for each type of scattering, *Ctsj*

taken from reference [61]; the Grüneisen parameter, *γ<sup>i</sup>*

t

**4.4. The thermal conductivity of NWs**

*τj*

<sup>−</sup><sup>1</sup> <sup>=</sup>*Ctsj* <sup>×</sup>*Gj*

(*ω*)× *Fj*

*η*0 [67].

the sample [68], according to which the probability function takes the form:

*p Exp <sup>v</sup>*

, 2 sin

( ) <sup>1</sup> 1 1

1 1

 h

*P*(*η*)*dη*. This is the form used by Mingo et al. for approaching the

2

(24)

(23)

, was approximated to a constant val‐

. Therefore, the tem‐

 q

thermal conductivity in the low temperature range for rough Silicon NWs with a roughness

Moreover, based on the dependence of the probability function *p* with frequency and rough‐ ness, Soffer addressed a different solution in the presence of a temperature gradient along

é ù æ ö æ ö = -ê ú ç ÷ ç ÷ è ø è ø ë û

where *η* is the asperity parameter defined by Ziman, and *θ* the angle of incidence of pho‐ nons. This form given by Soffer is the one used in our simulations because we tested that it was the one providing the best results. Finally, adding the limitation in length to the boun‐ dary relaxation time, the final expression for the boundary relaxation time can be written as

0

Using this relaxation time, the thermal conductivity can be calculated over the full tempera‐ ture range. The solution of the complex equation 16 was carried out by the Gauss-Kronrod local adaptive method, fixing a tolerance <10-9 for enhancing the accuracy. This method pro‐ vides calculation several times faster than other common numerical methods applied to this kind of problems. In order to test the model, first we carried out a prediction of the thermal conductivity of bulk silicon and germanium, obtaining an excellent fit to the experimental

The parameters used for the calculations are given in table 1. The values for *Γ* and *γ<sup>i</sup>* were

ue. Actually, this is crude approach, since the Grüneisen parameter has a dependence with temperature, especially in the low temperature range [44]. The rough simplification of a con‐ stant value for this parameter arises from the dependence with the temperature and fre‐ quency of the three phonon scattering relaxation times, which is generically expressed as

perature dependence of each three-phonon scattering process can be completely included in

(*T* ), namely the product of some function of *ω*, some function of the tem‐

*<sup>p</sup> L v*

*B i*

*p* - - - æ ö - = L+ ç ÷ + è ø

w

*i*

*λ* 4*π*

**Figure 5.** Fitting of the experimental thermal conductivities of bulk Si (A), and bulk Ge(B) by our modified Callaway-Holland model. Experimental data taken from [14].

The values of the cut-off temperatures, *θ<sup>i</sup>* , and the effective size of the sample remains as free parameters in the resolution of the equations. The effective size is characterized by the lengths L and *Λ*0, which take the same value as in the bulk for a sake of simplicity.


**Table 1.** Parameters used in the calculations of the thermal conductivity of bulk Si and bulk Ge.

The equivalent cut-off temperature obtained as *θ* =3(*θ<sup>L</sup>* <sup>−</sup><sup>1</sup> <sup>+</sup> <sup>2</sup>*θ<sup>T</sup>* −1 )−<sup>1</sup> [69] has a value of 495 K for Si, and 307 K for Ge. In the original Callaway model the cut-off temperature matches the Debye temperature, albeit this parameter has been treated in different ways in further works. The Debye temperature for Si determined from the specific heat takes a value of 645 K [61]. If one uses the sound velocities of the acoustic branches instead of the heat capacity, the cut-off temperature scales up to 710 K [61]; whereas, if one estimates this temperature through the zone-boundary frequencies of the transversal and longitudinal acoustic pho‐ nons, the value drops dramatically to 298 K [61]. One can also consider the fact that the Call‐ away model ignores the contributions of optical phonons, because of the continuum approximation. Fortunately, this limitation can be circumvented by increasing the cut-off temperatures [44]. Thus, the use of the cut-off temperature as a free adjustable parameter implicitly includes the optical phonon contribution.

Finally, the equivalent sample diameter, being of the order of millimeters, does not have a big impact in the thermal conductivity for temperatures above 300 K. This allows to adjust the temperature cut-offs for matching the experimental values in this temperature range. Therefore, the equivalent sample diameter is commonly used to matching the thermal con‐ ductivity in the low temperature range [61].

On the other hand, one can assume that the phonon confinement effects only show relevant effects below 25 nm in diameter [15]; this means that one can treat the NWs with diameter larger than 25 nm in a bulk-like approach.

Having established the excellent agreement with the experimental data for bulk Si and Ge, the same procedure was applied to Si NWs, and the results were compared to those meas‐ ured in [11]. The parameters used for these calculations are given in Table 2. The agreement with the experimental results was excellent over the full temperature range for NWs with‐ out phonon confinement (D>25nm). While the results, as expected, show imprecise below the phonon confinement threshold (D< 25nm). The pseudo-linear experimental behavior of the 22 nm NW cannot be predicted by this bulk-like model; which suggests that alternative approaches such as the one conducted in [36] are required for describing the thermal trans‐ port in NWs of reduced diameter, exhibiting confinement effects. These results are summar‐ ized in Figure 6.

**Figure 6.** Fitting to the experimental thermal conductivity of Si NWs with different diameters, showing the excellent fit for all of them, except for the 22 nm NW. Experimental data taken from [11]

Several authors have shown that the speed of sound of the different phonon branches is lower in semiconductor NWs than in their corresponding bulk counterparts, even for NWs with diameters larger than the one for phonon confinement [25, 27, 34, 40, 69]. The values of the sound velocities present differences depending upon the method followed to obtaining them; therefore, the sound velocities are left as free parameters when adjusting the experi‐ mental thermal conductivity data. The cut-off temperatures for semiconductor NWs must be also lower than their bulk counterparts [30, 34, 40, 69]. This is consistent with the values of the cut-off temperatures and the sound velocities found by fitting the experimental thermal conductivity data, Table 2.


**Table 2.** Parameters used in the calculations of the thermal conductivity of smooth Si NWs of [11].

temperatures [44]. Thus, the use of the cut-off temperature as a free adjustable parameter

Finally, the equivalent sample diameter, being of the order of millimeters, does not have a big impact in the thermal conductivity for temperatures above 300 K. This allows to adjust the temperature cut-offs for matching the experimental values in this temperature range. Therefore, the equivalent sample diameter is commonly used to matching the thermal con‐

On the other hand, one can assume that the phonon confinement effects only show relevant effects below 25 nm in diameter [15]; this means that one can treat the NWs with diameter

Having established the excellent agreement with the experimental data for bulk Si and Ge, the same procedure was applied to Si NWs, and the results were compared to those meas‐ ured in [11]. The parameters used for these calculations are given in Table 2. The agreement with the experimental results was excellent over the full temperature range for NWs with‐ out phonon confinement (D>25nm). While the results, as expected, show imprecise below the phonon confinement threshold (D< 25nm). The pseudo-linear experimental behavior of the 22 nm NW cannot be predicted by this bulk-like model; which suggests that alternative approaches such as the one conducted in [36] are required for describing the thermal trans‐ port in NWs of reduced diameter, exhibiting confinement effects. These results are summar‐

**Figure 6.** Fitting to the experimental thermal conductivity of Si NWs with different diameters, showing the excellent

Several authors have shown that the speed of sound of the different phonon branches is lower in semiconductor NWs than in their corresponding bulk counterparts, even for NWs with diameters larger than the one for phonon confinement [25, 27, 34, 40, 69]. The values of the sound velocities present differences depending upon the method followed to obtaining them; therefore, the sound velocities are left as free parameters when adjusting the experi‐

fit for all of them, except for the 22 nm NW. Experimental data taken from [11]

implicitly includes the optical phonon contribution.

ductivity in the low temperature range [61].

larger than 25 nm in a bulk-like approach.

ized in Figure 6.

250 Nanowires - Recent Advances

On the other hand, if one assumes that the crystallinity and the impurity concentration are the same for rough and smooth NWs, the only relevant difference between both types of NWs is the surface roughness; therefore, the thermal conductivity reduction in rough NWs must be exclusively related to the contribution of the roughness to the boundary scattering. The roughness is included in the boundary scattering through the asperity parameter *η*; choosing a value of *η*= 1 nm, typically reported for smooth NWs [40], the value of *p* is close to zero for most of the phonon frequencies, which means that even for such small roughness the boundary scattering is close to the diffusive limit in agreement with the result of refer‐ ence [56]. Therefore, further increase of *η* cannot result in a substantial increase of the diffu‐ sive scattering contribution, and the dramatic reduction of the thermal conductivity, experimentally observed in rough NWs [2], cannot be accounted for by the alone diffusive boundary scattering. The breakdown of the Casimir limit observed in [57] allows us to intro‐ duce in the calculation the concept of an effective Casimir length (*Λ*<sup>0</sup> ' <*Λ*0) smaller than the classical MFP given by the diameter of the NW consequence of the shrinkage of the acoustic phonon MFP in rough NW. The prediction of the thermal conductivity of rough NWs re‐ quires the evaluation of the effective Cassimir length *Λ*<sup>0</sup> ' ; for this purpose we used the exper‐ imental data of a Si NW with 115 nm diameter and roughness of *η*=5. 7 nm [2]. Assuming that the only difference with the smooth NW of the same diameter lies on the roughness, all the other parameters showed in Table 2 already used for the calculation of the thermal con‐ ductivity of the smooth NW of the same dimension (115 nm) remained unchanged. In Fig‐ ure 7-A the thermal conductivity calculated for the rough NW with an effective Casimir length *Λ*<sup>0</sup> ' =115 nm gives a value for the thermal conductivity at 300 K only 1 W/mK less than the thermal conductivity of the 115 smooth NW; therefore, the very increase of the asperity parameter cannot account for the abrupt decrease of the thermal conductivity in rough NWs, as long as the phonon MFP is not reduced below the Casimir limit; however, as the MFP is reduced, the thermal conductivity falls down dramatically, reaching the experimen‐ tal value reported in [2] for an effective Cassimir length of *Λ*<sup>0</sup> ' ≈13 nm, Figure 7-A.

**Figure 7.** Calculated thermal conductivities of a rough Si NW ( 115 nm diameter) for different effective Casimir lengths, showing fitting to experimental data ( taken from [2]) for Λ<sup>0</sup> ' =13 nm (A); fitting of the experimental thermal conductivities for rough Si NWs with different diameters, the fitting parameters are summarized in Table. 3 (B), experi‐ mental data taken from [2].

This large MFP reduction in rough NWs cannot be explained in terms of the surface scatter‐ ing mechanism described by Ziman [28], and must be considered in a first instance as an admissible calculation tool for obtaining good experimental data approach as was shown in [17]. The thermal conductivities calculated for NWs with the same geometric characteristics as those measured in [2] are shown in Figure 7-B, evidencing a very good fitting to the ex‐ perimental data, using the parameters listed in Table 3. The effective Casimir lengths ob‐ tained by fitting the experimental values show a significant reduction with respect to the NW diameters, the ideal Casimir lengths; similar reductions have been also reported by oth‐ er authors [17].


**Table 3.** Parameters used in the calculations of the thermal conductivity of rough NW of [2].

#### **4.5. A predictive expression for the thermal conductivity of Si NWs**

A big constraint for the applications of NWs in which the thermal conductivity plays a ma‐ jor role is the lack of experimental data over a large range of diameters and temperatures, or in its defect the absence of a model supplying such information. In view of the difficulties inherent to thermal conductivity measurements at the nanoscale, and the variety of diame‐ ters and surface roughness, the development of a model capable of predicting the thermal conductivity in a wide range of diameters and surface roughness acquires a capital interest.

We have shown that one can achieve an excellent fit of the experimental thermal conductivi‐ ty of semiconductor NWs using a modified Callaway-Holland formalism. This agreement should permit us the prediction of the thermal conductivity for NW diametersand the tem‐ perature ranges for which experimental data are not available. Particularly relevant is the case of the high temperature range, above 400 K, for which there are not experimental meas‐ urements available. Since the model for bulk Si provides a very good result in the high tem‐ perature range, it is reasonable to assume that the same model for NWs without phonon confinement works well in the high temperature range.

In our model, the fitting to the thermal conductivity is achieved using the speeds of sound and the cut-off temperatures for transverse and longitudinal modes, as free parameters. By relating those parameters to the NW diameter, one could predict the thermal conductivity for any NW diameter of a given roughness. In the case of the smooth NWs, a very good re‐ sult was obtained for NWs with a diameter above the phonon confinement limit. Unfortu‐ nately, there are only three diameters for which reliable experimental data are available. However, additional data exist when one considers rough NWs; assuming that the reduc‐ tion in the thermal conductivity of these NWs with respect to smooth NWs is exclusively related to the contribution of the roughness to the boundary scattering, which does not modify the sound velocities, nor the cut-off temperatures, one can extend the range of avail‐ able experimental data. This hypothesis was confirmed with rough NWs of 115 nm diame‐ ter, for which an excellent fitting was achieved by acting only on the boundary relaxation time, keeping all the other parameters the same as the ones used for the smooth 115 nm NW (see section 4. 4.). Therefore, one can use the values of the adjustable parameters, speed of sound, and cut off-temperature, of both smooth and rough NWs to build up the predictive expression for the thermal conductivity.

**Figure 7.** Calculated thermal conductivities of a rough Si NW ( 115 nm diameter) for different effective Casimir

conductivities for rough Si NWs with different diameters, the fitting parameters are summarized in Table. 3 (B), experi‐

This large MFP reduction in rough NWs cannot be explained in terms of the surface scatter‐ ing mechanism described by Ziman [28], and must be considered in a first instance as an admissible calculation tool for obtaining good experimental data approach as was shown in [17]. The thermal conductivities calculated for NWs with the same geometric characteristics as those measured in [2] are shown in Figure 7-B, evidencing a very good fitting to the ex‐ perimental data, using the parameters listed in Table 3. The effective Casimir lengths ob‐ tained by fitting the experimental values show a significant reduction with respect to the NW diameters, the ideal Casimir lengths; similar reductions have been also reported by oth‐

**D (nm) L (m) Λ0 (m) η (m) θ***L* **(K) θ***T* **(K)** *vL* **(m/s)** *vT* **(m/s) γ***<sup>L</sup>* **γ***<sup>T</sup>* 5x10-9 13x10-9 5. 7 x10-9 537. 5 302. 9 7962 4741 1. 1 0. 66 5x10-9 8. 8x10-9 5. 7x10-9 530. 1 300. 2 7745 4584 1. 1 0. 66 5x10-9 3. 3x10-9 5. 7 x10-9 505. 9 274 7159 4160 1. 1 0. 66

A big constraint for the applications of NWs in which the thermal conductivity plays a ma‐ jor role is the lack of experimental data over a large range of diameters and temperatures, or in its defect the absence of a model supplying such information. In view of the difficulties inherent to thermal conductivity measurements at the nanoscale, and the variety of diame‐ ters and surface roughness, the development of a model capable of predicting the thermal conductivity in a wide range of diameters and surface roughness acquires a capital interest.

**Table 3.** Parameters used in the calculations of the thermal conductivity of rough NW of [2].

**4.5. A predictive expression for the thermal conductivity of Si NWs**

'

=13 nm (A); fitting of the experimental thermal

lengths, showing fitting to experimental data ( taken from [2]) for Λ<sup>0</sup>

mental data taken from [2].

252 Nanowires - Recent Advances

er authors [17].

The values of the speed of sound and cut-off temperature estimated for each NW diameter for which experimental data are available are plotted in Figure 8 (dots). In this figure one can observe that the behavior of the both velocities, the speed of sound of the longitudinal acoustic branch and the transverse acoustic branch respectively, have the same smooth form. Similar behavior is observed for the cut-off temperatures for each branch. This allows us to assume that those free parameters can be phenomenologically expressed as a function of the NW diameter. In the same figure, we show the nonlinear regression for each free pa‐ rameter, showing in all the cases an adjustment of *r* <sup>2</sup> >0. 99 to a generic function expressed as

$$f\left(D\right) = \frac{1+aD}{b+cD} \tag{25}$$

Therefore, with these phenomenological relations one is allowed to interpolate the values of the free parameters used in the calculation of the thermal conductivity of Si NWs in a range of diameters between the phonon confinement limit to slightly above the hundred of nms in diameter, using for that the adjustment parameters given in Table 4. Note that these parame‐ ters only represent numbers suitable for describing the sound velocities and the temperature cut-offs as a function of the NW diameter.

**Figure 8.** Plots of the cut-off temperatures, and sound speeds for the longitudinal and acoustic phonon branches vs the NW diameter. The symbols represent the values obtained by fitting the available experimental data. The lines are the fits using the phenomenological three parameter equation 25 in the text body.


**Table 4.** Parameter values from the non linear regressions to the data of Figure 8.

Using these parameters one can construct the thermal conductivity vs T plots for any NW diameter. These plots are shown in Figure 9-A. The thermal conductivity vs the NW diame‐ ter for two fixed temperatures, 300K and 600K, are represented in Figure 9-B; one observes that at high temperature the influence of the diameter on the thermal conductivity is flat‐ tened out with respect to what happens at lower temperatures; which roughly can be inter‐ preted in terms of the balance between boundary scattering processes and three phonon processes. In the low temperature range (T<300K) the boundary scattering is dominant, while in the high temperature range the three phonon scattering processes increase its weight.

**Figure 9.** Calculated thermal conductivity for arbitrary diameters using the parameters given in Table 3 (A). Thermal conductivities at 300K and 600K as a function of the NW diameter (B)

#### **5. Summary and perspectives**

**Figure 8.** Plots of the cut-off temperatures, and sound speeds for the longitudinal and acoustic phonon branches vs the NW diameter. The symbols represent the values obtained by fitting the available experimental data. The lines are

**a b c**

*vL* -2,362x10-3 1,478x10-4 -4,91310-7

*vT* -2,563x10-3 2,566x10-4 -9,41410-7

θ*<sup>L</sup>* 1,668 x10-1 3,671x10-3 2,969x10-4

θ*<sup>T</sup>* -5,153 x10-2 2,395x10-3 -1,621x10-4

Using these parameters one can construct the thermal conductivity vs T plots for any NW diameter. These plots are shown in Figure 9-A. The thermal conductivity vs the NW diame‐ ter for two fixed temperatures, 300K and 600K, are represented in Figure 9-B; one observes that at high temperature the influence of the diameter on the thermal conductivity is flat‐ tened out with respect to what happens at lower temperatures; which roughly can be inter‐ preted in terms of the balance between boundary scattering processes and three phonon processes. In the low temperature range (T<300K) the boundary scattering is dominant, while in the high temperature range the three phonon scattering processes increase its

the fits using the phenomenological three parameter equation 25 in the text body.

**Table 4.** Parameter values from the non linear regressions to the data of Figure 8.

weight.

254 Nanowires - Recent Advances

An overview of the thermal transport in semiconductor NWs has been presented. Both ex‐ perimental and theoretical aspects have been discussed, evidencing that the theoretical mod‐ els need of reliable thermal conductivity measurements to contrast the models, in order to advance in the understanding of the thermal transport in semiconductor NWs. Several clues need to be addressed in the close future; in particular, the role of optical phonons scattering, the scattering mechanisms in low diameter NWs, in connection with the phonon confine‐ ment phenomena; and a comprehensive view of the boundary scattering in rough surfaces, for which a breakdown of the Casimir limit occurs; also, the modeling of the thermal trans‐ port in alloyed NWs has to be addressed because of the importance of alloys for engineering the thermal conductivity. We have developed a model based on a modified Callaway-Hol‐ land formalism, which agglomerates the contributed ideas provided by different authors. This model permits an excellent fitting of the experimental results over an extended temper‐ ature range. Using the fitting of this model to the available experimental results one estab‐ lished empiric relations that permit to build up a predictive tool, which opens the access to the thermal conductivity of any Si NW diameter above the phonon confinement limit. This formalism can be extended to any other semiconductor without loss of generality, provided that reliable experimental thermal conductivity data for a few NWs diameters are available.

#### **Acknowledgement**

This work was funded by the Spanish Government (MAT-2007-66181-C03 and MAT-2010-20441-C02)

## **Author details**


## **References**


[12] Hippalgaonkar K, Huang B, Chen R, Sawyer K, Ercius P, Majumdar A. Fabrication of Microdevices with Integrated Nanowires for Investigating Low-Dimensional Phonon Transport. Nano Lett. 2010;10 4341.

**Author details**

256 Nanowires - Recent Advances

, J. Jiménez1

Nature 2008;451, 163.

and T. Rodríguez2

Nanowire Building Blocks. Science 2001;291, 851.

1 GdS Optronlab, Ed I+D, University of Valladolid, Valladolid, Spain

2 Electronic Technology, ETSIT, Polytechnic University of Madrid, Madrid, Spain

[1] Cui Y, Lieber C.M. Functional Nanoscale Electronic Devices Assembled using Silicon

[2] Hochbaum A.I, Chen R, Delgado R.D, Liang W, Garnett E.C, Najarian M, Majumdar A, Yang P. Rough Silicon Nanowires as High Performance Thermoelectric Materials.

[3] Kayes B.M, Atwater H.A, Lewis N.S. Comparison of the device physics principles of planar and radial p-n junction nanorod solar cells. J. Appl. Phys. 2005;97, 114302.

[4] Xie X.Q, Liu W.F, Oh J.I, Shen W.Z. Optical absorption in c-Si/a-Si:H core/shell nano‐ wire arrays for photovoltaic applications. Appl. Phys. Lett. 2011;99, 033107.

[5] Singh N, Buddharaju K.D, Manhas S.K, Agarwal A, Rustagi S.C, Lo G.Q, Balasubra‐ manian N, Kwong D.L. Si, SiGe Nanowire Devices by Top–Down Technology and

[6] Westover T, Jones R, Huang J.Y, Wang G, Lai E, Talin A.A. Photoluminescence, Ther‐ mal Transport, and Breakdown in Joule-Heated GaN Nanowires. Nano Lett. 2009;9,

[7] Shi L, Yao D, Zhang G, Li B. Large thermoelectric figure of merit in Si1−xGex nano‐

[8] Mingo N. Thermoelectric figure of merit of II–VI semiconductor nanowires. Appl.

[9] Mingo N. Thermoelectric figure of merit and maximum power factor in III–V semi‐

[10] Shi L, Li D, You C, Jang W, Kim D, Yao Z, Kim P, Majumdar A. Measuring Thermal and Thermoelectric Properties of One-Dimensional Nanostructures Using a Micro‐

[11] Li D, Wu Y, Kim P, Shi L, Yang P, Majumdar A. Thermal conductivity of individual

Their Applications. IEEE Trans. Electron Devices 2008;55, 3107.

wires. Appl. Phys. Lett. 2010;96, 173108.

conductor nanowires. Appl. Phys. Lett. 2004;84, 2652.

fabricated Device. J. Heat Transfer 2003;125, 881.

silicon nanowires. Appl. Phys. Lett. 2003;83, 2934.

Phys. Lett. 2004;85, 5986.

J. Anaya1

**References**

257.


[45] Alvarez F.X, Jou1 D. Size and frequency dependence of effective thermal conductivi‐ ty in nanosystems. J. Appl. Phys, 2008;103, 094321.

[27] Zou J, Balandin A. Phonon heat conduction in a semiconductor nanowire. J. Appl.

[29] Volz S.G, Chen G. Molecular dynamics simulation of thermal conductivity of silicon

[30] Mingo N. Calculation of Si nanowire thermal conductivity using complete phonon

[31] Mingo N, Yang L. Predicting the Thermal Conductivity of Si and Ge Nanowires.

[32] Herring C. Role of Low-Energy Phonons in Thermal Conduction. Phys. Rev 1954;95,

[33] Chung J.D, McGaughey A.J.H, Kaviany M. Role of Phonon Dispersion in Lattice Thermal Conductivity Modeling. Transactions of the ASME 2004;126, 376.

[34] Huang M, Chong W, Chang T. The lattice thermal conductivity of a semiconductor

[35] Liang L.H, Li B. Size-dependent thermal conductivity of nanoscale semiconducting

[36] Murphy P.G, Moore J.E. Coherent phonon scattering effects on thermal transport in

[37] Tian Z, Esfarjani K, Shiomi J, Henry A.S, Chen G. On the importance of optical pho‐ nons to thermal conductivity in nanostructures. Appl. Phys. Lett 2011;99, 053122.

[38] Ward A, Broido D.A. Intrinsic phonon relaxation times from first-principles studies

[39] Kazan M, Pereira S, Coutinho J, Correia M.R, Masri P. Role of optical phonon in Ge

[40] Kazan M, Guisbiers G, Pereira S, Correia M, Masri P, Bruyant A, Volz S, Royer P. Thermal conductivity of silicon bulk and nanowires: Effects of isotopic composition,

[41] Callaway J. Model for Lattice Thermal Conductivity at Low Temperatures. Phys.

[43] Carruthers P. Theory of Thermal Conductivity of Solids at Low Temperatures. Rev.

phonon confinement, and surface roughness. J. Appl. Phys 2010;107, 083503.

[42] Holland M.G. Analysis of Lattice Thermal Conductivity. Phys. Rev 1963;132, 2461.

[44] Srivastava G.P. The Physics of Phonons, Bristol:Adam Hilger; 1990.

of the thermal conductivities of Si and Ge. Phys. Rev B 2010;81, 085205.

thin semiconductor nanowires. Phys. Rev. B 2007;76, 155313.

thermal conductivity. Appl. Phys. Lett 2008;92, 211903.

[28] Ziman J.M. Electrons and Phonons New York: Oxford University Press; 1967.

Phys. 2001;89, 2932.

Nano Lett. 2003;3, 1713.

Rev. 1959;113, 1046.

Mod. Phys 1961;33, 92.

954.

258 Nanowires - Recent Advances

nanowires. Appl. Phys. Lett. 1999;75 2056.

nanowire. J. Appl. Phys. 2006;99 114318.

systems. Phys. Rev. B 2006;73, 153303.

dispersion relations. Phys. Rev. B 2003;68, 113308.


## **Carbon Based Nanowires**

[61] Morelli D.T., Heremans J.P, Slack G.A. Estimation of the isotope effect on the lattice thermal conductivity of group IV and group III-V semiconductors. Phys. Rev B

[62] Peierls R. E. Quantum Theory of Solids New York: Oxford University Press; 1955.

[64] Klemens P.G. Solid State Physics, Vol 7, p.1 New York: Academic; 1958.

mal Conductance of Silicon Nanowire. Nano Lett. 2009;9, 1861.

[63] Sharma P.C, Frank Rose M. Three-phonon scattering processes and their role in pho‐ non thermal conductivity of silicon. Journal Solid State Chemistry 1998;73, 92.

[65] Slack G. A. Thermal Conductivity of CaF2, MnF2, CoF2, and ZnF2 Crystals. Phys.

[66] Asen-Palmer M. Bartkowski K, Gmelin E, Carona M, Zhernov A.P, Inyushkin V.A, Taldenkov A, Ozhogin V.I, Itoh K.M, Haller E.E. Thermal conductivity of germani‐

[67] Heron J. S, Fournier T, Mingo N, Bourgeois O. Mesoscopic Size Effects on the Ther‐

[68] Soffer S.B. Statistical Model for the Size Effect in Electrical Conduction. J. Appl. Phys.

[69] Omar M.S, Taha H.T. Lattice dislocation in Si nanowires. Physica B 2009;404, 5203.

um crystals with different isotopic compositions. Phys. Rev. B 1997;56, 9431.

202;66, 195304.

260 Nanowires - Recent Advances

Rev. 1961;122, 1451.

1967;38, 1710.

## **Carbon Nanotubes Filled With Ternary Chalcohalides**

Marian Nowak and Marcin Jesionek

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52590

## **1. Introduction**

Since the discovery by S. Iijima in 1991 [1] various potential applications have been pro‐ posed for carbon nanotubes (CNTs): sensors, field emission displays, nanometer-sized semi‐ conductor devices and hydrogen storage media. There is a huge literature stream related to nanotube research. On a fundamental level, there are still challenges to mass-produce con‐ trolled nanostructures at reasonable cost and new features. There is also a great demand for control versatile electronic characteristics of CNTs. One strategy is to use the CNTs them‐ selves, controlling useful properties *via* their radii and morphologies. An alternative ap‐ proach leading to new features of CNTs, i.e. directional action on their versatile electronic characteristics, is based on filling them with condensed substances from a wide range of ma‐ terials [2-8]. CNTs are sp2 graphene carbon cylinders capable of hosting a variety of species, including 1D crystals of metals, metal salts and oxides; semiconductors; superconductors; and chains of fullerene or endofullerene molecules (see Refs. cited in [7,8]). Such objects are distinguished in their unique physical and chemical properties from both hollow nanotubes and the encapsulated substances, which permits one to purpose-tailor "nanowires" and "nanotubes" with unique physical and chemical properties.

Among materials encapsulated within CNTs are both elementary substances and the binary compounds formed from the group 15–16–17 elements. Figure 1 presents a typical high res‐ olution transmission electron microscopy (HRTEM) image of single-walled carbon nanotube (SWNT) filled with one of the metal halogenides, i.e., BaI2@SWNT. However, one should note the possible filling of SWNTs with the alone helical iodine chains (Figure 2). Of course, the multi-walled carbon nanotubes (MWNTs) can be filled with the elemental forms (e.g., Bi, Sb, S, Se, Te, I2), halogenides (e.g., consistent of Cl, Br, I2), oxides (e.g., Sb2O3), and chalcoge‐ nides (e.g., Sb2S3, CdS, SnSe, SnTe, HgTe), too. A comprehensive list of such carbon nano‐ structures has been reviewed by A. Eliseev at al. [7] and P. Lukanow et al. [8].

© 2012 Nowak and Jesionek; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Nowak and Jesionek; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

**Figure 1.** a) HRTEM image of BaI2/SWNT composite. b) Detail from boxed region in (a), for which image noise was reduced with a deconvolution filter. c) Structure plot based on a 2D peak- mapping analysis of (b); end-on view on the right. d) Simulated HRTEM image, e) complete structure model of BaI2/SWNT composite. f) Coordination model of 1D BaI2 chain. g) 5- coordinate, h) 6-coordinate BaI6 polyhedra and ball-and-stick models derived from (c). i) and j) Models showing derived equatorial bond angles and distances for BaI4 units along the center of the 1D BaI2 chain (ionic radii in (j) indicated by diffuse spheres) (Reproduced from [9] with permission. Copyright Wiley-VCH Verlag GmbH).

**Figure 2.** a) Higher resolution Z-contrast image of a SWNT containing two strands of iodine. (b) Maximum entropy processed image to reduce the noise. Two constrictions can be seen, consistent with the projection of a double helix configuration with the two chains spiraling along the inside walls of the nanotube as shown in the overlay in (c). The period of the helix is about 5 nm, and the maximum separation of the two strands is ~0.65 nm, much less than the 1.3 -1.4 nm diameter of the SWNT (Reproduced from [10] with permission. Copyright the American Physical Society).

This review summarizes results of recent investigations [11-14] of carbon nanotubes (CNTs) filled with ternary chalcohalides formed from the group 15–16–17 elements, i.e. antimony sulfoiodide (SbSI) and antimony selenoiodide (SbSeI) known as SbSI-type materials. Some recent original data are presented, too. It should be underlined that the bulk SbSI being a semiconducting ferroelectric has an unusually large number of very attractive and suitable properties (see Refs. cited in [15-17]). Among them there are the photoferroelectricity, pyro‐ electric, pyrooptic, piezoelectric, electromechanical, electrooptic, photorefractive and nonlin‐ ear optical effects. Therefore SbSI is taken into consideration as a valuable material for many applications (see the literature cited in [15-17]).

SbSI type materials have been synthesized in different ways (see Refs. in [15-17]. The most often used method consists in fusion of stoichiometric amounts of antimony, sulfur and io‐ dine or antimony iodide, and antimony sulphide with antimony iodide. The procedure re‐ quires high temperature (723 - 873 K) and long reaction time (1-3 days). Recently [15,17], a novel sonochemical method for direct preparation of semiconducting and ferroelectric SbSI nanowires has been established. The determined [15,17] value of the indirect forbidden en‐ ergy band gap of SbSI gel EgIf=1.829(27) eV is well compared to the bulk value of band gap of SbSI reported in the literature (see Refs. in [15,17]). The maximum of dielectric constant ε=1.6·10<sup>4</sup> of SbSI nanowires was observed at Curie temperature Tc=292(1) K [17,18] that well corresponds with the phase transition in bulk SbSI crystals. It should be underlined that SbSeI can be sonochemically prepared in the form of crystalline nanowires [17,19], too.

### **2. Preparation of SbSI@CNTs and SbSeI@CNTs**

**Figure 1.** a) HRTEM image of BaI2/SWNT composite. b) Detail from boxed region in (a), for which image noise was reduced with a deconvolution filter. c) Structure plot based on a 2D peak- mapping analysis of (b); end-on view on the right. d) Simulated HRTEM image, e) complete structure model of BaI2/SWNT composite. f) Coordination model of 1D BaI2 chain. g) 5- coordinate, h) 6-coordinate BaI6 polyhedra and ball-and-stick models derived from (c). i) and j) Models showing derived equatorial bond angles and distances for BaI4 units along the center of the 1D BaI2 chain (ionic radii in (j) indicated by diffuse spheres) (Reproduced from [9] with permission. Copyright Wiley-VCH Verlag GmbH).

264 Nanowires - Recent Advances

**Figure 2.** a) Higher resolution Z-contrast image of a SWNT containing two strands of iodine. (b) Maximum entropy processed image to reduce the noise. Two constrictions can be seen, consistent with the projection of a double helix configuration with the two chains spiraling along the inside walls of the nanotube as shown in the overlay in (c). The period of the helix is about 5 nm, and the maximum separation of the two strands is ~0.65 nm, much less than the 1.3 -1.4 nm diameter of the SWNT (Reproduced from [10] with permission. Copyright the American Physical Society).

There are known a few methods for filling CNTs with different substances [4,6-8]: catalytic synthesis of nanotubes using the metals as catalysts, capillary drawing-in of molten materi‐ als or materials dissolved in solvents having a low surface tension, saturation with metal vapor as well as electrochemical methods based on passing the electrical current through an electro‐ lyte containing dissolved metal atoms. In this paper we present another method for insert‐ ing materials into the inner cavity of a nanotube. Our method is based on sonochemistry. It is well known [20-25] that ultrasound can induce new reactivities leading to the formation of unexpected chemical species. What makes sonochemistry unique is the remarkable phenom‐ enon of cavitation [25,26]. Comparing sonochemical method of preparing materials with the traditional ones, it can be seen that ultrasound irradiation can be used at room temperature and ambient pressure to promote heterogeneous reactions that normally occur only under extreme conditions of hundreds of atmospheres and degrees (e.g., [20-22]).

**Figure 3.** Schematic illustration of the set-up for ultrasonic filling of CNTs with SbSI and SbSeI (RS– ultrasonic reactor; DC30-K20 – refrigerated circulator bath; W1, W2 – thermal cycle; C – stand; S –sample; V – Pyrex glass cylinder; T – temperature controller; t – temperature sensor) [13].

Both SbSI and SbSeI were prepared in CNTs ultrasonically from the constituents (the ele‐ ments Sb, I, and S or Se) [11-14]. Methanol and ethanol served as the solvents for these reac‐ tions, respectively. All the reagents used in experiments were of analytical purity and were used without further purification. Antimony (99.95 %), selenium (99.5 %) and multi–walled CNTs (659258–10G, 90+%) were purchased from Sigma–Aldrich. Sublimated sulfur (pure p.a.), iodine (pure p.a.), absolute methanol (pure p.a.), and absolute ethanol (pure p.a.) were purchased from POCH S.A. (Gliwice, Poland). In a typical procedure of fabrication SbSI@CNTs [11], the elemental mixture with stoichiometric ratio of e.g. 0.380 g Sb, 0.099 g S and 0.394 g I, was immersed with 0.282 g of CNTs in 40 ml absolute methanol, which was contained in a 54 ml Pyrex glass cylinder of 20 mm inside diameter. In the first procedure of fabrication SbSeI@CNTs [14], the elemental mixture with stoichiometric ratio of e.g. 0.156 g Sb, 0.101 g Se and 0.165 g I2, was immersed with 0.102 g of CNTs in 10 ml absolute ethanol, which was contained in a 54 ml Pyrex glass cylinder of 20 mm inside diameter. Further‐ more, the syntheses of SbSeI in CNTs have been done in [14] for twice and four times greater contents of Se in the mixture (the obtained products are discussed in the next section). Ves‐ sels with the substrates were closed during experiments [11, 14] to prevent volatilization of the precipitants in long time tests. Bottoms of the vessels were planar or semispherical and 1 mm in thickness. Figure 3 presents scheme of the technological set-up. The cylinders were partly submerged in water in an ultrasonic reactor (InterSonic IS–UZP–2, frequency 35 kHz, with 75 W electrical power and 2.6 W/cm2 power density guaranteed by the manufacturer). Temperature of the bath was established using the Haake DC–K20 refrigerated circulator bath (Thermo Scientific). The sonochemical processes were continued for 3 h at 323 K tem‐ perature of the water in the ultrasonic reactor. During the sonications sols were formed. It was observed that the color of the suspension changed gradually indicating the growth process of the SbSI and SbSeI. To control this process, measurements of optical diffusive re‐ flectance Rd(λ) of the sample were performed (Figure 4). It was assumed that the sonochemi‐ cal process is finished when the spectral characteristics of Rd(λ) do not change with time.

**Figure 4.** Change of diffusive reflectance during the sonication of Sb, S and I in methanol (a) and of Sb, Se and I in ethanol (b) ( – after 5 min. of sonication, T=323 K; – 20 min. – 40 min. – 60 min.; – 80 min.;, – 120 min.; – 180 min.; – after washing) [13].

**Figure 3.** Schematic illustration of the set-up for ultrasonic filling of CNTs with SbSI and SbSeI (RS– ultrasonic reactor; DC30-K20 – refrigerated circulator bath; W1, W2 – thermal cycle; C – stand; S –sample; V – Pyrex glass cylinder; T –

Both SbSI and SbSeI were prepared in CNTs ultrasonically from the constituents (the ele‐ ments Sb, I, and S or Se) [11-14]. Methanol and ethanol served as the solvents for these reac‐ tions, respectively. All the reagents used in experiments were of analytical purity and were used without further purification. Antimony (99.95 %), selenium (99.5 %) and multi–walled CNTs (659258–10G, 90+%) were purchased from Sigma–Aldrich. Sublimated sulfur (pure p.a.), iodine (pure p.a.), absolute methanol (pure p.a.), and absolute ethanol (pure p.a.) were purchased from POCH S.A. (Gliwice, Poland). In a typical procedure of fabrication SbSI@CNTs [11], the elemental mixture with stoichiometric ratio of e.g. 0.380 g Sb, 0.099 g S and 0.394 g I, was immersed with 0.282 g of CNTs in 40 ml absolute methanol, which was contained in a 54 ml Pyrex glass cylinder of 20 mm inside diameter. In the first procedure of fabrication SbSeI@CNTs [14], the elemental mixture with stoichiometric ratio of e.g. 0.156 g Sb, 0.101 g Se and 0.165 g I2, was immersed with 0.102 g of CNTs in 10 ml absolute ethanol, which was contained in a 54 ml Pyrex glass cylinder of 20 mm inside diameter. Further‐ more, the syntheses of SbSeI in CNTs have been done in [14] for twice and four times greater contents of Se in the mixture (the obtained products are discussed in the next section). Ves‐ sels with the substrates were closed during experiments [11, 14] to prevent volatilization of the precipitants in long time tests. Bottoms of the vessels were planar or semispherical and 1 mm in thickness. Figure 3 presents scheme of the technological set-up. The cylinders were partly submerged in water in an ultrasonic reactor (InterSonic IS–UZP–2, frequency 35 kHz,

Temperature of the bath was established using the Haake DC–K20 refrigerated circulator bath (Thermo Scientific). The sonochemical processes were continued for 3 h at 323 K tem‐ perature of the water in the ultrasonic reactor. During the sonications sols were formed. It was observed that the color of the suspension changed gradually indicating the growth process of the SbSI and SbSeI. To control this process, measurements of optical diffusive re‐ flectance Rd(λ) of the sample were performed (Figure 4). It was assumed that the sonochemi‐ cal process is finished when the spectral characteristics of Rd(λ) do not change with time.

power density guaranteed by the manufacturer).

temperature controller; t – temperature sensor) [13].

266 Nanowires - Recent Advances

with 75 W electrical power and 2.6 W/cm2

When the sonication processes were finished, a dark sols were obtained. They were centri‐ fuged using the MPW–223e centrifuge, MPW Med. Instruments (Poland), to extract the prod‐ ucts. Then the liquids above the sediments were replaced with pure methanol or ethanol to wash the precipitates in the cases of SbSI@CNTs and SbSeI@CNTs, respectively. The centrifu‐ gation and washing were performed 5 times in both cases. At the end methanol as well as ethanol were evaporated from the samples during the drying in air at room temperature, so a brown-purple (Figure 5a) and dark brown–purple (Figure 5b) substances were obtained in the cases of SbSI@CNTs and SbSeI@CNTs, respectively.

**Figure 5.** Dried MWCNTs filled ultrasonically with SbSI in methanol and SbSeI in ethanol [13].

## **3. Morphology, composition and crystal structure of SbSI@CNTs and SbSeI@CNTs**

Characterization of the multi–walled CNTs filled with SbSI or SbSeI [11-14] was accomplish‐ ed using different techniques, such as powder X–ray diffraction (XRD) performed on a JEOL JDX-7S X-ray diffractometer with graphite monochromatized Cu Kα radiation (λ = 0.154056 nm), scanning electron microscopy (SEM) and energy dispersive X–ray spectroscopy (EDS) performed on a Hitachi S-4200 microscope with EDS Thermo Scientific spectrometer, high– resolution transmission electron microscopy (HRTEM) and selected area electron diffraction (SAED) performed on a JEOL-JEM 3010 microscope.

**Figure 6.** The powder XRD pattern of dried CNTs filled with SbSI ultrasonically in methanol [11].

**Figure 7.** The powder XRD patterns of dried multi–walled CNTs filled with SbSeI ultrasonically in ethanol with differ‐ ent compositions of the reactants [14] (1- 0.101 g Se, 2- 0.202 g Se, 3- 0.404 g Se; 0.156 g Sb, 0.165 g I2, and 0.102 g of CNTs in 10 ml absolute ethanol; described peaks correspond to the crystalline SbSeI [27,28], C [29,30] and SbI3 [31,32]).

The powder XRD pattern of the MWNTs filled with SbSI is shown in Figure 6. The welldefined, sharp diffraction lines suggest the well-crystallized substance. It was found that the diffraction lines can be divided into three groups. In the first group, containing most of the lines, the peaks can be indexed to be a pure orthorhombic phase for SbSI with the cell con‐ stants *a=0*.858 nm, *b=*1.017 nm, and *c=*0.414 nm. The identification was done using the PCSI‐ WIN computer program and the data from JCPDS-International Centre for Diffraction Data 2000. The intensities and positions of the peaks are in good agreement with literature values for SbSI [33]. The second group of diffraction lines can be indexed to be a carbon phase P63mc with the cell constants *a=0*.2470 nm and *c=*0.6790 nm [30]. The third group of a few addition‐ al X-ray diffraction lines for 2θ equal 16.51°, 18.66° and 29.00° has given the following val‐ ues of *d hkl*: 0.536 nm, 0.475 nm, and 0.308 nm, respectively. These lines may be explained as follows. The XRD investigations [34] of SbSI sonochemically produced in methanol show the coexistence of phases with Pna21 and *Pnam* crystal symmetry that are characteristic for ferro‐ electric and paraelectric domains, respectively. Hence, such data obtained at 298 K represent a structure of just below or very near the transition temperature of SbSI. Another explana‐ tion for the additional X-ray diffraction lines can be given taking into account existence of unknown phase in the investigated material. This problem must be studied in the future.

resolution transmission electron microscopy (HRTEM) and selected area electron diffraction

**Figure 6.** The powder XRD pattern of dried CNTs filled with SbSI ultrasonically in methanol [11].

**Figure 7.** The powder XRD patterns of dried multi–walled CNTs filled with SbSeI ultrasonically in ethanol with differ‐ ent compositions of the reactants [14] (1- 0.101 g Se, 2- 0.202 g Se, 3- 0.404 g Se; 0.156 g Sb, 0.165 g I2, and 0.102 g of CNTs in 10 ml absolute ethanol; described peaks correspond to the crystalline SbSeI [27,28], C [29,30] and SbI3 [31,32]).

The powder XRD pattern of the MWNTs filled with SbSI is shown in Figure 6. The welldefined, sharp diffraction lines suggest the well-crystallized substance. It was found that the diffraction lines can be divided into three groups. In the first group, containing most of the lines, the peaks can be indexed to be a pure orthorhombic phase for SbSI with the cell con‐ stants *a=0*.858 nm, *b=*1.017 nm, and *c=*0.414 nm. The identification was done using the PCSI‐ WIN computer program and the data from JCPDS-International Centre for Diffraction Data 2000. The intensities and positions of the peaks are in good agreement with literature values for SbSI [33]. The second group of diffraction lines can be indexed to be a carbon phase P63mc with the cell constants *a=0*.2470 nm and *c=*0.6790 nm [30]. The third group of a few addition‐

(SAED) performed on a JEOL-JEM 3010 microscope.

268 Nanowires - Recent Advances


**Table 1.** Comparison of interplanar spacings determined by XRD (Figure 7), SAED (Figure 14) and HRTEM (Figure 12) of multi–walled CNTs filled with SbSeI sonochemically in ethanol with literature data for CNTs, SbSeI orthorhombic crystals, and SbI3 monoclinic crystals.

The powder XRD patterns of the CNTs filled ultrasonically with SbSeI are shown in Figure 7. One can see that the best results (the lowest intensity of SbI3 peaks) have been obtained for four times greater content of Se in the mixture (compare curves in Figure 7). These results indicate that the CNTs should be filled with SbSeI with the excess of Se. The well–defined, sharp diffraction lines suggest the well–crystallized substance (curve 3 in Figure 7). It was found that the diffraction lines can be divided into three groups (Table 1). In the first group, containing most of the lines, the peaks can be indexed to be a pure orthorhombic phase for SbSeI with the cell constants *a=0.8698* nm, *b=1.0412* nm, and *c=0.4127* nm [27,28]. The intensi‐ ties and positions of the peaks are in good agreement with literature values for SbSeI [27,28]. The second group of diffraction lines can be indexed to be a carbon phase P63mc with the cell constants *a=0.2470* nm and *c=0.6790* nm [29,30]. The third group of a few additional X–ray diffraction lines can be attributed to the monoclinic antimony iodide (SbI3) with the cell constants *a=0.7281* nm, *b=1.0902* nm, *c=0.8946* nm, and *β=109.930 °* [31,32]. It can be explained as fol‐ lows. In ethanol iodine reacts with antimony forming antimony iodide (1). This substance is soluble in alcohols [35]. When the alcohol is evaporated, the SbI3 crystallizes. It is probable that (despite the washing of the CNTs sonochemically filled with SbSeI) some portion of SbI3 still exists in the product. This problem should be solved in the future.

**Figure 8.** The typical SEM micrographs of dried gels of SbSI@CNTs [11] and SbSeI@CNTs [14].

**Figure 9.** Typical TEM images of an individual SbSI@CNT [11] and SbSeI@CNT [14].

The typical SEM micrographs of the dried SbSI@CNTs and SbSeI@CNTs prepared sono‐ chemically in methanol and ethanol, respectively, are shown in Figures 8 and 9. The samples used for SEM and EDS observations were prepared by dispersing a small quantity of the CNTs filled with SbSI or SbSeI in ethanol followed by ultrasonic vibration for 10 min. One or two drops of the nanoparticle solutions were deposited on naturally oxidized silicon single crystal plate and dried in 120 min at room temperature in vacuum. One can see SbI3 nano‐ particles around some of the SbSeI@CNTs (Figure 9). Probably, these nanoparticles have been grown from the solution (SbI3 dissolved in ethanol) during drying of the product. The SbI3 by-product is the rest of intermediate compound obtained during sonochemical synthe‐ sis of SbSeI [36]. The dimension of SbI3 nanoparticle has been about 30 nm. The chemical composition and crystal structure of these nanoparticles have been proved [14].

The powder XRD patterns of the CNTs filled ultrasonically with SbSeI are shown in Figure 7. One can see that the best results (the lowest intensity of SbI3 peaks) have been obtained for four times greater content of Se in the mixture (compare curves in Figure 7). These results indicate that the CNTs should be filled with SbSeI with the excess of Se. The well–defined, sharp diffraction lines suggest the well–crystallized substance (curve 3 in Figure 7). It was found that the diffraction lines can be divided into three groups (Table 1). In the first group, containing most of the lines, the peaks can be indexed to be a pure orthorhombic phase for SbSeI with the cell constants *a=0.8698* nm, *b=1.0412* nm, and *c=0.4127* nm [27,28]. The intensi‐ ties and positions of the peaks are in good agreement with literature values for SbSeI [27,28]. The second group of diffraction lines can be indexed to be a carbon phase P63mc with the cell constants *a=0.2470* nm and *c=0.6790* nm [29,30]. The third group of a few additional X–ray diffraction lines can be attributed to the monoclinic antimony iodide (SbI3) with the cell constants

lows. In ethanol iodine reacts with antimony forming antimony iodide (1). This substance is soluble in alcohols [35]. When the alcohol is evaporated, the SbI3 crystallizes. It is probable that (despite the washing of the CNTs sonochemically filled with SbSeI) some portion of SbI3

[31,32]. It can be explained as fol‐

*a=0.7281* nm, *b=1.0902* nm, *c=0.8946* nm, and *β=109.930 °*

270 Nanowires - Recent Advances

still exists in the product. This problem should be solved in the future.

**Figure 8.** The typical SEM micrographs of dried gels of SbSI@CNTs [11] and SbSeI@CNTs [14].

**Figure 9.** Typical TEM images of an individual SbSI@CNT [11] and SbSeI@CNT [14].

**Figure 10.** The EDS spectra of dried multi–walled CNTs filled ultrasonically with SbSI [11] and SbSeI [14] (the investi‐ gated material was deposited on naturally oxidized Si single crystal).


The EDS spectra of dried multi–walled CNTs filled ultrasonically with SbSI and SbSeI are presented in Figure 10. The measured atomic concentrations of Sb, S, Se, I and C (without Si and O from the supporting wafer) are presented in Table 2. Taking into consideration the large excess of Sb and the deficit of S or Se as well as I detected by EDS, it should be under‐ lined that the EDS analysis performed on sonochemically produced alone SbSeI gel con‐ firmed an elemental atomic ratio of 0.41:0.26:0.33 for Sb, Se, and I [19]. Therefore, as it was suggested in [19], probably a separate phase of Sb and I is present, and presumably it is ad‐ sorbed on the surface of the SbSeI. A similar case was recently discovered in the XPS investi‐ gations of sonochemically prepared SbSI nanowires [37]. It is improbable that the excess Sb is present within the ordered SbSI and SbSeI nanowires inside the CNTs, since the interpla‐ nar spacings corresponds to stoichiometric materials (Figures 11 and 12). In [37] the antimo‐ ny subiodide was suggested as the hypothetical material of the surface layer on SbSI nanowires. Such Sb3I subiodide was obtained sonochemically [38]. Probably, there is the same compound on the surface of sonochemically prepared SbSeI nanowires. The existence of this substance will be investigated in the near future.

**Figure 11.** HRTEM image of an individual MWCNT filled with SbSI sonochemically in methanol. The fringe spacings of 0.319(2) nm (1) and 0.209(2) nm (2) correspond to the interplanar distances between the (220) planes of SbSI crystal and (101) planes of carbon nanotube, respectively [11].

The TEM (Figure 9) and HRTEM images (Figures 11 and 12) of an individual CNTs sono‐ chemically filled with SbSI and SbSeI reveal that the products consist of coaxial nanocables. The lateral dimensions of the SbSI@CNTs and SbSeI@CNTs have been in the ranges from 30 to 200 nm and from 20 to 170 nm, respectively [11,14]. Lengths of these nanocables reach up to several micrometers in both cases. The HRTEM images of individual CNTs sonochemical‐ ly filled with SbSI (Figure 11) exhibit good crystallinity of the SbSI and its clear (220) lattice fringes parallel to the nanocable axis. It indicates the growth of SbSI inside the CNT in [001] direction. The interplanar spacing is about 0.319(2) nm, which coincide with the interplanar spacing 0.32494 nm of (220) planes of the orthorhombic structure of conventional SbSI [33]. Figure 11 shows also the lattice fringes of the CNT walls. The fringe spacings of 0.209(2) nm match with the 0.21390 nm interplanar distances between the (101) planes of carbon nano‐ tubes [30]. All these results correspond well with the XRD patterns (Figure 6) of the CNTs sonochemically filled with SbSI. The SAED pattern (Figure 13) recorded on the end of multi– walled CNT filled with SbSI (presented in Figure 9a) indicates the interplanar spacings ap‐ propriate for CNTs as well as SbSI crystals (see Table 3).

The EDS spectra of dried multi–walled CNTs filled ultrasonically with SbSI and SbSeI are presented in Figure 10. The measured atomic concentrations of Sb, S, Se, I and C (without Si and O from the supporting wafer) are presented in Table 2. Taking into consideration the large excess of Sb and the deficit of S or Se as well as I detected by EDS, it should be under‐ lined that the EDS analysis performed on sonochemically produced alone SbSeI gel con‐ firmed an elemental atomic ratio of 0.41:0.26:0.33 for Sb, Se, and I [19]. Therefore, as it was suggested in [19], probably a separate phase of Sb and I is present, and presumably it is ad‐ sorbed on the surface of the SbSeI. A similar case was recently discovered in the XPS investi‐ gations of sonochemically prepared SbSI nanowires [37]. It is improbable that the excess Sb is present within the ordered SbSI and SbSeI nanowires inside the CNTs, since the interpla‐ nar spacings corresponds to stoichiometric materials (Figures 11 and 12). In [37] the antimo‐ ny subiodide was suggested as the hypothetical material of the surface layer on SbSI nanowires. Such Sb3I subiodide was obtained sonochemically [38]. Probably, there is the same compound on the surface of sonochemically prepared SbSeI nanowires. The existence

**Figure 11.** HRTEM image of an individual MWCNT filled with SbSI sonochemically in methanol. The fringe spacings of 0.319(2) nm (1) and 0.209(2) nm (2) correspond to the interplanar distances between the (220) planes of SbSI crystal

The TEM (Figure 9) and HRTEM images (Figures 11 and 12) of an individual CNTs sono‐ chemically filled with SbSI and SbSeI reveal that the products consist of coaxial nanocables. The lateral dimensions of the SbSI@CNTs and SbSeI@CNTs have been in the ranges from 30 to 200 nm and from 20 to 170 nm, respectively [11,14]. Lengths of these nanocables reach up to several micrometers in both cases. The HRTEM images of individual CNTs sonochemical‐ ly filled with SbSI (Figure 11) exhibit good crystallinity of the SbSI and its clear (220) lattice fringes parallel to the nanocable axis. It indicates the growth of SbSI inside the CNT in [001] direction. The interplanar spacing is about 0.319(2) nm, which coincide with the interplanar

of this substance will be investigated in the near future.

272 Nanowires - Recent Advances

and (101) planes of carbon nanotube, respectively [11].

**Figure 12.** HRTEM image of an individual multi–walled CNT filled with SbSeI sonochemically in ethanol. The fringe spacings of 0.332(5) nm (sign 1) and 0.432(13) nm (sign 2) correspond to the interplanar distances between the (002) planes of carbon nanotube and (200) planes of SbSeI crystal, respectively [14].

**Figure 13.** SAED pattern of MWCNT filled with SbSI ultrasonically in methanol [11] (shown in Figure 9). The diffraction patterns correspond to the interplanar distances presented in Table 3.


**Table 3.** Comparison of interplanar spacings determined by SAED (Figure 13) of multi–walled CNT filled with SbSI ultrasonically in methanol with literature data for CNTs and SbSI bulk crystals.

The HRTEM images of an individual CNTs sonochemically filled with SbSeI (Figure 12) ex‐ hibit good crystallinity of the SbSeI and its clear (200) lattice fringes parallel to the nanocable axis. It indicates the growth of SbSeI inside the CNT in [001] direction. The interplanar spac‐ ing is about 0.432(13) nm, which coincide with the interplanar spacing 0.4349 nm of (200) planes of the orthorhombic structure of conventional SbSeI [28]. Figure 12 shows also the lattice fringes of the CNT walls. The fringe spacings of 0.332(5) nm match with the 0.3395 nm interplanar distances between the (002) planes of carbon nanotubes [30]. All these results correspond well with the XRD patterns (Figure 7, Table 1) of the CNTs sonochemically filled with SbSeI. The SAED pattern (Figure 14) recorded on the multi–walled CNT filled with SbSeI (presented in Figure 12) indicates the interplanar spacings appropriate for SbSeI crys‐ tals (see Table 1). Surprisingly, the presence of SbSeI nanowires or SbSeI material on the out‐ er walls of the MWCNT has not been observed in the HRTEM and TEM images. Probably, the process of capillary suction of molecules into carbon nanotubes facilitates the process of SbSeI nanocrystal growth.

**Sign Results of the**

274 Nanowires - Recent Advances

**SAED dhkl (nm)**

1 reflex 0.4360 –

3 reflex 0.1446 –

7 reflex 0.1190 –

9 circle 0.1155 0.11606

ultrasonically in methanol with literature data for CNTs and SbSI bulk crystals.

SbSeI nanocrystal growth.

**Literature data** for C [29, 30] for SbSI [27, 33] dhkl (nm) (hkl) dhkl (nm) (hkl)

> 0.43402 0.42450

> 0.14244 0.14244

> 0.12031 0.11960

> > – – –

(120) (200)

(422) (531)

(143) (181)

> – – –

– –

– –

– –

(112) (105) (006)

–

–

–

0.11464 0.11316

8 circle 0.1210 0.12350 (110) – –

10 circle 0.1036 0.10425 (202) – – 11 circle 0.07874 0.07954 (122) – – 12 circle 0.07038 0.07720 (206) – –

**Table 3.** Comparison of interplanar spacings determined by SAED (Figure 13) of multi–walled CNT filled with SbSI

The HRTEM images of an individual CNTs sonochemically filled with SbSeI (Figure 12) ex‐ hibit good crystallinity of the SbSeI and its clear (200) lattice fringes parallel to the nanocable axis. It indicates the growth of SbSeI inside the CNT in [001] direction. The interplanar spac‐ ing is about 0.432(13) nm, which coincide with the interplanar spacing 0.4349 nm of (200) planes of the orthorhombic structure of conventional SbSeI [28]. Figure 12 shows also the lattice fringes of the CNT walls. The fringe spacings of 0.332(5) nm match with the 0.3395 nm interplanar distances between the (002) planes of carbon nanotubes [30]. All these results correspond well with the XRD patterns (Figure 7, Table 1) of the CNTs sonochemically filled with SbSeI. The SAED pattern (Figure 14) recorded on the multi–walled CNT filled with SbSeI (presented in Figure 12) indicates the interplanar spacings appropriate for SbSeI crys‐ tals (see Table 1). Surprisingly, the presence of SbSeI nanowires or SbSeI material on the out‐ er walls of the MWCNT has not been observed in the HRTEM and TEM images. Probably, the process of capillary suction of molecules into carbon nanotubes facilitates the process of

4 reflex 0.3732 – – 0.38465 (011) 5 reflex 0.3470 0.33950 (002) 0.35036 (111) 2 reflex 0.2189 0.21390 (100) 0.21663 (330) 6 0.2089 0.20402 (101) 0.20800 (002)

**Figure 14.** Electron diffraction pattern of multi–walled CNT filled with SbSeI ultrasonically in ethanol [14] in the orien‐ tation close to the [210] zone axis (a) and its simulated diagram (b). The diffraction patterns correspond to the inter‐ planar distances presented in Table 1.

## **4. Mechanisms of sonochemical preparation of SbSI@CNTs and SbSeI@CNTs**

It is known that the mode of insertion dictates the nature and morphology of the obtained filling of CNTs [39]. When the filling is induced via solution–deposition, small discrete en‐ capsulates are obtained, whereas when it is obtained via capillarity, continuously filled CNTs are observed. Probably the last happens when CNTs are filled sonochemically by SbSI [11] as well as by SbSeI [14]. As in the case of sonochemically produced alone SbSI [15] and SbSeI [19], the transient high–temperature and high–pressure field produced during ultra‐ sound irradiation provide a favorable environment for the 1D growth of the SbSI and SbSeI nanocrystals from elements inside multi–walled CNTs in ethanol, though the bulk solution surrounding the collapsing bubbles is at relatively low temperature (T=323 K) and atmos‐ pheric pressure.

The probable reaction routes of synthesis of SbSI and SbSeI in CNTs, and the mechanisms of nanowires formation using elemental Sb, S or Se and I in the presence of methanol and etha‐ nol under ultrasonic irradiation can be summarized as follows:

**a.** iodine, I2, dissolved in methanol or ethanol reacts with antimony and forms the antimo‐ ny triiodide, SbI3, also dissolved in ethanol [40]

$$\text{2Sb} + \text{3I}\_2 \rightarrow \text{2SbI}\_3 \tag{1}$$

**b.** dehydrogenation, dehydration as well as decomposition of ethanol (2) and methanol (3) in or close to the cavitation bubbles leads to the formation of hydrogen and water [41,42]

$$\begin{aligned} \text{C}\_2\text{H}\_3\text{OH} &\leftrightarrow \text{C}\_2\text{H}\_4\text{O} + \text{H}\_2 \\ \text{C}\_2\text{H}\_3\text{OH} &\leftrightarrow \text{C}\_2\text{H}\_4 + \text{H}\_2\text{O} \\ \text{C}\_2\text{H}\_3\text{OH} &\leftrightarrow \text{C}\_2\text{H}\_3\text{O} + \text{H} \end{aligned} \tag{2}$$

$$\begin{aligned} \text{CH}\_3\text{OH} &\leftrightarrow \text{CH}\_3\text{'}+\text{OH}^\cdot, \\ \text{OH}^\cdot+\text{CH}\_3\text{OH} &\leftrightarrow \text{H}\_2\text{O}+\text{CH}\_2\text{OH}^\cdot, \\ \text{CH}\_3\text{'}+\text{CH}\_3\text{OH} &\leftrightarrow \text{CH}\_4+\text{CH}\_2\text{OH}^\cdot, \\ \text{CH}\_2\text{OH}^\cdot+\text{M} &\leftrightarrow \text{CH}\_2\text{O}+\text{H}^\cdot+\text{M} \ , \\ \text{CH}\_2\text{O}+\text{M} &\leftrightarrow \text{H}^\cdot+\text{CHO}^\cdot+\text{M} \ , \\ \text{CHO}^\cdot+\text{M} &\leftrightarrow \text{CO}+\text{H}^\cdot+\text{M} \ , \\ \text{H}^\cdot+\text{CH}\_3\text{OH} &\leftrightarrow \text{H}\_2+\text{CH}\_2\text{OH}^\cdot \end{aligned} \tag{3}$$

**c.** the sonolysis of water yields the H• and OH• radicals

$$(\text{H}\_2\text{O})\rangle \text{H}^\* + \text{OH}^\* \tag{4}$$

**d.** the ultrasonic irradiation facilitates the reduction of chalcogens to the active forms of S-2 and Se-2 (see Refs. in [24,43]) that react with the in-situ generated H● radicals forming H2S and H2Se [44]

$$\text{H S} + 2\text{H}^\* \rightarrow \text{H}\_2\text{S} \tag{5}$$

$$\text{Fe} + 2\text{H}^\* \rightarrow \text{H}\_2\text{Se} \tag{6}$$


$$\rm{SbbI}\_3 + \rm{H\_2S} \rightarrow \rm{SbSI} + \rm{H\_2} + \rm{I\_2} \tag{7}$$

$$\rm SbI\_3 + H\_2Sc \to SbSeI + H\_2 + I\_2 \tag{8}$$

**h.** SbSI or SbSeI molecules, under the microjets and shockwaves formed during collapsing of the bubbles, are pushed towards each other in CNTs and are held by chemical forces. Therefore, the nuclei of SbSI or SbSeI are formed as a result of the interparticle collisions (see [24]).

25 24 2 25 24 2 25 25

3 3

CH OH CH OH ,

« + + «+ + «+

2 2

+ «+

• • • • 3 22

H O)))H + OH <sup>2</sup>

S 2H H S2

Se 2H H Se <sup>2</sup>

the solution is stable in the internal cavity of CNTs [48,49].

**g.** H2S or H2Se reacts with SbI3 to yield SbSI and SbSeI molecules

**e.** opening of the CNT ends due to the acids (H2S, H2Se) [7] or the action of ultrasounds

**f.** CNT's suck the released H2S or H2Se molecules and SbI3 in ethanol needed to build SbSI and SbSeI. This is allowed by the capillary effect [47]. It is well known [48,49], that open CNTs are impregnated with excess of the precursor solutions under ultrasonic condi‐ tions to ensure that most of the tubes can be filled by the impregnating solution. From the thermodynamic point of view, the dissolution of the impregnating solution sticking to external walls of CNTs into the washing medium necessitates the condition of the solvation energy gain smaller than zero [48,49]. On the other hand, the energy gain for the capillary filling of a CNT must be smaller than the solvation energy to ensure that

**d.** the ultrasonic irradiation facilitates the reduction of chalcogens to the active forms of S-2 and Se-2 (see Refs. in [24,43]) that react with the in-situ generated H● radicals forming

· · (4)

· + ® (5)

· + ® (6)

2

**c.** the sonolysis of water yields the H• and OH• radicals

H2S and H2Se [44]

276 Nanowires - Recent Advances

[45,46].

C H OH C H O H , C H OH C H H O , C H OH C H O H

« + « + « +

• •

• •

• • 3 22 • • 3 3 42 • •

+« ++ +« + + +« ++

 OH CH OH H O CH OH , CH CH OH CH CH OH , CH OH M CH O H M , CH O M H CHO M , CHO M CO H M , H CH OH H CH OH

(2)

(3)


The (7) and (8) were the basic reactions used in different methods of preparation of bulk SbSI-type crystals [52]. Despite the tubes filling, the observed nanowire type morphology of the product (Figures 11 and 12) is possibly due to the inherent chain type structure and growth habit of the SbSI as well as SbSeI crystals [50].

#### **5. Optical properties of SbSI@CNTs and SbSeI@CNTs**

Optical absorption spectra have been extensively used as one of the most important tools for probing the energy gaps (Eg) and band structures of semiconductors [53]. There are several methods for measuring them. These include diffuse reflectance spectroscopy (DRS). When a material, consisting of many particles or nanoparticles, is illuminated some of the impinging radiation penetrates the sample and some is reflected from its surface. The portion that pen‐ etrates the sample is scattered at a large number of points in its path as well as it is transmit‐ ted through the particles a number of times. Only the part of this radiation that is returned to the surface of the sample and comes back out is considered to be diffuse reflection. This component has been investigated extensively by many authors [54-56]. DRS is a suitable, not destructive, and simple method of investigation, especially important in examinations of po‐ rous, nanocrystaline materials and gels [57]. It is impossible to examine such materials ap‐ plying specular reflection and it is also extremely difficult to determine the pathlength in an optical transmittance of them. The following factors are related to high spectral quality of diffuse reflectance:


The optical diffuse reflection spectroscopy (DRS) of SbSI@CNTs and SbSeI@CNTs was car‐ ried out [11,14] on a spectrophotometer SP-2000 (Ocean Optics Inc.) equipped with an inte‐ grating sphere ISP-REF (Ocean Optics Inc.). Spectra were recorded at room temperature, from 350 to 1000 nm. The standard WS-1 (Ocean Optics Inc.) was used as a reference. The diffuse reflectance values were converted to the Kubelka–Munk function (known to be pro‐ portional to the absorption coefficient) shown by [54-56]

$$F\_{K-M}(R\_d) = \frac{(1 - R\_d)^2}{2R\_d} \sim \alpha \tag{9}$$

where Rd describes the coefficient of diffuse reflectance and α is the absorption coefficient of light in the investigated material.

**Figure 15.** Comparison of the diffuse reflectance spectra of the multi–walled CNTs filled with SbSI (a) [11] and SbSeI (b) [14] (■) ultrasonically (in methanol and ethanol, respectively), and of the empty multi–walled CNTs (□) in metha‐ nol and ethanol.

In Figure 15 the diffuse reflectance spectra of CNTs filled with SbSI and SbSeI are compared with the spectrum registered for hollow CNTs. In the first cases one can see the characteris‐ tic for semiconducting materials edges of fundamental absorption around 615 nm and 687 nm. However, the diffuse reflectance decreases also with increasing wavelengths (Figure 15), probable due to the large amount of free carriers absorbing light. Figure 16 presents the spectra of Kubelka–Munk functions derived from the diffuse reflectance data presented in Figure 15.

**1.** dilution of the sample with a non-absorbing matrix ensures a deeper penetration of the incident beam into the sample which increases the contribution of the scattered compo‐

**2.** smaller particles improve the quality of DRS spectra because the contribution of reflec‐

The optical diffuse reflection spectroscopy (DRS) of SbSI@CNTs and SbSeI@CNTs was car‐ ried out [11,14] on a spectrophotometer SP-2000 (Ocean Optics Inc.) equipped with an inte‐ grating sphere ISP-REF (Ocean Optics Inc.). Spectra were recorded at room temperature, from 350 to 1000 nm. The standard WS-1 (Ocean Optics Inc.) was used as a reference. The diffuse reflectance values were converted to the Kubelka–Munk function (known to be pro‐

> <sup>2</sup> (1 ) () ~ <sup>2</sup> *d*

*d*

where Rd describes the coefficient of diffuse reflectance and α is the absorption coefficient of

**Figure 15.** Comparison of the diffuse reflectance spectra of the multi–walled CNTs filled with SbSI (a) [11] and SbSeI (b) [14] (■) ultrasonically (in methanol and ethanol, respectively), and of the empty multi–walled CNTs (□) in metha‐

In Figure 15 the diffuse reflectance spectra of CNTs filled with SbSI and SbSeI are compared with the spectrum registered for hollow CNTs. In the first cases one can see the characteris‐ tic for semiconducting materials edges of fundamental absorption around 615 nm and 687 nm. However, the diffuse reflectance decreases also with increasing wavelengths (Figure 15), probable due to the large amount of free carriers absorbing light. Figure 16 presents the spectra of Kubelka–Munk functions derived from the diffuse reflectance data presented in

a


*KM d*

*<sup>R</sup> F R R* -

nent in the spectrum and minimizes the specular reflection component;

tion from the surface is reduced.

278 Nanowires - Recent Advances

light in the investigated material.

nol and ethanol.

Figure 15.

portional to the absorption coefficient) shown by [54-56]

**Figure 16.** Comparison of fitted spectra of Kubelka–Munk functions calculated for diffuse reflectances of multi–wal‐ led CNTs filled with SbSeI (■) with the data reported for multi–walled CNTs filled with SbSI (□). Solid curves represent the least square fitted theoretical dependences (description in the text; values of the fitted parameters are given in Table 4) [14].


**Table 4.** Comparison of parameters of multi–walled CNTs filled with SbSI and SbSeI with the literature data reported for SbSeI and SbSI nanowires produced sonochemically (EgIa – indirect allowed energy gap; EgIf – indirect forbidden energy gap; Eph – phonon energy; EU – Urbach energy; A0 – constant absorption; A50 A60, A120 – proportionality factors determined from the fitting of the spectra of FK–M functions evaluated from the DRS data).

The Kubelka–Munk (FK–M) spectra were least square fitted with theoretical dependences ap‐ propriate for different mechanisms of absorption [57]. The best fitting was obtained for the sum of indirect allowed absorption with absorption/emission of phonons (α1), free carrier absorption (α2) and constant absorption term (α3) (see the literature in [53,57]):

$$
\alpha = \alpha\_1 + \alpha\_2 + \alpha\_3 \tag{10}
$$

where

$$\alpha\_{1} = A\_{\rm s0} \frac{\left(h\nu - E\_{gla} + E\_{ph}\right)^{2}}{\exp\left(\frac{E\_{ph}}{k\_{B}T}\right) - 1} \qquad \text{for}$$

$$E\_{gla} - E\_{ph} < h\nu \le E\_{gla} + E\_{ph}$$

$$\alpha\_{\rm i} = A\_{\rm s0} \left| \frac{\left(h\nu - E\_{gla} + E\_{ph}\right)^2}{\exp\left(\frac{E\_{ph}}{k\_BT}\right) - 1} + \frac{\left(h\nu - E\_{gla} - E\_{ph}\right)^2}{1 - \exp\left(-\frac{E\_{ph}}{k\_BT}\right)} \right| \tag{12}$$

$$\text{for} \qquad h\nu > E\_{gla} + E\_{ph}$$

$$
\alpha\_2 = A\_{123} \cdot \mathcal{X}^2 \tag{13}
$$

$$
\alpha\_3 = A\_0 \tag{14}
$$

EgIa represents the indirect allowed energy gap, Eph is the phonon energy, A0, A50, A125 are constant parameters, hν is the photon energy, T is the temperature, kB is the Boltzmann con‐ stant. The determined values of these parameters are given in Table 4. In the same table they are compared with the literature data reported for SbSeI [19] and SbSI [57] nanowires pro‐ duced sonochemically.

It should be noted that absorption spectra [58-62] of bulk SbSeI single crystals have been fit‐ ted as indirect allowed absorption without phonon statistics – normalized for hν (α4)

$$\alpha\_4 = \frac{A\_{41}}{h\nu} \left(h\nu - E\_{g\text{la}}\right)^2 \qquad \qquad \text{for } h\text{v} > E\_{g\text{la}} \tag{15}$$

where EgIa represents the indirect allowed energy gap, A41 is the proportionality factor. It should be noted that the indirect allowed optical energy gap described by formulae (17) was discovered also in doped single crystals SbSeI:Co [61], SbSeI:V [60], SbSeI:Cr [59], and SbSeI:Ni [58]. Therefore, we fitted the same spectrum of FK–M of the CNTs filled with SbSeI (Figure 16) with theoretical dependence appropriate for the sum of indirect allowed absorp‐ tion without phonon statistics – normalized for hν (α4), free carrier absorption (α2) and con‐ stant absorption term (α3). Values of the fitted parameters are given in Table 4, too. However, the value of sum of squared residuals (χ<sup>2</sup> ) in the last fitting is over six hundred times larger then in the case of fitting with formula (10) (Table 4).

sum of indirect allowed absorption with absorption/emission of phonons (α1), free carrier

2

=++ <sup>123</sup> (10)

(11)

(12)

 a a

absorption (α2) and constant absorption term (α3) (see the literature in [53,57]):

aa

( )

*hv E E*


*gIa ph ph B*

*<sup>A</sup> for <sup>E</sup>*

( ) ( ) 2 2

î þ èø è ø

ì ü ï ï -+ --

*hv E E hv E E*

*gIa ph gIa ph ph ph B B*

*E E k T k T for hv E E*

2

 l

EgIa represents the indirect allowed energy gap, Eph is the phonon energy, A0, A50, A125 are constant parameters, hν is the photon energy, T is the temperature, kB is the Boltzmann con‐ stant. The determined values of these parameters are given in Table 4. In the same table they are compared with the literature data reported for SbSeI [19] and SbSI [57] nanowires pro‐

It should be noted that absorption spectra [58-62] of bulk SbSeI single crystals have been fit‐

where EgIa represents the indirect allowed energy gap, A41 is the proportionality factor. It should be noted that the indirect allowed optical energy gap described by formulae (17) was discovered also in doped single crystals SbSeI:Co [61], SbSeI:V [60], SbSeI:Cr [59], and

ted as indirect allowed absorption without phonon statistics – normalized for hν (α4)

<sup>4</sup> ( ) *gIa gIa <sup>A</sup> hv E for hv E hv*

41 2

*gIa ph*

a= - > (15)

2 123 = × *A* (13)

3 0 = *A* (14)

> +

exp 1 1 exp

a = í ý <sup>+</sup> æö æ ö ï ï ç÷ ç ÷ - --

exp 1

ç ÷ è ø

*k T*

*gIa ph gIa ph*


a

a

*E E hv E E*

a = æ ö

1 50

1 50

duced sonochemically.

*A*

where

280 Nanowires - Recent Advances

One should remember that the FK–M spectrum of sonochemically produced alone SbSeI nanowires [19] was best fitted for indirect forbidden absorption (α5), Urbach ruled absorp‐ tion (α6) and constant absorption term (α3), where

$$\alpha\_{\rm s} = A\_{60} \left( h\mathbf{v} - E\_{\rm gf} \right)^{\rm s} \qquad \text{ for } h\mathbf{v} > E\_{\rm gf} \tag{16}$$

$$\alpha\_{\delta} = A\_{U} \exp\left[\frac{h\nu}{E\_{U}}\right] \tag{17}$$

Therefore, we fitted the same spectrum (Figure 16) of FK–M of the CNTs filled with SbSeI with theoretical dependence appropriate for the sum of α5, α2 and α<sup>3</sup> (the fitting did not take into account the α6 because it is in contradiction with the experimental data described by α2). One can see that the value of sum of squared residuals (χ<sup>2</sup> ) in the last fitting is over forty five times larger than in the case of fitting with the indirect allowed absorption (Table 4).

The observed free carrier absorption of light in the case of CNTs filled with SbSI and SbSeI is evoked by the CNTs material because it is absent in the case of alone SbSI and SbSeI nano‐ wires [19,57]. Instead of it the Urbach absorption was reported in the latter cases. May be the latter mechanism of absorption exists also in CNTs filled with SbSI and SbSeI but it is hid‐ den by the free carrier absorption. It is known (see [47]) that a metal atom intercalated inside the internal cavity of a CNT displays a tendency towards the transfer of some part of the valence electrons to the outer surface of the nanotube, where unoccupied electronic states exist. As a result of such a transfer there arises an additional mechanism of electrical con‐ duction, related to the travel of an electron about those states.

The nanocrystalline SbSeI filling the CNTs is a semiconductor with little smaller energy gap of 1.61 eV than the alone SbSeI nanowires (Table 4). The established type of indirect allowed energy gap of SbSeI in CNTs is different from the indirect forbidden energy gap of sono‐ chemically produced alone SbSeI nanowires [19]. However, one should remember that ab‐ sorption spectra [58-62] of bulk SbSeI single crystals have been fitted with indirect allowed absorption without phonon statistics – normalized for hν. Therefore, the indirect allowed energy gap of SbSeI in CNTs is very probable and the light absorption in the alone SbSeI nanowires should be studied again.

## **6. Electrical and photoelectrical properties of SbSI@CNTs and SbSeI@CNTs**

In [13] for the first time electrical investigations have been performed on circular samples of SbSI@CNT gel prepared sonochemically. The diameter and thickness of the sample were 4 mm and 1.5 mm, respectively. The largest opposite surfaces of the sample were covered with a silver paste (SPI Supplies) and electrical connection from these electrodes was made by copper wire. The measurements were performed applying impedance analyzer Agilent 4294A in the frequency (f) range 200 Hz – 20 MHz with amplitude of 0.01 V and zero bias. The impedance investigations were made in vacuum in the temperature region from 273 K to 353 K using a LN2 cryostat VPF-700 (Janis Research Company, Inc.). The metallic cham‐ ber served as a Faraday cage in order to eliminate undesired external interference. The tem‐ perature was controlled using 331 Temperature Monitor, (Lake Shore). Measurements were performed in darkness to avoid the excess carriers that can be photogenerated in SbSI@CNT. LabView program has been developed for computer-controlling the experiment, data ac‐ quisition and analysis.

Figures 17a and 17b present influence of temperature on Bode plots for SbSI@CNT gel pre‐ pared ultrasonically in methanol. The impedance amplitude (|Z|) decreases while the phase (̈) increases with increasing frequency. Using formulae: Re*Z* = |*Z* |cos*ϕ*and Im*Z* = |*Z* |sin*ϕ*, the Nyquist plots (Figure 17c) have been constructed. The two arcs in Ny‐ quist plots, in the temperature range from 273 K to 293 K, suggest the need of using non De‐ bye model to interpret the data. Hence, the equivalent circuit to interpret the impedance of SbSI@CNT gel should contain the constant phase element (CPE) with impedance given by

$$Z\_{CPE} = \left[ A \left( j\rho o \right)^{n} \right]^{-1} \tag{18}$$

where: parameter A is a constant, j2 = -1, ω = 2πf, the exponent n equals 1 for ideal capacitor, 0.5 in the case of diffusion processes, 0 for ideal resistor, and -1 for ideal inductor [63]. The porous materials have values of n from the range 0.9 ÷ 1 [63].

For temperatures higher than T=293 K only one arc in Nyquist plots is observed (Figure 17c). Probably, it is evoked by the ferroelectric-paraelectric phase transition of SbSI near Curie temperature TC=292(1) K [17,18].

The experimental data have been best fitted (Figure 17) employing complex nonlinear curve fitting using ZView 2 program [64]. Fitting of the Nyquist plots has been done using equiva‐ lent circuits reported for different models (e.g. Voigt circuit, Maxwell circuit, and their com‐ binations). The best fitting was obtained using equivalent circuit shown in Figure 18. It is consistent of serial connection of two parallel circuits (resistors: R1, R2 and CPE elements: CPE1, CPE2) with inductance L. Probably, the first parallel circuit (R1 – CPE1) represents properties of SbSI component while the second parallel circuit (R2 – CPE2) describes proper‐ ties of CNTs and the electrodes. The inductance L is dedicated to the presence of CNTs [65,66].

**Figure 17.** Influence of temperature on Bode (a, b) and Nyquist (c) plots for SbSI@CNT gel prepared ultrasonically in methanol [13] (■ – 273 K, – 283 K, – 293 K, – 303 K, – 313 K, – 323 K, – 333 K, – 343 K, – 353 K). Solid curves represent the fitted theoretical dependences calculated for an equivalent circuit presented in Figure 18. The fitted parameters are given in Figures 19-21.

**Figure 18.** Equivalent circuit for SbSI@CNT gel prepared ultrasonically in methanol [13].

**6. Electrical and photoelectrical properties of SbSI@CNTs and**

In [13] for the first time electrical investigations have been performed on circular samples of SbSI@CNT gel prepared sonochemically. The diameter and thickness of the sample were 4 mm and 1.5 mm, respectively. The largest opposite surfaces of the sample were covered with a silver paste (SPI Supplies) and electrical connection from these electrodes was made by copper wire. The measurements were performed applying impedance analyzer Agilent 4294A in the frequency (f) range 200 Hz – 20 MHz with amplitude of 0.01 V and zero bias. The impedance investigations were made in vacuum in the temperature region from 273 K to 353 K using a LN2 cryostat VPF-700 (Janis Research Company, Inc.). The metallic cham‐ ber served as a Faraday cage in order to eliminate undesired external interference. The tem‐ perature was controlled using 331 Temperature Monitor, (Lake Shore). Measurements were performed in darkness to avoid the excess carriers that can be photogenerated in SbSI@CNT. LabView program has been developed for computer-controlling the experiment, data ac‐

Figures 17a and 17b present influence of temperature on Bode plots for SbSI@CNT gel pre‐ pared ultrasonically in methanol. The impedance amplitude (|Z|) decreases while the phase (̈) increases with increasing frequency. Using formulae: Re*Z* = |*Z* |cos*ϕ*and Im*Z* = |*Z* |sin*ϕ*, the Nyquist plots (Figure 17c) have been constructed. The two arcs in Ny‐ quist plots, in the temperature range from 273 K to 293 K, suggest the need of using non De‐ bye model to interpret the data. Hence, the equivalent circuit to interpret the impedance of SbSI@CNT gel should contain the constant phase element (CPE) with impedance given by

( ) <sup>1</sup> *<sup>n</sup>*

ë û (18)

w - = é ù

where: parameter A is a constant, j2 = -1, ω = 2πf, the exponent n equals 1 for ideal capacitor, 0.5 in the case of diffusion processes, 0 for ideal resistor, and -1 for ideal inductor [63]. The

For temperatures higher than T=293 K only one arc in Nyquist plots is observed (Figure 17c). Probably, it is evoked by the ferroelectric-paraelectric phase transition of SbSI near Curie

The experimental data have been best fitted (Figure 17) employing complex nonlinear curve fitting using ZView 2 program [64]. Fitting of the Nyquist plots has been done using equiva‐ lent circuits reported for different models (e.g. Voigt circuit, Maxwell circuit, and their com‐ binations). The best fitting was obtained using equivalent circuit shown in Figure 18. It is consistent of serial connection of two parallel circuits (resistors: R1, R2 and CPE elements: CPE1, CPE2) with inductance L. Probably, the first parallel circuit (R1 – CPE1) represents properties of SbSI component while the second parallel circuit (R2 – CPE2) describes proper‐ ties of CNTs and the electrodes. The inductance L is dedicated to the presence of CNTs [65,66].

*Z Aj CPE*

porous materials have values of n from the range 0.9 ÷ 1 [63].

temperature TC=292(1) K [17,18].

**SbSeI@CNTs**

282 Nanowires - Recent Advances

quisition and analysis.

Every semiconducting ferroelectric SbSI nanowire (in CNT) can be represented by parallel connection of resistance and capacitance. There appear a few groups of possible explanation of physical origin of CPEs:


In the present case, we are able to offer (after [70]) some additional suggestions, as follows. The CPE represents time-dependent phenomena which, in the case of polarization fluctua‐ tions, may be an indicator of the cooperative nature of the fluctuations of individual dipoles. Specifically, fluctuating dipoles interact with their surroundings.

**Figure 19.** Comparison of the temperature dependences of resistance parameters of the equivalent circuits describing Nyquist plots registered in the cases of SbSI@CNT ultrasonically prepared in methanol [13] ( – R1, ◆ – R2 ) (Figures 17 and 18), and SbSI@CNT ( – R1, ■ – R2) and SbSeI@CNT ( – R1, □ – R2) ultrasonically prepared in ethanol.

**Figure 20.** Comparison of the temperature dependences of capacitance parameters of the equivalent circuits describ‐ ing Nyquist plots registered in the cases of SbSI@CNT ultrasonically prepared in methanol [13] (◆– n1, A1; – n2, A2) (Figures 17 and 18), and SbSI@CNT (▪ – n, A; – C) and SbSeI@CNT (■ – C1, – C2) ultrasonically prepared in ethanol.

In order to interpret experimental data, it is essential to have a model equivalent circuit that provides a realistic representation of the sample. The investigated SbSI@CNT and SbSeI@CNT gel is a complicated object that contains carbon nanotubes (filled with SbSI or SbSeI) connected with each other in parallel and in series, as well as interfaces (i.e. nano‐ wires - nanowires or nanowires - electrodes). Of course the models should not possess too many components, because the fitting of many variables will be unreliable. Therefore the in‐ terpretation of the components of the equivalent electric circuits presented in Figure 18 is very difficult. To obtain simpler image of the observed phenomena experiments with single SbSI@CNT and SbSeI@CNT are needed.

tions, may be an indicator of the cooperative nature of the fluctuations of individual dipoles.

**Figure 19.** Comparison of the temperature dependences of resistance parameters of the equivalent circuits describing Nyquist plots registered in the cases of SbSI@CNT ultrasonically prepared in methanol [13] ( – R1, ◆ – R2 ) (Figures 17

**Figure 20.** Comparison of the temperature dependences of capacitance parameters of the equivalent circuits describ‐ ing Nyquist plots registered in the cases of SbSI@CNT ultrasonically prepared in methanol [13] (◆– n1, A1; – n2, A2) (Figures 17 and 18), and SbSI@CNT (▪ – n, A; – C) and SbSeI@CNT (■ – C1, – C2) ultrasonically prepared in ethanol.

and 18), and SbSI@CNT ( – R1, ■ – R2) and SbSeI@CNT ( – R1, □ – R2) ultrasonically prepared in ethanol.

Specifically, fluctuating dipoles interact with their surroundings.

284 Nanowires - Recent Advances

**Figure 21.** Comparison of the temperature dependences of inductance parameters of the equivalent circuits describ‐ ing Nyquist plots registered in the cases of SbSI@CNT ultrasonically prepared in methanol [13] (◆) (Figures 17 and 18), and SbSI@CNT ( ) and SbSeI@CNT ( ) ultrasonically prepared in ethanol.

Figure 22 presents comparison of the temperature dependences of relaxation time of SbSI@CNT ultrasonically prepared in methanol with the relaxation time of SbSI@CNT and SbSeI@CNT ultrasonically prepared in ethanol. One can see the influence of the liquid used during sonochemical filling of CNTs on properties of the product. In the case of SbSI@CNT ultrasonically prepared in methanol there are two relaxation times in temperatures lower than T=293 K. Above this temperature only one relaxation time is needed to describe the properties of SbSI@CNT. Such behavior can be connected with the ferroelectric-paraelectric phase transition of SbSI near Curie temperature TC=292(1) K [17,18]. This phenomenon needs additional investigations.

**Figure 22.** Comparison of the temperature dependences of relaxation time of SbSI@CNT ultrasonically prepared in methanol (◆ – τ1, – τ2 ) (Figure 17), and SbSI@CNT ( ) and SbSeI@CNT ( ) ultrasonically prepared in ethanol [13].

It should be noted, that similar equivalent circuits have been well fitted to the Nyquist plots registered also in the cases of SbSI@CNT and SbSeI@CNT ultrasonically prepared in etha‐ nol. Comparison of the temperature dependences of the fitted parameters are given in Fig‐ ures 19-21.

To overcome the mentioned above problems with interfaces (i.e., nanotubes - nanotubes and nanotubes - electrodes), an essential prerequisite is to build reliable interconnections be‐ tween the CNTs and the external electrical circuits. To address this need, various chemical and physical processes have been explored to build such interconnections [71]. Changxin Chen et al. [72,73] have presented ultrasonic nanowelding process, with which one can fab‐ ricate reliable bonding between SWCNTs and metal electrodes. Ultrasonic nanowelding was carried out in a FB-128 ultrasonic wire bonder [72,73]. An Al2O3 single crystal with a 2,500 μm2 circular pressing surface and a RMS roughness of 0.2 nm was mounted onto the bonder to act as the welding head. A clamping force of 78.4 mN was applied to press the welding head against the nanotube and electrodes. At the same time an ultrasonic vibration with a frequency of 60 kHz and an ultrasonic power of about 10 mW was applied to the welding head through an ultrasonic transducer. The ultrasonic energy was transferred to the bond‐ ing interface through the ultrasonic welding head. Thus the ends of SWCNTs and electrodes were welded together under the combined action of the ultrasonic energy and a clamping force. The welding process was carried out at room temperature for a duration of 0.2 s.

Figure 23 presents SEM and AFM images of carbon nanotubes on Si/SiO2 substrate with Au electrodes after ultrasonic welding. In our ultrasonic nanowelding set-up the generator ADG70-100P-230-NO 70 kHz (Rinco Ultrasonics) with converter C 70-2 (Rinco Ultrasonics) has been used. The sonotrodes have been equipped with SiC single crystals. The preliminary electrical tests of SbSI@CNTs ultrasonic nanowelded with different microelectrodes on dif‐ ferent substrates are successful. The obtained results will be published in a near future.

**Figure 23.** SEM and AFM images of carbon nanotubes on Si/SiO2 substrate with Au electrodes after ultrasonic welding.

## **5. Conclusions**

**Figure 22.** Comparison of the temperature dependences of relaxation time of SbSI@CNT ultrasonically prepared in methanol (◆ – τ1, – τ2 ) (Figure 17), and SbSI@CNT ( ) and SbSeI@CNT ( ) ultrasonically prepared in ethanol [13].

It should be noted, that similar equivalent circuits have been well fitted to the Nyquist plots registered also in the cases of SbSI@CNT and SbSeI@CNT ultrasonically prepared in etha‐ nol. Comparison of the temperature dependences of the fitted parameters are given in Fig‐

To overcome the mentioned above problems with interfaces (i.e., nanotubes - nanotubes and nanotubes - electrodes), an essential prerequisite is to build reliable interconnections be‐ tween the CNTs and the external electrical circuits. To address this need, various chemical and physical processes have been explored to build such interconnections [71]. Changxin Chen et al. [72,73] have presented ultrasonic nanowelding process, with which one can fab‐ ricate reliable bonding between SWCNTs and metal electrodes. Ultrasonic nanowelding was carried out in a FB-128 ultrasonic wire bonder [72,73]. An Al2O3 single crystal with a 2,500 μm2 circular pressing surface and a RMS roughness of 0.2 nm was mounted onto the bonder to act as the welding head. A clamping force of 78.4 mN was applied to press the welding head against the nanotube and electrodes. At the same time an ultrasonic vibration with a frequency of 60 kHz and an ultrasonic power of about 10 mW was applied to the welding head through an ultrasonic transducer. The ultrasonic energy was transferred to the bond‐ ing interface through the ultrasonic welding head. Thus the ends of SWCNTs and electrodes were welded together under the combined action of the ultrasonic energy and a clamping force. The welding process was carried out at room temperature for a duration of 0.2 s.

Figure 23 presents SEM and AFM images of carbon nanotubes on Si/SiO2 substrate with Au electrodes after ultrasonic welding. In our ultrasonic nanowelding set-up the generator ADG70-100P-230-NO 70 kHz (Rinco Ultrasonics) with converter C 70-2 (Rinco Ultrasonics) has been used. The sonotrodes have been equipped with SiC single crystals. The preliminary

ures 19-21.

286 Nanowires - Recent Advances

The presented very simple, sonochemical synthesis of nanophase SbSI or SbSeI in CNTs at low temperature is a convenient, fast, mild, efficient and environmentally friendly route for producing novel type of hybrid nanomaterials. The resulting SbSI@CNTs and SbSeI@CNTs composites are highly anisotropic 1D structures whose electronic and optical properties are considerably modified with respect to the encapsulating nanotube. The fabricated SbSI and SbSeI fillings of CNTs are single crystalline in nature and in the form of nanowires.

Since the sonochemical process was carried out at ambient temperature and pressure, it may be predicted that upscaling of this method will lead to large quantities of SbSI@CNTs and SbSeI@CNTs. Further studies on the properties of the sonochemically prepared SbSI and SbSeI in CNTs are underway. Taking into account the recently reported sonochemical prep‐ aration of pure SbS1-xSexI nanowires [36], one should expect the sonochemical preparation of quaternary SbS1-xSexI semiconductors within CNTs. The possibility of change the energy gap (and other parameters) with changing molar composition (x) of the quaternary compounds [36] gives opportunity to tail the functional properties of CNTs filled with SbS1-xSexI. It should be remember that the physical properties of such quaternary compounds, formed as solid solutions, can be tailored with stoichiometric composition. It should be an advantage of using such material in different devices. It seems that this approach can be extended to the preparation of some other ternary and quaternary nanomaterials formed from the group 15-16-17 elements within CNTs. It should provide a "bottom–up" approach for the manufac‐ ture of future nanoscale devices.

This review summarizes all so far published results of investigations on SbSI@CNTs and SbSeI@CNTs. One can see that the properties of these 1D hybrid nanomaterials still remain little known. Meantime, these nanostructures should provide promising materials for funda‐ mental investigations on nanoscale ferroelectricity and piezoelectricity as well as materials for some applications.

Obviously, the presented new materials as the other one-dimensional semiconductor nano‐ structures should receive considerable attention from the scientific and engineering com‐ munities due to their potentially useful novel electronic properties.

The ultrasonic nanowelding of SbSI@CNTs and SbSeI@CNTs with metal microelectrodes meets the need of future large-scale, simply, convenient, fast, and efficient technology to build reliable interconnections between the CNTs filled with ternary chalcohalides formed from the group 15–16–17 elements and the external electrical circuits.

## **Acknowledgements**

The authors are indebted to Dr. P. Szperlich, Prof. J. Szala, Prof. S. Kochowski, Dr. A. Starc‐ zewska, Dr. M. Kępińska, and Dr. T. Rzychoń of the Silesian University of Technology, Ka‐ towice (Poland), and Prof. D. Stróż and Prof. R. Wrzalik of the University of Silesia, Katowice (Poland) for collaboration in investigations of carbon nanotubes filled with ternary chalcohalides. This review was partially supported by the NCN (Poland) under contract No. UMO-2011/01/B/ST5/06273.

## **Author details**

Marian Nowak\* and Marcin Jesionek

\*Address all correspondence to: Marian.Nowak@polsl.pl

Institute of Physics, Silesian University of Technology, Poland

### **References**


of using such material in different devices. It seems that this approach can be extended to the preparation of some other ternary and quaternary nanomaterials formed from the group 15-16-17 elements within CNTs. It should provide a "bottom–up" approach for the manufac‐

This review summarizes all so far published results of investigations on SbSI@CNTs and SbSeI@CNTs. One can see that the properties of these 1D hybrid nanomaterials still remain little known. Meantime, these nanostructures should provide promising materials for funda‐ mental investigations on nanoscale ferroelectricity and piezoelectricity as well as materials

Obviously, the presented new materials as the other one-dimensional semiconductor nano‐ structures should receive considerable attention from the scientific and engineering com‐

The ultrasonic nanowelding of SbSI@CNTs and SbSeI@CNTs with metal microelectrodes meets the need of future large-scale, simply, convenient, fast, and efficient technology to build reliable interconnections between the CNTs filled with ternary chalcohalides formed

The authors are indebted to Dr. P. Szperlich, Prof. J. Szala, Prof. S. Kochowski, Dr. A. Starc‐ zewska, Dr. M. Kępińska, and Dr. T. Rzychoń of the Silesian University of Technology, Ka‐ towice (Poland), and Prof. D. Stróż and Prof. R. Wrzalik of the University of Silesia, Katowice (Poland) for collaboration in investigations of carbon nanotubes filled with ternary chalcohalides. This review was partially supported by the NCN (Poland) under contract No.

[1] Iijima, S. (1991). Helical microtubules of graphitic carbon. *Nature*, 354, 56-58.

[2] Ajayan, P. M., & Iijima, S. (1993). Capillarity-induced filling of carbon nanotubes. *Na‐*

munities due to their potentially useful novel electronic properties.

from the group 15–16–17 elements and the external electrical circuits.

ture of future nanoscale devices.

for some applications.

288 Nanowires - Recent Advances

**Acknowledgements**

UMO-2011/01/B/ST5/06273.

and Marcin Jesionek

\*Address all correspondence to: Marian.Nowak@polsl.pl

Institute of Physics, Silesian University of Technology, Poland

**Author details**

Marian Nowak\*

**References**

*ture*, 361, 333-334.


[31] Pohl, S., & Saak, W. (1984). Zur Polymorphie von Antimontriiodid. Die Kristallstruk‐ tur von monoklinem SbI3 . *Z. Kristallogr*, 169, 177-184.

[16] Nowak, M., Mroczek, P., Duka, P., Kidawa, A., Szperlich, P., Grabowski, A., Szala, J., & Moskal, G. (2009). Using of textured policrystalline SbSI in actuators. *Sensors and*

[17] Nowak, M. (2010). Photoferroelectric nanowires. *in: N. Lupu (Ed.) Nanowires Science and Technology. Rijeka: InTech*, 269-308, http://www.intechopen.com/articles/show/

[18] Szperlich, P., Nowak, M., Bober, Ł., Szala, J., & Stróż, D. (2009). Ferroelectric proper‐ ties of ultrasonochemically prepared SbSI ethanogel. *Ultrason. Sonochem*, 16, 398-401.

[19] Nowak, M., Kauch, B., Szperlich, P., Jesionek, M., Kępińska, M., Bober, Ł., Szala, J., Moskal, G., Rzychoń, T., & Stróż, D. (2009). Sonochemical Preparation of SbSeI Gel.

[20] Suslick, K. S., & Price, G. J. (1999). Applications of ultrasound to materials chemistry.

[21] Jin, Ho. Bang, & Suslick, K. S. (2010). Applications of Ultrasound to the Synthesis of

[22] Abedini, R., & Mousavi, S. M. (2010). Preparation and enhancing of materials using ultrasound technique:polymers, catalysts and nano-structure particles. *Petroleum &*

[23] Mason, T. J. (2003). Sonochemistry and sonoprocessing: the link, the trends and

[24] Gedanken, A. (2004). Using sonochemistry for the fabrication of nanomaterials,. *Ul‐*

[25] Cravotto, G., & Cintas, P. (2006). Power ultrasound in organic synthesis: moving cav‐ itational chemistry from academia to innovative and large-scale applications. *Chem.*

[26] Suslick, K. S., Didenko, Y., Fang, M. M., Hyeon, T., Kolbeck, K. J., Mc Namara, W. B., Mdleleni, M. M., & Wong, M. (1999). Acoustic cavitation and its chemical consequen‐

[27] Ibanez, A., Jumas, J. C., Olivier-Fourcade, J., Philippot, E., & Maurin, M. (1983). On chalcogenide-iodates of antimony SbXI (X=S, Se, Te): structure and Mossbauer spec‐

[28] Antimony Selenide Iodide (2000). JCPDS-International Centre for Diffraction Data.

[29] Hassel, O., & Mark, H. (1924). Über die Kristallstruktur des Graphits. *Z. Phys*, 25,

[30] Carbon (2000). JCPDS-International Centre for Diffraction Data. PCPDFWIN Card

title/photoferroelectric-nanowires, accessed 21 June 2012).

Nanostructured Materials. *Adv. Mater*, 22, 1039-1059.

(probably) the future,. *Ultrason. Sonochem*, 10, 175-179.

*Actuators A*, 150, 251-256.

290 Nanowires - Recent Advances

*Ultrason. Sonochem*, 16, 546-551.

*Annu. Rev. Mater. Sci*, 29, 295-326.

*Coal*, 52, 81-98.

*trason. Sonochem*, 11, 47-55.

ces. *Phil. Trans. Roy. Soc. A*, 357, 335-353.

tra of 121Sb. *J. Solid State Chem*, 48, 272-283.

PCPDFWIN Card File , 2(76-1354)

*Soc. Rev*, 35, 180-196.

317-337.

File , 2(75-1621)


[61] Soonie, Jeon., Gijun, Cho., Wha-tek, Kim., & Sook-Il, Kwun. (1988). *Solid State Com‐ mun*, 68, 1043-1046.

[46] Chunwei, Yang., Xinguo, Hu., Dianlong, Wang., Changsong, Dai., Liang, Zhang., Haibo, Jin., & Simeon, Agathopoulos. (2006). Ultrasonically treated multi-walled car‐ bon nanotubes (MWCNTs) as PtRu catalyst supports for methanol electrooxidation.

[47] Eletskii, A. V. (2004). Sorption properties of carbon nanostructures. *Physics-Uspekhi*,

[48] Qiang, F. U., Weinberg, G., & Dang-sheng, S. U. (2008). Selective filling of carbon nanotubes with metals by selective washing. *New Carbon Materials*, 23, 17-20.

[49] De Gennes, P. G., Brochard-Wyart, F., & Quere, D. (2003). Capillarity and Wetting

[50] Voutsas, G. P., & Rentzeperis, P. J. (1986). The crystal structure of the quaternary

[51] Molnar, B., Johannes, R., & Haas, W. (1965). Properties of single-crystal SbSI. *Bull.*

[52] Gerzanich, E. I., Lyakhovitskaya, V. A., Fridkin, V. M., & Popovkin, B. A. (1982). *in: E. Kaldis (Ed.), Current Topics in Materials Science, Amsterdam: North-Holland*, 10,

[53] Pankove, J. I. (1971). Optical processes in semiconductors. *New Jersey: Prentice-Hall*

[55] Philips-Invernizzi, B., Dupont, D., & Caze, C. (2001). Bibliographical review for re‐

[56] Dahm, D. J., & Dahm, K. D. (2007). Interpreting Diffuse Reflectance and Transmit‐ tance: A Theoretical Introduction to Absorption Spectroscopy of Scattering Materials.

[57] Nowak, M., Kauch, B., & Szperlich, P. (2009). Determination of energy band gap of nanocrystalline SbSI using diffuse reflectance spectroscopy,. *Rev. Sci. Instrum*, 80,

[58] Dong-Woon, Shin., Seung-Cheol, Hyun., Sang-An, Park., Yoeng-Geun, Kim., Chang-Dae, Kim., & Wha-Tek, Kim. (1994). Optical properties of undoped and Ni-doped V-

[59] Seung-Cheol, Hyun., Young-Geun, Kim., Mi-Yang, Kim., Jeong-Dae, Koh., Byong-Seo, Park., & Wha-Tek, Kim. (1995). Optical properties of undoped and chromium-

[60] Sang-An, Park., Mi-Yang, Kim., Ji-Young, Lim., Byong-Seo, Park., Jeong-Dae, Koh., & Wha-Tek, Kim. (1995). Optical Properties of Undoped and V-Doped VA-VIA-VIIA

Phenomena: Drops, Bubbles, Pearls, Waves. *Berlin: Springer*, 51-52.

[54] Kortum, G. (1969). Reflectance Spectroscopy. *Berlin: Springer-Verlag*.

flectance of diffusing media. *Opt. Eng*, 40, 1082-1092.

VI-VII single crystals. *J. Phys. Chem. Solids*, 55, 825-830.

Single Crystals. *Phys. Status Solidi B*, 187, 253-260.

doped VA-VIA-VIIA single crystals. *J. Mater. Sci*, 30, 6113-6117.

compound SbSe0.75S0.25I. *Z. Kristallogr*, 178, 289-295.

*J. Power Sources*, 160, 187-193.

47, 1119-1154.

292 Nanowires - Recent Advances

*Am. Phys. Soc*, 10, 109.

*Norfolk: IM Publications LLP*.

55-190.

046107 .

*Inc*.


## **Synthesis and Characterization of Advanced Carbon-Based Nanowires – Study of Composites Actuation Capabilities Containing These Nanowires as Fillers**

V. Salles, L. Seveyrat, T. Fiorido, L. Hu, J. Galineau, C. Eid, B. Guiffard, A. Brioude and D. Guyomar

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54384

## **1. Introduction**

Electromechanical properties in polymers can be employed to create a large number of sen‐ sors and actuators [1-2]. For example they could replace the piezoelectric materials common‐ ly used in Micro Electromechanical Systems (MEMS). Even if the electromechanical coupling is relatively weak for polymers, they can generate high strains due to electrostric‐ tive and Maxwell effects which are a quadratic function of the applied electric field as op‐ posed to a linear function for piezoelectric materials. Other advantages of the polymers are their ease of processability, flexibility and cheapness.

Actuation capabilities of polymers depend on their intrinsic electrical, dielectric and me‐ chanical properties. In order to optimize these properties, one way consists on the incorpo‐ ration of nano-objects in the polymer matrix. Due to the nanometer-scale of the fillers dimensions, an important change of the properties can be obtained with relatively low ra‐ tios. In addition, the effect is more pronounced for conductive nano-objets since higher per‐ mittivities are expected to be obtained with a reduced amount of fillers [3]. Carbon-based nanocomposites are good candidates for electroactive nanocomposites: some studies were performed particularly on carbon black, carbon nanotubes, graphite or graphene fillers [4-9]. This part of the chapter is devoted to the influence of carbon-based nanowires (SiC@C, SiC@SiO2, Fe3C@C) as fillers on polyurethanes. The microstructural, dielectric, mechanical, electromechanical and electromagnetic properties are presented and discussed in the follow‐ ing sections.

© 2012 Salles et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Salles et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Among the available electroactive polymers, polyurethanes are of great interest for a wide range of actuator applications due to their significant electric field-induced strains [10-11].

The total electrical field induced strain S can be expressed from electrostricition and Max‐ well stress [12-13].

$$\mathbf{S} = \mathbf{S}\_{electrostriation} + \mathbf{S}\_{Maxwell} = \mathbf{M}\_{33}^{"\ast} \boldsymbol{E}^2 \tag{1}$$

With M\*33 is the apparent electromechanical coefficient and E the applied electrical field.

The Mawwell stress effect comes from the interaction force between the positive and nega‐ tive charges present on the electrodes

$$\mathcal{S}\_{\text{Maxwell}} = -\frac{\stackrel{\cdot}{\mathcal{E}\_r}\mathcal{E}\_0}{Y} E^2 \tag{2}$$

With ε'<sup>r</sup> the real part of the polymer relative permittivity, ε<sup>0</sup> the vacuum permittivity, and Y the Young's modulus.

The electrostriction represents the coupling between the electrical polarization and mechani‐ cal response in the material and is related to the electrostrictive coefficient Q and the relative permittiy ε'r.

$$\mathcal{S}\_{electrosfriction} = \mathsf{Q} \, \varepsilon\_0^2 \left(\boldsymbol{\varepsilon}\_r^\prime - 1\right)^2 \, \boldsymbol{E}^2 \tag{3}$$

It was shown in previous studies [4, 8, 14] that in such polymers as PU, the Maxwell stress can be neglected.

Based on experimental studies [12, 14-15], the Q coefficient is found to be inversely propor‐ tional to the product of the Young's Modulus and permittivity according to:

$$Q \propto \frac{1}{\varepsilon\_0 \varepsilon\_r^{\prime} Y} \tag{4}$$

As a consequence, an empirical proportionality relationship is also obtained for the M coefficient :

$$M \propto \frac{\varepsilon\_0 (\varepsilon\_r' - 1)^2}{\varepsilon\_r' \, Y} \tag{5}$$

This chapter is divided into two main parts. The fabrication procedures and the characteri‐ zation of the as-prepared nanowires are detailed in the first part whereas the second part is dedicated to show how these nanomaterials can act on the electromechanical properties of EAPs when they are introduced as fillers, demonstrating their high potentiality to fabricate innovative actuators.

## **2. Fabrication and characterization of carbon-based nanowires**

Among the available electroactive polymers, polyurethanes are of great interest for a wide range of actuator applications due to their significant electric field-induced strains [10-11].

The total electrical field induced strain S can be expressed from electrostricition and Max‐

With M\*33 is the apparent electromechanical coefficient and E the applied electrical field.

The Mawwell stress effect comes from the interaction force between the positive and nega‐

'

*Y* e e

With ε'<sup>r</sup> the real part of the polymer relative permittivity, ε<sup>0</sup> the vacuum permittivity, and Y

The electrostriction represents the coupling between the electrical polarization and mechani‐ cal response in the material and is related to the electrostrictive coefficient Q and the relative

It was shown in previous studies [4, 8, 14] that in such polymers as PU, the Maxwell stress

Based on experimental studies [12, 14-15], the Q coefficient is found to be inversely propor‐

0 1 '

*r*

As a consequence, an empirical proportionality relationship is also obtained for the M

*r r*

e

e e

*Y*

2 <sup>0</sup> ( ' 1) '

2 ' 22

*Maxwell S E*

<sup>0</sup> ( 1) *electrostriction <sup>r</sup> S QE* = e e

tional to the product of the Young's Modulus and permittivity according to:

*Q* e e

*M*

*r* 0 2

\* 2 *electrostriction Maxwell* <sup>33</sup> *SS S ME* = += (1)

= - (2)

(3)

*<sup>Y</sup>* <sup>µ</sup> (4)


well stress [12-13].

296 Nanowires - Recent Advances

the Young's modulus.

permittiy ε'r.

can be neglected.

coefficient :

tive charges present on the electrodes

Among the different fabrication processes able to lead to one-dimensional nanostructures, one can distinguish two main approaches: one based on vapour deposition and one other based on polymer pyrolysis, corresponding to two well-known methods presented in this paper: Chemical Vapor Deposition (CVD) and electrospinning (ES) respectively. The interest in studying such two processes lies on fabrication of nanofibers which tend to be mono-crys‐ talline in the first case and more polycrystalline in the second one, with nano-sized crystals.

#### **2.1. Elaboration and characterization of silicon carbide nanowires by CVD**

We report here a simple process to fabricate SiC-based nanowires coated with either silica, SiC@SiO2, or carbon, SiC@C. In both cases, SiC nanocables (NC) were prepared from solid precursors of Si and C, *i.e.* SiO(s) and polypropylene (PP) respectively [16]. These reactants are much cheaper compared to high-purity gaseous reactants used in conventional CVD syntheses. In a typical experiment, the experiments were carried out in a furnace equipped with an alumina tube. PP was placed in an alumina boat, followed by a second alumina boat containing an equimolar mixture of Si(s) and SiO2(s) or simply a SiO(s) powder, which was partially covered by a graphite condensation plate (Figure 1). All of the experiments were conducted under argon at a very low flow rate (10 mL.min–1), at 1400 °C during 10 h.

**Figure 1.** a) Experimental setup for the growth of SiC-based nanowires; b) Picture of the two alumina boats contain‐ ing the reagents and the graphite condensation plate [16]

At high temperature, a reaction between Si and SiO2 involves the formation of gaseous SiO which is able to react with carbonated species to initiate the nanowires growth. While using a PP/SiO weight ratio of 1 (maximum), SiC@SiO2 NCs are formed whereas a ratio of 2 leads to the growth of SiC@C NWs (Figures 2 and 3). It is thus possible to control the silica thick‐ ness from 2 to 10 nm by adjusting the PP/SiO weight ratio from 1 to 0.25 for instance.

The as-obtained NWs have a diameter ranging from 20 to 40 nm. The NWs core consists of 3C-SiC, the cubic polytype of SiC, and the shell is either composed of amorphous silica (Fig‐ ure 2) or turbostratic carbon (Figure 3). The high-purity of such structures was confirmed by EELS analyses and Raman investigations allowed to determine the P-type doping of the SiCbased NWs.

**Figure 2.** SiC@SiO2 NCs prepared from 1 g PP and 4 g SiO(s). a) Slightly out-of-focus TEM image of the crude product, where the amorphous silica layer can be easily distinguished from the SiC core (the corresponding well-focused image is included as an insert); b) HRTEM image showing a 10 nm silica layer around the SiC core [16]

**Figure 3.** a,b) HRTEM images of a 30 nm SiC@C NC prepared by increasing the amount of PP to 8 g for every 4 g SiO(s) [16]

#### **2.2. Elaboration and characterization of iron carbide nanowires by electrospinning**

At high temperature, a reaction between Si and SiO2 involves the formation of gaseous SiO which is able to react with carbonated species to initiate the nanowires growth. While using a PP/SiO weight ratio of 1 (maximum), SiC@SiO2 NCs are formed whereas a ratio of 2 leads to the growth of SiC@C NWs (Figures 2 and 3). It is thus possible to control the silica thick‐

The as-obtained NWs have a diameter ranging from 20 to 40 nm. The NWs core consists of 3C-SiC, the cubic polytype of SiC, and the shell is either composed of amorphous silica (Fig‐ ure 2) or turbostratic carbon (Figure 3). The high-purity of such structures was confirmed by EELS analyses and Raman investigations allowed to determine the P-type doping of the SiC-

**Figure 2.** SiC@SiO2 NCs prepared from 1 g PP and 4 g SiO(s). a) Slightly out-of-focus TEM image of the crude product, where the amorphous silica layer can be easily distinguished from the SiC core (the corresponding well-focused image

**Figure 3.** a,b) HRTEM images of a 30 nm SiC@C NC prepared by increasing the amount of PP to 8 g for every 4 g SiO(s) [16]

is included as an insert); b) HRTEM image showing a 10 nm silica layer around the SiC core [16]

ness from 2 to 10 nm by adjusting the PP/SiO weight ratio from 1 to 0.25 for instance.

based NWs.

298 Nanowires - Recent Advances

Electrospinning is a process which has been mainly studied and used during the last decade to prepare nanowires/nanofibers. We also have noticed an increasing interest of the scientif‐ ic community during this period (Figure 4).

**Figure 4.** Histogram representing the number of scientific articles published per year during the last decade (Research performed on 4th June 2012 with "Web of Science", with "electrospinning" AND "nanofiber")

A lot of effort has been focused on the electrospinning process due to its versatility, which permits the formation of polymeric fibers as well as, after treatment, oxide and non-oxide inorganic fibers at submicro- and nanoscales [17-18]. This process is easy to scale-up to large amounts and has been recently employed to develop non-oxide ceramic nanowires. In this field, the synthesis of SiC [19-20], B4C [21], B4C/SiC [22], GaN [23] and BN [24] fibers has been reported for diameters smaller than 1μm. This production technique allows to precise‐ ly control the chemical composition of sub-microscale and nanoscale fibers which are seri‐ ous candidate for applications ranging from smart textile to biomedical, including automotive and environmental areas.

Electrospinning consists in electrostatic stretching of a molten polymer or a solution contain‐ ing a polymer solubilized in a solvent (the most common procedure). A high-voltage (HV) is applied between a spinneret (needle) and a metallic target (Figure 5).

With a low and continuous flow rate of solution through the needle, the solution drops without application of voltage but when the electric field is high enough to overcome the surface tension, a Taylor cone is formed at the needle tip and the solution is then accelerated toward the target in the form of a polymer jet. With stable working conditions, a continuous filament can be produced and collected as a solid 1D-material since the main part of the sol‐ vent evolves during the jet acceleration, in case of wet spinning. solvent evolves during the jet acceleration. Its final diameter, conventionally in the submicrometer scale, is about 1000 times lower than the inner spinneret diameter. Numerous parameters directly influence the morphology of the as-spun fibers/filaments. It is possible to fabricate dense, porous, beaded, flattened or perfectly cylindrical fibers simply by adjusting the solution properties :


and/or the spinning conditions :


**•** moisture degree.

**Figure 5.** Schema of a typical electrospinning setup

**•** surface tension,

300 Nanowires - Recent Advances

**•** viscoelastcity,

**•** high-voltage,

**•** type of target,

**•** working distance,

**•** electrical conductivity,

**•** polymer concentration,

and/or the spinning conditions :

**•** polymer/solution flow rate,

**•** inner needle diameter,

**•** solvent(s) volatility,

With a low and continuous flow rate of solution through the needle, the solution drops without application of voltage but when the electric field is high enough to overcome the surface tension, a Taylor cone is formed at the needle tip and the solution is then accelerated toward the target in the form of a polymer jet. With stable working conditions, a continuous filament can be produced and collected as a solid 1D-material since the main part of the sol‐ vent evolves during the jet acceleration, in case of wet spinning. solvent evolves during the jet acceleration. Its final diameter, conventionally in the submicrometer scale, is about 1000 times lower than the inner spinneret diameter. Numerous parameters directly influence the morphology of the as-spun fibers/filaments. It is possible to fabricate dense, porous, beaded,

flattened or perfectly cylindrical fibers simply by adjusting the solution properties :

In case of fabrication of inorganic fibers by electrospinning, a specific precursor is employed [17-24]. Iron-based materials (iron oxide and iron carbide) can be fabricated from a mixture of polyvinyl pyrrolidone (PVP), bringing the spinnability property, and iron (II) acetate (FeAc2) in acetic acid and ethanol (Figure 6) [25].

**Figure 6.** Experimental process with the different nanostructures obtained [25]

After stirring, to obtain a homogeneous and brown colored solution, the polymer solution is loaded into a syringe (5 mL) before being spun on a metallic target. After optimization of all spinning parameters, homogeneous filaments characterized by a smooth surface and an average diameter ranging from 200 to 500 nm can be obtained according to the precursor content (*i.e.* FeAc2/PVP ratio) in the initial solution (Figure 7). An average value was calcu‐ lated on 100 fibers of each sample giving a linear evolution of the fiber diameter versus the FeAc2/PVP ratio: 440 nm, 395 nm, 260 nm and 220 nm for ratios of 2.63, 1.75, 0.87 and 0.43 repectively. This phenomenon is mainly attributed to a decrease of solid content when the ratio is decreased since the PVP content is the same in all the solutions (corresponding to 7 wt%). We can notice that iron acetate is stored and sampled under argon in order to prevent it from degradation over time.

A pyrolysis step was then carried out under argon and hydrogen (95% Ar and 5% H2) with a heating rate of 3°C/min and an holding time of 4h at 450°C before cooling down. This ther‐ mal treatment allowed the organic part decomposition as well as the precursor conversion into an inorganic material. The thermal behaviour of the filaments during the pyrolysis step is still under investigation but we know that the last weight loss is due to PVP gaseous evo‐ lution up to 400-450 °C. Above this temperature, there is no more weight variation of the filaments. Temperatures above 800 °C were not studied.

**Figure 7.** SEM images of the various PVP/FeAc2 nanofibers with different wt ratios (FeAc2/PVP). (a) ratio = 2.63; (b) ratio = 1.75; (c) ratio = 0.87; (d) ratio = 0.43 [25]

The weight loss occuring during treatment is logically accompanied by a shrinkage in the macroscopic sample dimensions (already high-lighted and described in a previous work [24]), as well as in the filament diameter (Figure 8).

A pyrolysis step was then carried out under argon and hydrogen (95% Ar and 5% H2) with a heating rate of 3°C/min and an holding time of 4h at 450°C before cooling down. This ther‐ mal treatment allowed the organic part decomposition as well as the precursor conversion into an inorganic material. The thermal behaviour of the filaments during the pyrolysis step is still under investigation but we know that the last weight loss is due to PVP gaseous evo‐ lution up to 400-450 °C. Above this temperature, there is no more weight variation of the

**Figure 7.** SEM images of the various PVP/FeAc2 nanofibers with different wt ratios (FeAc2/PVP). (a) ratio = 2.63; (b)

ratio = 1.75; (c) ratio = 0.87; (d) ratio = 0.43 [25]

filaments. Temperatures above 800 °C were not studied.

302 Nanowires - Recent Advances

**Figure 8.** HRTEM of Fe3C nanoparticles surrounded with graphitic planes. Inset: TEM images of Fe3C nanofibers pre‐ pared with FeAc2/PVP = 0.87 [25]

As XRD analysis was inadequate (no signal) to characterize the type of material composing constituting the inorganic filaments, transmission electronic microscopy (TEM) was per‐ formed on several filaments (Figure 8). The first main information is that the filaments are made of nanoparticles homogeneously dispersed in a matrix made of a different material. High-resolution TEM (HR-TEM) was then used to observe that the small nanoparticles have a spherical shape, and interplanar distances of 3.37 and 2.14 Å both measured on the same nanoparticle indicating the presence of pure iron carbide, corresponding to crystalline planes (002) and (200), repectively, of the Fe3C orthorhombic phase. Moreover, these cemen‐ tite domains are surrounded by well-known concentric graphitic planes forming a coating layer of 2-3 nm on the particle surface. The reduction in a hydrogen atmosphere of the aselectrospun PVP/FeAc2 fibers is responsible for the formation of polycrystalline Fe3C nano‐ fibers embedded in carbon graphitic planes, ensuring stability and cohesion of the Fe3C nanoparticles. The presence of graphiic carbon was confirmed by Raman spectroscopy with D and G bands at 1360 and 1594 cm-1 respectively [25].

The functionalization of organic matrices for electroactuation application has been studied and is presented below, using either SiC-based or Fe3C-based nanowires as inorganic fillers. Carbon-based fibers, SiC@C and Fe3C, were expected to be interesting because of their car‐ bon coating which should enhance the surface electrical conductivity of the fillers, thus pro‐ moting the charge displacement at a nano- or a submicro-scale inside the polymer matrix. This should have a positive effect on the electrical polarization of the polymer host, directly linked to its ability to be deformed when it is subjected to an electric field.

## **3. Integration of the nanowires into polymer and study of composites actuation capabilities**

#### **3.1. Elaboration of nanocomposites**

The two polymers were polyether-based aromatic thermoplastic polyurethane (PU), pur‐ chased from Lubrizol Corporation: Estane 58888 NAT 0.21 (PU88) and Estane X-4977 NAT039 (PU75). They differ by the ratio of hard to soft segments. Hard segments (HS) com‐ prise the 4.4' methylene bis(phenyl isocyanate) (MDI) and 1,4-butanediol (BDO) and soft segments are poly(tetramethlene oxide) (PTMO). The PU88 has a density of 1.13 g/cm3 , a hardness of 88 shore A and the HS content is about 46%; the PU75 has a density of 1.07 g/ cm3 , a hardness of 75 shore A and the HS content is less, near 26% [26]. HS amount is an important factor because the physical properties depend directly on the degree of phase sep‐ aration between the hard and the soft phases.

**Figure 9.** Elaboration of the PU88-SiC nanocomposite solution by solution casting method and spin coating [26]

PU88-SiC@C and PU88-SiC@SiO2 nano-composite films were prepared following a solution casting method (Figure 9). Before use, the PU granules were heated at 80°C for 3 h to re‐ move any traces of water. They were then put in N,N-dimethylformamide (DMF, Sigma-Al‐ drich D158550, 99%). The solution was maintained at 80°C for 2 h. The SiC nanowires were dispersed into DMF using an ultrasonic processor (Hielsher UP400S, 400 W, 24 kHz, 7 mm diameter sonotrode) under the following experimental conditions depending on the type of the nanowire: 10 to 20 min duration, 80% amplitude, 80% pulse. This solution was added to that of the polymer and the resultant mixture was heated at 80 °C for 1 h under mechanical stirring, until a homogeneous and viscous solution was obtained. This solution was de‐ gassed for 24 hours at room temperature before spin coating on glass plates using a Laurell WS400 BX spin coater. This was a two steps process of 10 and 15 seconds with different rota‐ tional speed depending on the desired thickness: 200/300 rpm for 80 μm and 500/800 rpm for 35 μm. The films were then placed in an oven at 60°C for 24 hours and then removed from the glass substrate. A second heating treatment at 125°C (below the melting tempera‐ ture) for 3 hours was performed in order to eliminate any residual traces of solvent.

The elaboration method was improved for the PU75-Fe3C@C nanocomposites by using closed recipients during the dissolution and mixing steps so that the evaporation of the DMF solvent was avoided. In addition, the solution was applied on glass plates with an Elc‐ ometer 3700 doctor blade film applicator (Figure 10). This technique produced large surface films with homogeneous thickness. The following thermal treatments were the same as de‐ scribed above.

**Figure 10.** Deposition of the PU75-Fe3C nanocomposite solution with Elcometer 3700 doctor blade applicator

For both methods (Figures 9 and 10), the weight fractions of fillers varied between 0 and 4 %.

It is important to note that spin coating is widely used as a shaping method for EAPs that are introduced inside actuators devices, however, the method so-called "Doctor Blade" al‐ lows to prepare polymer or composite films with larger dimensions which is interesting for studies necessitating numerous samples for analyses (electrostriction, mechanical test, DSC, SEM,...), with verification of reproducible results as well.

#### **3.2. Characterization of the nanocomposites**

**3. Integration of the nanowires into polymer and study of composites**

The two polymers were polyether-based aromatic thermoplastic polyurethane (PU), pur‐ chased from Lubrizol Corporation: Estane 58888 NAT 0.21 (PU88) and Estane X-4977 NAT039 (PU75). They differ by the ratio of hard to soft segments. Hard segments (HS) com‐ prise the 4.4' methylene bis(phenyl isocyanate) (MDI) and 1,4-butanediol (BDO) and soft segments are poly(tetramethlene oxide) (PTMO). The PU88 has a density of 1.13 g/cm3

hardness of 88 shore A and the HS content is about 46%; the PU75 has a density of 1.07 g/

**Figure 9.** Elaboration of the PU88-SiC nanocomposite solution by solution casting method and spin coating [26]

PU88-SiC@C and PU88-SiC@SiO2 nano-composite films were prepared following a solution casting method (Figure 9). Before use, the PU granules were heated at 80°C for 3 h to re‐ move any traces of water. They were then put in N,N-dimethylformamide (DMF, Sigma-Al‐ drich D158550, 99%). The solution was maintained at 80°C for 2 h. The SiC nanowires were dispersed into DMF using an ultrasonic processor (Hielsher UP400S, 400 W, 24 kHz, 7 mm diameter sonotrode) under the following experimental conditions depending on the type of the nanowire: 10 to 20 min duration, 80% amplitude, 80% pulse. This solution was added to that of the polymer and the resultant mixture was heated at 80 °C for 1 h under mechanical stirring, until a homogeneous and viscous solution was obtained. This solution was de‐

, a hardness of 75 shore A and the HS content is less, near 26% [26]. HS amount is an important factor because the physical properties depend directly on the degree of phase sep‐

, a

**actuation capabilities**

304 Nanowires - Recent Advances

cm3

**3.1. Elaboration of nanocomposites**

aration between the hard and the soft phases.

#### *3.2.1. Structural characterization of the nanocomposites*

Microstructural observations (Figure 11), carried out by SEM using a cryo-fracture prepara‐ tion, showed that the nanowires seem to be aligned preferentially in planes parallel to the surface of the film. Whatever the shaping method used, taking into account the external forces applied on the long chains of PU, it is assumed that there is no difference of the nano‐ wires organization in the polymer host. The homogeneity of this distribution was checked over the entire width, and from the bottom to the top surface.

**Figure 11.** SEM observations performed on samples PU75-Fe3C (2.5 wt% of Fe3C); Arrows indicate the presence nano‐ fillers with perpendicular (solid arrows) and non perpendicular (dashed arrows) orientations compared to the fracture plane

**Figure 12.** DSC thermograms on heating and cooling of pure PU88, PU88-2SiC@C and PU88-2SiC@SiO2 nanocompo‐ sites films

The Differential Scanning Calorimetry (DSC) technique was used to study the effect of the nanowires on the structure of the host polymer. The thermal analysis was performed with DSC 131 evo from Setaram, under nitrogen atmosphere. The samples (about 20 mg) were placed in closed 100-μl aluminum crucibles and cooled from ambient temperature down to -100°C, then heated to 210°C and finally cooled to RT. The heating and cooling ramps were performed at 10°C.min-1.

The thermograms for the pure PU88, and the PU88-2SiC@C and PU88-2SiC@SiO2 composi‐ tions are presented below (Figure 12).

The glass transition temperature for the soft segments (Tg), melting temperatures (TII and TIII), heat of fusion (ΔHm), crystallization temperature (Tc) and heat of crystallization (ΔHc) are summarized in Table 1.


**Table 1.** DSC results obtained for the PU88-SiC nanocomposite films

**Figure 11.** SEM observations performed on samples PU75-Fe3C (2.5 wt% of Fe3C); Arrows indicate the presence nano‐ fillers with perpendicular (solid arrows) and non perpendicular (dashed arrows) orientations compared to the fracture

**Figure 12.** DSC thermograms on heating and cooling of pure PU88, PU88-2SiC@C and PU88-2SiC@SiO2 nanocompo‐

plane

306 Nanowires - Recent Advances

sites films

The first thermal phenomenon near -45°C was related to the glass transition (TgSS) of the soft segments (SS). This value can give an estimation of the amount of hard segments (HS) dis‐ solved in the soft domains and as a consequence an indication of the purity of the soft phase [27-28]. When the fraction of nanowires was increased, no significant change of Tg was ob‐ served. It was thus concluded that the degree of HS-SS mixing was not modified by the in‐ corporation of nanowires.

Some effects of relaxation of the polymer chain, a modification at the SS/HS interface or ther‐ mal endotherm seemed to explain the behavior of heat flow near 60°C. At temperatures in the range 150-170°C, it was observed a bimodal endotherm that can be related to the micromixing of non-crysytalline or semi-crystalline hard and soft phases followed by the melting of crystalline HS [29]. Another interpretation of the bimodal peak could be the melting of crystalline hard segments: the two peaks represent the two characteristic lengths of HS [27]. In both cases, the increase of the samples' crystallinity would be accompanied by an increase of these temperatures and/or an increase of the total enthalpy of the 2 phenomena.

With the incorporation of SiC@C nanowires, the endothermal peaks remained unmodified, temperature and corresponding enthalpy did not change. With SiC@SiO2, one can observe only a slight decrease of the temperature of the first endotherm related to the micromixing of hard and soft phases.

During the cooling experiments, a more important influence of the nanowires on the crystal‐ lization behavior was observed. When the amount of nanowires was increased, the crystalli‐ zation temperature was raised but without a significant change in enthalpy. The nanowires seemed to act as nucleating agents improving the crystallization. There is a more important effect on the crystallization of the polyurethane with the incorporation of SiC@C compared to SiC@SiO2 (Figure 12).

The same study has been performed on nanocomposite films based on PU75 and Fe3C@C nanowires and the values of Tg and melting temperatures and heat of fusion are summar‐ ized in Table 2. The PU75 is less rigid than the PU88, as indicated by the value of the glass transition temperature which is 25°C under the PU88 one. With the incorporation of Fe3C@C, there is no modification of the thermal properties during heating and cooling, and so no morphological change of the PU75 polymer host.


**Table 2.** DSC results obtained for the PU75-Fe3C@C nanocomposite films

It can be concluded from these DSC results that, whatever the PU, no major microstructural modification of the matrix was observed with the incorporation of nanowires.

#### *3.2.2. Dielectric properties of the nanocomposites*

Prior to the electrical measurements, the samples were coated with 20 nm thick gold by sputtering. The relative permittivity ε', the dielectric loss factor tan δ and the real part of the electrical conductivity σ' were obtained with a Solartron 1255 impedance / gain-phase ana‐ lyzer and a 1296 dielectric interface. The voltage was set at 1 VRMS and a frequency range was performed in the range 0.1-105 Hz.

Figure 13 presents the electrical conductivity versus frequency of some PU88/SiC nanowires composites. In all samples, two frequency regions can be observed: one for which the con‐ ductivity was almost constant (low frequencies < 10Hz) and which is attributed to charge displacements and another one (high frequencies > 10Hz) where the conductivity increased with the frequency and corresponds to the dielectric losses [30].

Compared to the SiC@SiO2, the SiC@C nanowires lead to a slightly more important increase of the polyurethane conductivity. This is in goog agreement with the conductive carbon lay‐ er which covers the SiC wires, instead of the SiO2 one.

Synthesis and Characterization of Advanced Carbon-Based Nanowires – Study of Composites Actuation... http://dx.doi.org/10.5772/54384 309

**Figure 13.** The electrical conductivity versus frequency for pure PU and PU88-2SiC composites

With the incorporation of SiC@C nanowires, the endothermal peaks remained unmodified, temperature and corresponding enthalpy did not change. With SiC@SiO2, one can observe only a slight decrease of the temperature of the first endotherm related to the micromixing

During the cooling experiments, a more important influence of the nanowires on the crystal‐ lization behavior was observed. When the amount of nanowires was increased, the crystalli‐ zation temperature was raised but without a significant change in enthalpy. The nanowires seemed to act as nucleating agents improving the crystallization. There is a more important effect on the crystallization of the polyurethane with the incorporation of SiC@C compared

The same study has been performed on nanocomposite films based on PU75 and Fe3C@C nanowires and the values of Tg and melting temperatures and heat of fusion are summar‐ ized in Table 2. The PU75 is less rigid than the PU88, as indicated by the value of the glass transition temperature which is 25°C under the PU88 one. With the incorporation of Fe3C@C, there is no modification of the thermal properties during heating and cooling, and

Pure PU75 -68.6 153 174 2.3 PU75-2.5Fe3C@C -69.0 152 175 3.0 PU75-5Fe3C@C -71.7 152 174 2.4

It can be concluded from these DSC results that, whatever the PU, no major microstructural

Prior to the electrical measurements, the samples were coated with 20 nm thick gold by sputtering. The relative permittivity ε', the dielectric loss factor tan δ and the real part of the electrical conductivity σ' were obtained with a Solartron 1255 impedance / gain-phase ana‐ lyzer and a 1296 dielectric interface. The voltage was set at 1 VRMS and a frequency range was

Figure 13 presents the electrical conductivity versus frequency of some PU88/SiC nanowires composites. In all samples, two frequency regions can be observed: one for which the con‐ ductivity was almost constant (low frequencies < 10Hz) and which is attributed to charge displacements and another one (high frequencies > 10Hz) where the conductivity increased

Compared to the SiC@SiO2, the SiC@C nanowires lead to a slightly more important increase of the polyurethane conductivity. This is in goog agreement with the conductive carbon lay‐

modification of the matrix was observed with the incorporation of nanowires.

Tg (°C) TII (°C) TIII (°C) ΔHm (J/g)

of hard and soft phases.

308 Nanowires - Recent Advances

to SiC@SiO2 (Figure 12).

so no morphological change of the PU75 polymer host.

**Table 2.** DSC results obtained for the PU75-Fe3C@C nanocomposite films

Hz.

with the frequency and corresponds to the dielectric losses [30].

er which covers the SiC wires, instead of the SiO2 one.

*3.2.2. Dielectric properties of the nanocomposites*

performed in the range 0.1-105

The frequency/conductivity curves show that even if the conductivity is slightly enhanced with the incorporation of nanowires, the percolation threshold is not reached even with higher concentrations of SiC (3% tested) or SiO2 (4% tested). This threshold could be estimat‐ ed as the content located between that for which the conductivity still depended on the fre‐ quency and that for which it became completely independent of the frequency.

Figure 14 exhibits the permittivity versus frequency of some PU88/SiC nanowires composites

**Figure 14.** The relative permittivity versus frequency for pure PU and PU88-2SiC composites.

Whatever the sample, the permittivity increases as the frequency is lowered, which is in good agreement with Maxwell Wagner mechanism [31]. The composites present an higher permittivity than the pure polyurethane one: for example, at 0.1 Hz, it increases from 9 for pure PU to 12 for PU88-2SiC@SiO2 and to 18 for PU88-2SiC@C.

The effect is more pronounced for the SiC@C compared to SiC@SiO2 certainly because the first type of filler has a conductive carbon layer on the surface instead of the insulating SiO2 layer.

For the PU75-Fe3C system, the same dielectric characterization was made (Figure 15).

**Figure 15.** The conductivity (a) and relative permittivity (b) versus frequency for composites with varying SiC contents (a) PU88-SiC@C and (b) PU88-SiC@SiO2

The conductivity curves for the PU75 system shows like for the PU88 two regions depend‐ ing on the value of the frequency. Whatever the frequency, the conductivity is not really modified with or without Fe3C.

Compared to the SiC@C filler, the Fe3C@C has a very slight influence on the permittivity of the samples at low frequency. But at a frequency near 1 kHz, the gain was of the same order of magnitude for the two fillers i.e. 1.35 fold.

#### *3.2.3. Mechanical properties of the nanocomposites*

The mechanical behaviour of the composite films was evaluated on 25 x 10 mm2 samples, us‐ ing a house made tensile test (Figure 16). On one end, the sample was clamped onto a New‐ port platin and the other end onto a force sensor. The stage was moved using a function generator at a frequency of 50 Hz and the dynamic force was acquired on an oscilloscope. Measurements were performed over a deformation range from 0 to 100%. The Young's modu‐ lus was determined at the beginning of the curve strain / stress for elongations less than 4%.

Figure 17 exhibits the evolution of the Young's modulus of the nanocomposites films as a function of the content of nanowires.

Synthesis and Characterization of Advanced Carbon-Based Nanowires – Study of Composites Actuation... http://dx.doi.org/10.5772/54384 311

**Figure 16.** Experimental set-up for measuring the mechanical properties of the nanocomposite films

**Figure 17.** The Young's modulus versus the content of nanowires for PU88-Si@C and PU75-Fe3C@C

For the two polyurethane systems, it is observed an expected little increase of the Young's modulus with the incorporation of nanowires. For the same content of nanowires, for exam‐ ple 2.5%, there is a 1.3 gain factor for PU88-SiC@C and 1.2 gain factor for PU75-Fe3C@C.

#### *3.2.4. Electromechanical capabilities*

Whatever the sample, the permittivity increases as the frequency is lowered, which is in good agreement with Maxwell Wagner mechanism [31]. The composites present an higher permittivity than the pure polyurethane one: for example, at 0.1 Hz, it increases from 9 for

The effect is more pronounced for the SiC@C compared to SiC@SiO2 certainly because the first type of filler has a conductive carbon layer on the surface instead of the insulating SiO2

**Figure 15.** The conductivity (a) and relative permittivity (b) versus frequency for composites with varying SiC contents

The conductivity curves for the PU75 system shows like for the PU88 two regions depend‐ ing on the value of the frequency. Whatever the frequency, the conductivity is not really

Compared to the SiC@C filler, the Fe3C@C has a very slight influence on the permittivity of the samples at low frequency. But at a frequency near 1 kHz, the gain was of the same order

ing a house made tensile test (Figure 16). On one end, the sample was clamped onto a New‐ port platin and the other end onto a force sensor. The stage was moved using a function generator at a frequency of 50 Hz and the dynamic force was acquired on an oscilloscope. Measurements were performed over a deformation range from 0 to 100%. The Young's modu‐ lus was determined at the beginning of the curve strain / stress for elongations less than 4%.

Figure 17 exhibits the evolution of the Young's modulus of the nanocomposites films as a

samples, us‐

The mechanical behaviour of the composite films was evaluated on 25 x 10 mm2

For the PU75-Fe3C system, the same dielectric characterization was made (Figure 15).

pure PU to 12 for PU88-2SiC@SiO2 and to 18 for PU88-2SiC@C.

layer.

310 Nanowires - Recent Advances

(a) PU88-SiC@C and (b) PU88-SiC@SiO2

modified with or without Fe3C.

of magnitude for the two fillers i.e. 1.35 fold.

*3.2.3. Mechanical properties of the nanocomposites*

function of the content of nanowires.

**a.** Electric field-induced thickness strain measured by a double-beam interferometer at low frequency (0.1Hz).

The electric field-induced thickness strain S was measured with a double-beam laser inter‐ ferometer (Agilent 10889B), with a precision in the order of 10 nm. The film samples of 25 mm diameter were placed on a horizontal stainless steel disc (20 mm in diameter), and a sec‐ ond brass disc placed on the upper side of the film allowing the application of a bipolar elec‐ tric field at 0.1 Hz. The voltage was supplied by a function generator (Agilent 33250A) amplified 1000-fold through a high-voltage lock-in amplifier (Trek 10/10B). A guard ring was added to the sample surface and grounded so as to eliminate any surface leakage cur‐ rents. Figure 18 gives photographs of the experimental set-up for strain measurements.

**Figure 18.** Experimental set-up for strain measurements with laser interferometer at 0.1 Hz.

Figure 19-a) presents the thickness strain amplitude – S3 (S3 is negative since the film is com‐ pressed) versus the electric field amplitude of pure and filled with 0.5% SiC@C PU88 films of different thicknesses.

All studied films exhibited the same type of strain variations (except the 20 μm – thick nano‐ composite because electric breakdown occurred above 18V/μm): a first region where the strain increases regularly with the electric field amplitude indeed and a second region at higher fields where an asymptotic value is reached. This is the result of the hyperelastic be‐ haviour of such PU materials.

The deformation depends on the thickness of the film: lower electric fields were required to obtain a given strain level before saturation as the film's thickness increased. The 20μmthick nanocomposite film presented the maximal thickness strain of about 18% at 18 V/μm. At moderate electric values, the greatest effect was observed for the 100 μm-thick PU-SiC@C film since the strain was equal to 10% at a field as low as 6 V/μm.

The electric field-induced thickness strain S was measured with a double-beam laser inter‐ ferometer (Agilent 10889B), with a precision in the order of 10 nm. The film samples of 25 mm diameter were placed on a horizontal stainless steel disc (20 mm in diameter), and a sec‐ ond brass disc placed on the upper side of the film allowing the application of a bipolar elec‐ tric field at 0.1 Hz. The voltage was supplied by a function generator (Agilent 33250A) amplified 1000-fold through a high-voltage lock-in amplifier (Trek 10/10B). A guard ring was added to the sample surface and grounded so as to eliminate any surface leakage cur‐ rents. Figure 18 gives photographs of the experimental set-up for strain measurements.

Laser interferometer

Oscilloscope Function generator

Current amplifier

Laser HT

**Figure 18.** Experimental set-up for strain measurements with laser interferometer at 0.1 Hz.

Figure 19-a) presents the thickness strain amplitude – S3 (S3 is negative since the film is com‐ pressed) versus the electric field amplitude of pure and filled with 0.5% SiC@C PU88 films

All studied films exhibited the same type of strain variations (except the 20 μm – thick nano‐ composite because electric breakdown occurred above 18V/μm): a first region where the strain increases regularly with the electric field amplitude indeed and a second region at higher fields where an asymptotic value is reached. This is the result of the hyperelastic be‐

Voltage amplifier

312 Nanowires - Recent Advances

Mirror

of different thicknesses.

haviour of such PU materials.

High voltage wire

Brass disc Electrode supérieur e

sample

Disc shaped polymer

Ground wire

**Figure 19.** Evolution of the strain amplitude versus the electric field amplitude for various thicknesses of pure PU88 and PU88-0.5SiC@C films at 0.1 Hz. (b) The evolution of the strain amplitude versus the thickness of pure PU88 and PU88-0.5SiC@C films at E= 6.5 V/µm [11]

As the relative permittivity was found to be higher in the SiC@C composites, the results showed an increase of the interfacial polarization induced by a pronounced space charges effect. Space charges accumulate in PU and more effectively in PU-SiC@C composites and it was still evidenced by the large dependence of the strain on the film thickness, particularly for low and medium electric field values (E < 10 V/μm). These charges are generally elec‐ trons injected in the bulk from the electrodes and/or impurities (ions) in the sample which are trapped to the different structural interfaces of the material. The consequence of the presence of space charges is the appearance of high electric field local densities, located at the various trapping sites (defects, interfaces), thus yielding a non-uniform field distribution in the film thickness and consequently a high deformation [32].

However, the film must be thick enough to present such a strain improvement by space charges effect, as it is shown in Figure 19-b which gives the evolution of strain at 6.5 V/μm electric field versus thickness. If the film was too thin, the strain was not increased because the interfaces that were able to trap the space charges started to overlap and as a conse‐ quence charge homogenization occured [14, 33]. When the film thickness increased, the strain also increased, with a more pronounced effect for the nanocomposite. It should be emphasized that the polyurethane exhibited a considerable ability to collect space charges certainly due to the two-phase structure composed of hard and soft segments inducing many interfaces. The presence of space charges were experimentally evidenced by studying the steady-state current – electric field characteristic of the films [11].

DSC measurements presented in previous section were performed in order to determine whether the SiC@C nanowires themselves or polymer chains/nanowires' interfaces behaved as additionnal space charge traps and/or if they increased the level of phase mixing between soft and hard segments and consequently the structural interface density. The DSC results show that there is no modification of the phase mixing so it can be concluded that space charge effects in the composites were not enhanced by an increase of the structural interface density, but rather by the SiC@C nanowires themselves, acting as charge traps.

Another study was performed with SiC@SiO2 nanowires as fillers with the laser interferome‐ ter method at low frequency [34]. The electric field induced thickness strain response has been investigated for the polystyrene-grafted silica coating of SiC NWs (PU88-2SiC@SiO2@PS) nanocomposites and compared to the pure polyurethane film and the PU88-2SiC@SiO2 nanocomposite without polystyrene grafting.

Polyurethane polymer filling with both types of nanowires leads to a higher electromechani‐ cal response than for the pure PU88 film. At a moderate electric field of 10 V/μm, 2SiC@SiO2@PS loading increased the strain level of pure PU88 by a factor of 2.2. The im‐ provement came partially from polystyrene grafting since the PU-2SiC@SiO2 films showed only a 1.7 times increase. It was shown in the article that since a higher strain was observed for the PU88-2SiC@SiO2@PS films, the polystyrene chains grafting seemed to act as a dis‐ persing agent, thus yielding a homogeneous distribution of NWs in the composite.

**b.** Deflection strain under electric field of a diaphragm type dielectric actuator working at high frequency (near 1kHz).

Another way to investigate the electromechanical properties of polymer films is to measure the expension in the 31 direction (which is directly proportional to the 33 contraction). It was chosen to investigate it using a diaphragm, which is also a good candidates for integration of these polymer films into Microsystems [35].

The diaphragm device (Figure 20) comprised a sample holder and the electroded sample.

The sample holder was fabricated in thick PMMA sheets. Two diameter holes of 5.5 mm and 20 mm were drilled into the two parts of the device. The 5.5 mm hole constituted the dia‐ phragm. Gold electrode was sputtered onto each part for electrical connections. Samples of 25 mm in diameter were electroded using gold sputtering as shown in the figure 19 and then aligned with the 5.5 mm hole and the electrode on the rear part of sample holder. The sample adhered onto the sample holder by electrostatic forces.

DSC measurements presented in previous section were performed in order to determine whether the SiC@C nanowires themselves or polymer chains/nanowires' interfaces behaved as additionnal space charge traps and/or if they increased the level of phase mixing between soft and hard segments and consequently the structural interface density. The DSC results show that there is no modification of the phase mixing so it can be concluded that space charge effects in the composites were not enhanced by an increase of the structural interface

Another study was performed with SiC@SiO2 nanowires as fillers with the laser interferome‐ ter method at low frequency [34]. The electric field induced thickness strain response has been investigated for the polystyrene-grafted silica coating of SiC NWs (PU88-2SiC@SiO2@PS) nanocomposites and compared to the pure polyurethane film and the

Polyurethane polymer filling with both types of nanowires leads to a higher electromechani‐ cal response than for the pure PU88 film. At a moderate electric field of 10 V/μm, 2SiC@SiO2@PS loading increased the strain level of pure PU88 by a factor of 2.2. The im‐ provement came partially from polystyrene grafting since the PU-2SiC@SiO2 films showed only a 1.7 times increase. It was shown in the article that since a higher strain was observed for the PU88-2SiC@SiO2@PS films, the polystyrene chains grafting seemed to act as a dis‐

**b.** Deflection strain under electric field of a diaphragm type dielectric actuator working at

Another way to investigate the electromechanical properties of polymer films is to measure the expension in the 31 direction (which is directly proportional to the 33 contraction). It was chosen to investigate it using a diaphragm, which is also a good candidates for integration

The diaphragm device (Figure 20) comprised a sample holder and the electroded sample.

**Figure 20.** Diaphragm device used for the measurement of the deflection strain under electrical field at high frequen‐

persing agent, thus yielding a homogeneous distribution of NWs in the composite.

density, but rather by the SiC@C nanowires themselves, acting as charge traps.

PU88-2SiC@SiO2 nanocomposite without polystyrene grafting.

high frequency (near 1kHz).

314 Nanowires - Recent Advances

cy [35]

of these polymer films into Microsystems [35].

The polymer film was excited in a pseudo piezoelectric mode by applying a DC electrical field in addition to the AC excitation. The sinusoidal voltage was applied using a function generator (Agilent 33200A) and amplified by the Trek Model 10/10 high voltage power am‐ plifier The DC component induces a constant polarization of the membrane, and thus, a pie‐ zoelectric behavior. In this study, all data were recorded in pseudo-piezoelectric mode and at the first resonant frequency. In addition, the DC bias voltage was chosen equal to the AC voltage magnitude. The membrane deformation was recorded using a Polytech OFV-5000 vibrometer equiped with a OFV-505 sensor head. The reflectivity of the gold electrode was used to monitor the deflection. The deflection strain is defined as Sd=S/d where d is the de‐ flection and h the sample thickness. The deflection is caused by the compressive strain un‐ dergone by the film rigidily clamped at the edge.

The deflection strain magnitude for PU75 loaded with 0, 2.5 and 5wt% of Fe3C@C has been studied as a function of the square applied electrical field magnitude [36]. The results are presented in figure 21.

**Figure 21.** The deflection strain versus the square applied electrical field for PU75-Fe3C@C nanocomposites films

At low and medium electric fields (E < 10 V/μm), the deflection strain increased quasi-line‐ arly with E2 , thus confirming the electrostrictive nature of the electromechanical response. For higher electric field, a saturation effect corresponding to an asymptotic constant value of strain was observed and can be attributed to the saturation of the electrical polarization [37].

The deflection is increased by 2 to 7 fold from pure PU75 to 2.5% load PU depending on the electrical field magnitude, and is higher than that measured on the same PU filled with car‐ bon black [35]. A considerable deflection strain of more than 130 % was found for the PU75-2.5Fe3C@C nanocomposite for a 18 V/μm electrical field.

With higher amount of Fe3C nanowires, the deflection decreases. This decrease could be re‐ lated to an agglomeration of the nanowires. Some complementary structural charateriza‐ tions are under progress by SEM and Focused Ion Beam in order to fully understand this behavior. Even if the global permittivity and dielectric loss do not increase dramatically, it is possible that some agglomerates appear locally inside the film and contribute to decrease the interfacial polarization and consequently the density of trapped charges. This effect yields a decrease in the deflection strain amplitude.

By using the same method, measurements have been performed on different PU88-SiC@C nanocomposites. Figure 22 exhibits the strain deflection versus the square of the electric field. Even if the polyurethane host is not the same for the study, it is possible to compare the influence of the two types of nanowires, ie SiC@C and Fe3C@C, on the actuation per‐ formances.

**Figure 22.** The deflection strain versus the square applied electrical field for PU88-SiC@C nanocomposites films

The electric field induced a deflection of PU88 films which was lower than that of the PU75 films; it is in agreement with a similar value of the relative permittivity in both cases and a lower value of Young's modulus of the PU75 films. Indeed the strain depends directly on the ratio of permittivity over Young's modulus as shown in the equation (5).

At high electrical fields, the PU88-3SiC@C shows a 25% deflection strain which is 6 times higher than for pure PU88. The gain in actuation is similar to that measured on PU75-2.5Fe3C@C. This is consistent with results presented in previous section: dielectric studies for both nanocomposites have shown an increase of the relative permittivity near 1 kHz with a gain of 1.35. Moreover, the mechanical reinforcement was also comparable for the 2 types of composite films.

### **4. Conclusions**

Two different methods of synthesis were investigated and exploited to develop new carbonbased nanowires, SiC@SiO2, SiC@C and Fe3C, used as fillers in PU matrices to improve the deflection capabilities of these electroactive polymers.

Polyurethane composites filled with low amounts of SiC@C and Fe3C@C nanowires (0 – 5 wt %) were prepared by two solvent methods and yielded to very large surface films in the range of 30-80 μm in thickness. It can be noted that the deposit way was different for the two systems: spin coating for PU88-SiC@C films and doctor blade application for PU75- Fe3C@C, both methods being used for preparation of actuators made of polymers.

The dielectric properties were studied at different frequencies and showed an increasing of the relative permittivity of the composites with the amount of nanowires. Permittivity and conductivity results showed also that the percolation threshold was not reached whatever the studied system. The mechanical properties were performed by using a simple tensile test and it was shown that the increase in the Young's modulus induced by the nanowires was only moderate.

Based on DSC analysis, only a very slight modification of the polyurethane host was detect‐ ed with the incorporation of the nanowires and no difference in the degree of hard seg‐ ments/soft segments mixing was found.

The actuation capabilities were studied by measuring the strain under an applied electric field with two methods.

The first method consisted in the measure of the electric field-induced thickness strain S with a double-beam laser interferometer at low frequency (0.1 Hz). At a moderate electrical field (6V/μm), it was found a 3 fold increase of the electrostriction coefficient for only 0.5%wt of SiC@C into PU88 polymer. The polyurethane is a good candidate for electrome‐ chanical applications, it has a good ability to collect space charges due to its special structure consisting of a mixing of hard and soft segments. It was shown that the nanowires acted as charge traps and then enhanced the electrical field induced deformation.

The second method is to measure the deflection strain of a diaphragm type dielectric actua‐ tor working at high frequency (near 1kHz). The deflection strain was considerably enhanced with the incorporation of either Fe3C@C and SiC@C nanowires into polyurethane films whatever the type of the polyurethane host. The maximum deflection strains measured at high electrical fields (18 V/μm) reached 130 % for the PU75-Fe3C@C nanocomposites and 25% for the PU88-3SiC@C. These values are considerable and at least 6 times higher than those obtained with corresponding pure polymer films.

Due to their evident large actuation capabilities, these new nanocomposites based on poly‐ urethane and carbon-based nanowires are very promising materials for micro electrome‐ chanical systems.

## **Acknowledgments**

With higher amount of Fe3C nanowires, the deflection decreases. This decrease could be re‐ lated to an agglomeration of the nanowires. Some complementary structural charateriza‐ tions are under progress by SEM and Focused Ion Beam in order to fully understand this behavior. Even if the global permittivity and dielectric loss do not increase dramatically, it is possible that some agglomerates appear locally inside the film and contribute to decrease the interfacial polarization and consequently the density of trapped charges. This effect

By using the same method, measurements have been performed on different PU88-SiC@C nanocomposites. Figure 22 exhibits the strain deflection versus the square of the electric field. Even if the polyurethane host is not the same for the study, it is possible to compare the influence of the two types of nanowires, ie SiC@C and Fe3C@C, on the actuation per‐

**Figure 22.** The deflection strain versus the square applied electrical field for PU88-SiC@C nanocomposites films

the ratio of permittivity over Young's modulus as shown in the equation (5).

The electric field induced a deflection of PU88 films which was lower than that of the PU75 films; it is in agreement with a similar value of the relative permittivity in both cases and a lower value of Young's modulus of the PU75 films. Indeed the strain depends directly on

At high electrical fields, the PU88-3SiC@C shows a 25% deflection strain which is 6 times higher than for pure PU88. The gain in actuation is similar to that measured on PU75-2.5Fe3C@C. This is consistent with results presented in previous section: dielectric studies for both nanocomposites have shown an increase of the relative permittivity near 1 kHz with a gain of 1.35. Moreover, the mechanical reinforcement was also comparable for

Two different methods of synthesis were investigated and exploited to develop new carbonbased nanowires, SiC@SiO2, SiC@C and Fe3C, used as fillers in PU matrices to improve the

yields a decrease in the deflection strain amplitude.

formances.

316 Nanowires - Recent Advances

the 2 types of composite films.

deflection capabilities of these electroactive polymers.

**4. Conclusions**

Authors thank the Institut Carnot for its financial support and Veronique Perrin from LGEF (EA682 INSA Lyon) for mechanical tests and thickness strain measurements with laser inter‐ ferometer. The authors also acknowledge the CTμ (Centre Technologique des Microstruc‐ tures) of the Université Lyon 1 for access to the SEM.

## **Author details**

V. Salles1 , L. Seveyrat2\*, T. Fiorido1 , L. Hu1 , J. Galineau2 , C. Eid1 , B. Guiffard2 , A. Brioude1 and D. Guyomar2

\*Address all correspondence to: laurence.seveyrat@insa-lyon.fr

1 Laboratoire des Multimatériaux et Interfaces, UMR 5615 CNRS, Université Lyon1, Univer‐ sité de Lyon, Bat. Berthollet, France

2 Laboratoire de Génie Electrique et Ferroélectricité, LGEF – INSA Lyon, France

## **References**


[9] Guiffard B, Seveyrat L, Sebald G, Guyomar D. Enhanced Electric Field-Induced Strain in Non-percolative Carbon Nanopowder/Polyurethane Composites. Journal of Physics D: Applied Physics 2006;39(14) 3053-3057.

**Author details**

318 Nanowires - Recent Advances

, L. Seveyrat2\*, T. Fiorido1

and D. Guyomar2

sité de Lyon, Bat. Berthollet, France

2008;33 173-181.

Science; 2008.

, L. Hu1

2 Laboratoire de Génie Electrique et Ferroélectricité, LGEF – INSA Lyon, France

\*Address all correspondence to: laurence.seveyrat@insa-lyon.fr

Review of Materials Research 2010;40 131-151.

Technology 2011;71(6) 885-892.

2008;20(11) 2074-2079.

, J. Galineau2

1 Laboratoire des Multimatériaux et Interfaces, UMR 5615 CNRS, Université Lyon1, Univer‐

[1] Bar-Cohen Y, Zhang Q. Electroactive Polymer Actuators and Sensors. MRS Bulletin

[2] Carpi F, Rossi D, Kornbluh R, Pelrine R, Sommer-Larsen P. Dielectric Elastomers as Electromechanical Transducers : Fundamentals, Materials, Devices, Models and Ap‐ plications of an Emerging Electroactive Polymer Technology. Amsterdam: Elsevier

[3] Nan CW, Shen Y, Ma J. Physical Properties of Composites Near Percolation. Annual

[4] Wongtimnoi K, Guiffard B, Bogner-Van de Moortèle A, Seveyrat L, Gauthier C, Cav‐ aillé JY. Improvement of Electrostrictive Properties of a Polyether-based Polyur‐ ethane Elastomer Filled with Conductive Carbon Black. Composites Science and

[5] Park C, Kang JH, Harrison JS, Costen RC, Lowther SE. Actuating Single Wall Carbon Nanotube-Polymer Composites: Intrinsic Unimorphs. Advanced Materials

[6] Zhang S, Zhang N, Huang C, Ren K, Zhang Q. Microstructure and Electromechanical Properties of Carbon Nanotubes/Poly(vinylidene fluoride-trifluoroethylene-chloro‐

[7] He F, Fan JT, Lau S. Thermal, Mechanical and Dielectric Properties of Graphite Rein‐ forced Poly(vinylidene fluoride) Composite. Polymer Testing 2008;27(8) 964-970.

[8] Seveyrat L, Chalkha A, Guyomar D, Lebrun L. Preparation of Graphene Nanoflakes/ Polymer Composites and Their Performances for Actuation and Energy Harvesting

fluoroethylene) Composites. Advanced Materials 2005;17(15) 1897-1901.

Applications. Journalf of Applied Physics 2012;111(10) 104904-104913.

, C. Eid1

, B. Guiffard2

,

V. Salles1

A. Brioude1

**References**


[35] Galineau J, Guiffard B, Seveyrat L, Lallart M, Guyomar D. Study and Modeling of an Electrostrictive Polyurethane Diaphragm Loaded with Conductive Carbon Black. Ac‐ cepted manuscript in Sensors and Actuators A: Physical http://dx.doi.org/10.1016/ j.sna.2012.09.021.

[22] Guron MM, Wei XL, Welna D, Krogman N, Kim MJ, Allcock H, Sneddon LG. Precer‐ amic Polymer Blends as Precursors for Boron-Carbide/Silicon-Carbide Composite

[23] Wu H, Sun Y, Lin D, Zhang R, Zhang C, Pan W. GaN Nanofibers based on Electro‐ spinning: Facile Synthesis, Controlled Assembly, Precise Doping, and Application as

High Performance UV Photodetector. Advanced Materials 2009;21(2) 227-231.

ymer-derived Ceramics Route. Nanoscale 2010;2 215-217.

al des Sciences Appliquées de Lyon ; 2011.

of Modern Physics 2000;72(3) 873-892.

Advanced Technologies 1998;9(6) 317-321.

Nanotechnology 2010;21 145610.

2010;21(12) 125701.

320 Nanowires - Recent Advances

1997;64(4) 803-817.

1986;19(3) 714-720.

99-109.

[24] Salles V, Bernard S, Brioude A, Cornu D, Miele P. A New Class of Boron Niride Fi‐ bers with Tunable Properties by Combining An Electrospinning Process and the Pol‐

[25] Eid C, Brioude A, Salles V, Plenet JC, Asmar R, Monteil Y, Khoury R, Khoury A, Miele P. Iron-based 1D Nanostructures by Electrospinning Process. Nanotechnology

[26] Wongtimnoi K. Polyurethanes électrostricitifs et nanocomposites : caractérisation et analyse des mécanismes de couplage électromécaniques. PhD thesis. Institut Nation‐

[27] Martin DJ, Meijs GF, Gunatillake PA, McCarthy SJ, Renwick GM. The Effect of Aver‐ age Soft Segment length on Morphology and Properties of a Serie of Polyurethane Elastomers. II. SAXS-DSC Annealing Study. Journal of Applied Polymer Science

[28] Chen KS, Leon Yu T. Chen YS, Lin TL, Liu WJ. Soft and Hard-segment Phases Segre‐ gation of Polyester-based Polyurethane. Journal of Polymer Research 2001;8(2)

[29] Koberstein JT, Russell TP. Simultaneous SAXS-DSC Study of Multiple Endothermic Behavior in Polyether-based Polyurethane Block Copolymers. Macromolecules

[30] Dyre JC, Schroder TB. Universality of AC Conduction in Disordered Solids. Reviews

[31] Kremer F, Schönhals A. Broadband Dielectric Spectroscopy. Springer-Verlag; 2003.

[32] Su J, Zhang QM, Ting RY. Space-charge-enhanced Electromechanical Response in Thin-film Polyurethane Elastomers. Applied Physics Letters 1997;71(3) 386-388.

[33] Su, J, Zhang QM, Wang PC, MacDiarmid AG, Wynne KJ. Preparation and Character‐ ization of Electrostrictive Films with Conductive Polymer Electrodes. Polymers for

[34] Rybak A, Warde M, Beyou E, Chaumont P, Bechelany M, Brioude A, Toury B, Cornu D, Miele P, Guiffard B, Seveyrat L, Guyomar D. Synthesis of Polystyrene Coated SiC Nanowires as Fillers in a Polyurethane Matrix for Electromechanical Conversion.

Ceramics and Ceramic Fibers. Chemistry of Materials 2009;21(8) 1708-1715.


**Section 4**

**Other Nanowires**

## **Cadmium Selenide-Gold (CdSe-Au) Hybrid Nanowires: Synthesis and Study of Charge Transfer**

Xiaowei Teng and Xiaoxiong Jia

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52828

## **1. Introduction**

Complex hybrid nanostructures, which are assembled directly from different components without the aid from organic molecular bridges, are of major scientific and technological in‐ terest as they provide a variety of opportunities in various applications. By coupling compo‐ nents, each with its different functionalities, the hybrid nanostructures offer entirely novel properties that do not appear in their monocomponent counterparts. Many hybrid nano‐ structures, such as semiconductor/noble metal, semiconductor/semiconductor, noble metal/ noble metal, noble metal/magnetic metal or metal oxide, and semiconductor/magnetic metal or metal oxide, recently were synthesized via the Polyol processes, photochemical routes, seeded growth, or site-selective lithographic approaches. However, there is a limited ability to finely control the size, morphology, phase-segregation, and position of the components in hybrid nanostructures, partially reflecting the structural complexity imposed by the random nucleation of the second component.

Among various metals, gold (Au) has been often studied in the hybrid system. The Au atom has the electron configuration of (4f)14(5d)10(6s)1 . Due to s-p-d hybridization, a small amount of itinerant holes existing at the d band in bulk Au has been observed. The corresponding hybridization resulted in the electron configuration as of (4f)14(5d)10-x(6s)1+x. Thus, d-band holes can be generated in Au, provide various interesting properties in magnetism, catalysis and photonics [1, 2, 3].

Semiconductor nanomaterial have remarkable technological impacts on biomedical fluores‐ cent labels, light emitting devices (LEDs), white-light laser sources, near-infrared emitters, solar cells, and tunable polarized lasers. Among various semiconductors, semiconductornoble metal hybrid structures are of particular interest for their great potential application in catalysis and photonics. The synthesis of II-VI and III-V semiconductors with precise

controlling of various shapes has been intensively studied over the past decade. CdSe, as one of the II-VI semiconductors, becomes a popular choice for researchers due to its wellstudied synthetic approaches and promising applications in biological labeling and lightemitting diodes. CdSe nanostructures can be synthesized with various morphologies, such as spherical particle, rod, arrow, tear drop and tetrapod. Among these shapes, CdSe nano‐ rods have been proved as good substrates for Au nanoparticles to grow on. Therefore, we selected CdSe nanorods as the host semiconductor to produce CdSe-Au hybrid nanorods in this work.

## **2. Synthesis of CdSe Nanorod**

#### **2.1. Shape Control of CdSe**

The synthesis of II-VI and III-V semiconductors with precise controlling of various shapes has been intensively studied over the past decade. CdSe, as one of the II-VI semiconductors, becomes a popular choice for researchers due to its well-studied synthetic approaches and promising applications in biological labeling and light-emitting diodes. CdSe nanostructures can be synthesized with various morphologies, such as spherical particle, rod, arrow, tear drop and tetrapod. CdSe nanocrystals can be made via the injection of reactant precursors into a hot excess surfactant solvent, such as mixture of trioctylphosphine oxide (TOPO, [CH3(CH2)7]3PO) and stearic acid (CH3(CH2)16COOH), which usually provides a fast nuclea‐ tion of CdSe seeds, followed by slow growth. Two injecting methods are mostly adopted. One is a mono-injection method, from which the spherical CdSe nanoparticles are formed. In this method, the temporal separation between nucleation and growth helps to achieve rel‐ atively narrow size distribution of nanoparticles. The other is a multiple-injection method, from which CdSe nanorods are produced.

In the early study of CdSe synthetic methods, Murray *et al*. provided a simple route to the production of nearly monodispersed CdE (E = S, Se, Te) semiconductor nanoparticles [4]. A pre-heated (~300 °C) surfactant solvent was prepared first followed by co-injection of orga‐ nometallic precursors (cadmium dimethyl, Cd(CH3)2 and Se trioctylphosphine, TOP, [CH3(CH2)7]3P) to produce a homogeneous nucleation using TOPO as a surfactant. The con‐ sequent growth at a moderate temperatures range (230~260 °C) resulted in the formation of uniform CdSe nanoparticles with diameters ranging from 1.5 to 11.5 nm.

Peng et al. also developed Murray's method and first achieved the shape control of CdSe nanostructures from CdSe nanoparticles to CdSe nanorods using Cd(CH3)2 as precursor [5]. However, instead of using pure TOPO, the mixture of hexylphosphonic acid (HPA, CH3(CH2)5P(O)(OH)2) and TOPO was used to serve as surfactant and reaction solvent. The function of HPA was to maintain the growth of CdSe nanorods in a controllable manner. Peng et al. also introduced multiple injections of organometallic precursors in the synthesis, which replenished monomers during the growth period of CdSe and resulted in the aniso‐ tropic growth of CdSe nanorods. By adopting HPA as surfactant and using multiple-injec‐ tion method, CdSe nanorods were successfully produced in their studies. It was indicated that the growth of hexagonal crystal structure (wurtzite) of CdSe was highly anisotropic. Some growth conditions favored crystallization along the <001> axis, usually called the c-ax‐ is of wurtzite structure, and resulted in the formation of the CdSe nanorods. However, the initial injection temperature adopted in this method was in the range from 340 °C to 360 °C. Also, this multiple-injection method is relatively complex to execute compared to the monoinjection method adopted in the synthesis of CdSe nanoparticles.

Shieh et al. further explored Peng's multiple-injection approach in the synthesis of CdSe nanorods using cadmium mono-oxide (CdO) as precursor, and n-tetradecylphosphonic acid (TDPA, CH3(CH2)13P(O)(OH)2) and TOPO as surfactant [6]. They adopted a lower injection temperature (260 °C). However, they had to preheat the Se precursors to 120 °C every time before the injection.

From above discussion we can extract three important factors in the synthesis of CdSe nano‐ crystals:


In this work, we developed a mono-injection method based on traditional approach used in the synthesis of CdSe nanocrystals. We successfully achieved the shape control of CdSe nanostructures from nanoparticles to nanorods by simply adjusting the molar ratios be‐ tween Se and Cd precursors. The growth models were proposed for different morphologies of CdSe and verified by the followed experimental results. The synthesized CdSe nanorods are utilized for the growth of Au nanoparticles which will be interpreted in details in the next section.

#### **2.2. Synthesis of CdSe**

controlling of various shapes has been intensively studied over the past decade. CdSe, as one of the II-VI semiconductors, becomes a popular choice for researchers due to its wellstudied synthetic approaches and promising applications in biological labeling and lightemitting diodes. CdSe nanostructures can be synthesized with various morphologies, such as spherical particle, rod, arrow, tear drop and tetrapod. Among these shapes, CdSe nano‐ rods have been proved as good substrates for Au nanoparticles to grow on. Therefore, we selected CdSe nanorods as the host semiconductor to produce CdSe-Au hybrid nanorods in

The synthesis of II-VI and III-V semiconductors with precise controlling of various shapes has been intensively studied over the past decade. CdSe, as one of the II-VI semiconductors, becomes a popular choice for researchers due to its well-studied synthetic approaches and promising applications in biological labeling and light-emitting diodes. CdSe nanostructures can be synthesized with various morphologies, such as spherical particle, rod, arrow, tear drop and tetrapod. CdSe nanocrystals can be made via the injection of reactant precursors into a hot excess surfactant solvent, such as mixture of trioctylphosphine oxide (TOPO, [CH3(CH2)7]3PO) and stearic acid (CH3(CH2)16COOH), which usually provides a fast nuclea‐ tion of CdSe seeds, followed by slow growth. Two injecting methods are mostly adopted. One is a mono-injection method, from which the spherical CdSe nanoparticles are formed. In this method, the temporal separation between nucleation and growth helps to achieve rel‐ atively narrow size distribution of nanoparticles. The other is a multiple-injection method,

In the early study of CdSe synthetic methods, Murray *et al*. provided a simple route to the production of nearly monodispersed CdE (E = S, Se, Te) semiconductor nanoparticles [4]. A pre-heated (~300 °C) surfactant solvent was prepared first followed by co-injection of orga‐ nometallic precursors (cadmium dimethyl, Cd(CH3)2 and Se trioctylphosphine, TOP, [CH3(CH2)7]3P) to produce a homogeneous nucleation using TOPO as a surfactant. The con‐ sequent growth at a moderate temperatures range (230~260 °C) resulted in the formation of

Peng et al. also developed Murray's method and first achieved the shape control of CdSe nanostructures from CdSe nanoparticles to CdSe nanorods using Cd(CH3)2 as precursor [5]. However, instead of using pure TOPO, the mixture of hexylphosphonic acid (HPA, CH3(CH2)5P(O)(OH)2) and TOPO was used to serve as surfactant and reaction solvent. The function of HPA was to maintain the growth of CdSe nanorods in a controllable manner. Peng et al. also introduced multiple injections of organometallic precursors in the synthesis, which replenished monomers during the growth period of CdSe and resulted in the aniso‐ tropic growth of CdSe nanorods. By adopting HPA as surfactant and using multiple-injec‐ tion method, CdSe nanorods were successfully produced in their studies. It was indicated

uniform CdSe nanoparticles with diameters ranging from 1.5 to 11.5 nm.

this work.

326 Nanowires - Recent Advances

**2. Synthesis of CdSe Nanorod**

from which CdSe nanorods are produced.

**2.1. Shape Control of CdSe**

In this work, we used Cd(Ac)2 as the Cd precursor, a mixture of TOPO, octadecylamine (ODA, CH3(CH2)17NH2) and stearic acid as reaction solvent and surfactants. Se powder was dissolved in TOP to serve as injection solution for the growth of CdSe nanocrystals. The in‐ jection temperature was 310 °C and the growth time period for CdSe nanocrystals was 25 min. In a typical synthesis of CdSe, a mixture of Cd(Ac)2 (90%, Strem Chemicals, 0.1132 g or 0.15 mmol), stearic acid (98%, Alfa Aesar, 0.0853 g or 0.3 mmol), TOPO (98%, Alfa Aesar, 2.5 g or 6.3 mmol) and ODA (90%, Acros, 2.5 g or 8.3 mmol) were added into a 50 ml three-neck round-bottom flask and then heated up via a heating mantle under vigorous stirring and ar‐ gon flow protection. During the heating process, the initial white solid mixture turned into a transparent colorless solution. When the solution temperature reached 310 °C, a Se solution made by dissolving 0.18, 0.25, 0.75 or 1.0 mmol Se powder (99%, Strem Chemicals) into 1.79, 2.02, 5.38 or 7.17 mmol TOP (90%, Alfa Aesar) was quickly injected into the flask (<< 1 s). A dark purple solution formed immediately. After heating for an additional 25 min, the reac‐ tion was stopped by the removal of the heating mantle. After the solution was cooled to around 60 °C, the reaction solution was equally distributed into two 14 ml plastic centrifuga‐ tion tubes. The final precipitation of CdSe was collected by washing it with ethanol and chloroform via a high speed centrifugation process.

**Figure 1.** (A) Isotropic growth of CdSe nanoparticles. (B) Anisotropic growth of CdSe nanorods. (001) is the crystal sur‐ face perpendicular to the c-axis in the CdSe structure.

During the synthesis of CdSe, Cd(Ac)2 first decomposed to form Cd atoms at high tempera‐ tures. These Cd atoms or clusters (aggregation of a few atoms) didn't nucleate before reach‐ ing the supersaturation. The formation of CdSe nanocrystals was triggered by injecting the Se/TOP solution into the flask. After injection, the Cd monomers reacted with Se to form CdSe monomers immediately, judging by the fast color changes upon the injection of Se into reaction solution. Once the concentration of CdSe reached supersaturation level, the CdSe nuclei (seeds) formed in the solution. In the meantime, CdSe monomers continued to depos‐ it on CdSe nuclei, and resulted in the further growth of CdSe nanostructures. The continu‐ ous growth of CdSe nanostructures depleted the CdSe monomers in the solution, which would eventually decrease to the equilibrium concentration (the lowest concentration of CdSe monomer in solution). In the synthesis of CdSe with the molar ratio of Se/Cd=5, both nanorods and nanoparticles were observed. It is known that the formation of CdSe nano‐ rods could be attributed to the elongation of c-axis in CdSe crystal structure. Namely, if the c-axis has a faster growth rate comparing to other axes in CdSe crystal structure, CdSe nano‐ rods will be formed via an anisotropic growth along c-axis; if all the axes have a similar growth rate, CdSe nanoparticles will form via an isotropic growth. The growth rates of dif‐ ferent axes can be also affected by the concentration of monomers in the solution. If the con‐ centration of monomers is very high, all the axes will have a similar growth rate, CdSe nanoparticles will be formed; if the concentration of the monomers is low, CdSe nanorods will be formed via a dominant growth along c-axis. The corresponding scheme for the iso‐ tropic and anisotropic growth of CdSe nanocrystals is shown in Figure 1.

dark purple solution formed immediately. After heating for an additional 25 min, the reac‐ tion was stopped by the removal of the heating mantle. After the solution was cooled to around 60 °C, the reaction solution was equally distributed into two 14 ml plastic centrifuga‐ tion tubes. The final precipitation of CdSe was collected by washing it with ethanol and

**Figure 1.** (A) Isotropic growth of CdSe nanoparticles. (B) Anisotropic growth of CdSe nanorods. (001) is the crystal sur‐

During the synthesis of CdSe, Cd(Ac)2 first decomposed to form Cd atoms at high tempera‐ tures. These Cd atoms or clusters (aggregation of a few atoms) didn't nucleate before reach‐ ing the supersaturation. The formation of CdSe nanocrystals was triggered by injecting the Se/TOP solution into the flask. After injection, the Cd monomers reacted with Se to form CdSe monomers immediately, judging by the fast color changes upon the injection of Se into reaction solution. Once the concentration of CdSe reached supersaturation level, the CdSe nuclei (seeds) formed in the solution. In the meantime, CdSe monomers continued to depos‐ it on CdSe nuclei, and resulted in the further growth of CdSe nanostructures. The continu‐ ous growth of CdSe nanostructures depleted the CdSe monomers in the solution, which would eventually decrease to the equilibrium concentration (the lowest concentration of CdSe monomer in solution). In the synthesis of CdSe with the molar ratio of Se/Cd=5, both nanorods and nanoparticles were observed. It is known that the formation of CdSe nano‐ rods could be attributed to the elongation of c-axis in CdSe crystal structure. Namely, if the c-axis has a faster growth rate comparing to other axes in CdSe crystal structure, CdSe nano‐ rods will be formed via an anisotropic growth along c-axis; if all the axes have a similar growth rate, CdSe nanoparticles will form via an isotropic growth. The growth rates of dif‐ ferent axes can be also affected by the concentration of monomers in the solution. If the con‐ centration of monomers is very high, all the axes will have a similar growth rate, CdSe

chloroform via a high speed centrifugation process.

328 Nanowires - Recent Advances

face perpendicular to the c-axis in the CdSe structure.

**Figure 2.** TEM images of CdSe nanoparticles and the corresponding sizing histogram. These CdSe nanoparticles were synthesized at 310 °C for 25 min using Cd(Ac)2 (0.15 mmol), CH3(CH2)16COOH (0.3 mmol), TOPO (2.5 g), ODA (2.5 g), Se (1 mmol) and TOP (7.17 mmol). The molar ratio of Se/Cd was 6.7. The average diameter of CdSe nanoparticles was 7.7 ± 1.9 nm.

In order to verify the effect of CdSe concentration on the morphologies of CdSe as proposed in Figure 1, a series of experiments was conducted using different Se/Cd molar ratios. To simplify the experimental procedure, we kept the amount of Cd(Ac)2 constant, and only ad‐ justed the amount of Se to change the molar ratios of Se/Cd in the syntheses. If the model we proposed is valid, high amount of Se will lead to the formation of CdSe nanoparticles via the isotropic growth; while fewer amount of Se will lead to the formation of CdSe nanorods by encouraging the growth along c-axis. The TEM images of resulting CdSe nanostructures are shown in Figure 2 and 3. When Se/Cd molar ratio was equal to 6.7, CdSe crystals grow iso‐ tropically during the reaction due to the abundant existence of monomers in the solution (Figure 2). When the Se/Cd molar ratios decreased to 1.7 and 1.2 in the syntheses, the CdSe nanorod systems were synthesized (Figures 3). The observations are well consistent with the growth models we proposed. Moreover, the experimental procedure presenting in this work is more easily to conduct. This method allows the synthesis to conduct at a moderate tem‐ perature. Also by simply changing the molar ratios between Se and Cd, different shaped CdSe nanocrystals can be successfully synthesized, which provides a new thought to control the shape of CdSe nanocrystals.

In summary, a simple mono-injection method was adopted in the synthesis of CdSe nano‐ crystals. CdSe nanoparticles and nanorods were produced by adjusting the molar ratios be‐ tween Se and Cd precursors. We proposed a plausible growth model in which the concentration of monomers in the solution played an important role in determining the final morphologies of CdSe nanocrystals. The model can be briefly described as high concentra‐ tions of CdSe monomer lead to an isotropic growth of CdSe nanoparticles; while low con‐ centrations of CdSe monomer lead to an anisotropic growth of CdSe nanorods. A series of experiments using different Se/Cd molar ratios (Se/Cd = 6.7, 1.7 and 1.2) was carried out to verify this growth model.

**Figure 3.** TEM images of CdSe nanorods and the corresponding sizing histograms. (Left) These CdSe nanorods were synthesized at 310 °C for 25 min using Cd(Ac)2 (0.15 mmol), CH3(CH2)16COOH (0.3 mmol), TOPO (2.5 g), ODA (2.5 g), Se (0.25 mmol) and TOP (2.02 mmol). The molar ratio of Se/Cd was 1.7. The length of the rod was 33.4 ± 13.0 nm, the width was 5.9 ± 0.5 nm. (Right) These CdSe nanorods were synthesized at 310 °C for 25 min using Cd(Ac)2 (0.15 mmol), CH3(CH2)16COOH (0.3 mmol), TOPO (2.5 g), ODA (2.5 g), Se (0.15 mmol) and TOP (1.79 mmol). The molar ratio of Se/Cd was 1.2. The length of the rod was 27.0 ± 14.4 nm, the width was 5.6 ± 0.9 nm.

## **3. Synthesis of CdSe-Au Nanorod**

#### **3.1. Material Synthesis**

The growth of metals on semiconductors as a means of increasing functionalities is an im‐ portant subject in material science. A preparation of such hybrids usually involves the injec‐ tion of a metal salt into the semiconductor nanocrystal solution. Mokari *et al*. first achieved a position-selective growth of Au nanoparticles onto CdSe nanorods by gradually adding gold chloride (AuCl3) solution into CdSe nanorods solution at room temperature in the pres‐ ence of dodecyldimethylammonium bromide (DDAB, [CH3(CH2)11]2N(CH3)2Br,) and dode‐ cylamine (CH3(CH2)11NH2) [7]. They claimed that amine reduced AuCl3 to form CdSe/Au hybrid with Au nanoparticles deposited on the tips of CdSe nanorods. Furthermore, by con‐ trolling the reaction times, the sizes of Au nanoparticles could be tuned from 2 to 4 nm. The most compelling aspect of Mokari's work was the high selective growth onto the rod's tips. They attributed this position-selective growth to higher surface energies of the tips, and more accessible sites due to imperfect passivation of surfactants on the tips of CdSe nano‐ rods. In their follow-up work, Mokari *et al*. further studied the growth process of Au nano‐ particles on CdSe nanorods [8]. At long reaction times, the concentration of AuCl3 in solution dropped to a critical value and the reaction transitioned from a growth state to a ripening state, resulting in the disappearance of small Au nanoparticles along the surface of nanorod and the growth of large Au nanoparticles on the tips of nanorod. From the above discussion we can conclude two important factors to achieve the growth of Au nanoparticles on CdSe nanorods:

tions of CdSe monomer lead to an isotropic growth of CdSe nanoparticles; while low con‐ centrations of CdSe monomer lead to an anisotropic growth of CdSe nanorods. A series of experiments using different Se/Cd molar ratios (Se/Cd = 6.7, 1.7 and 1.2) was carried out to

**Figure 3.** TEM images of CdSe nanorods and the corresponding sizing histograms. (Left) These CdSe nanorods were synthesized at 310 °C for 25 min using Cd(Ac)2 (0.15 mmol), CH3(CH2)16COOH (0.3 mmol), TOPO (2.5 g), ODA (2.5 g), Se (0.25 mmol) and TOP (2.02 mmol). The molar ratio of Se/Cd was 1.7. The length of the rod was 33.4 ± 13.0 nm, the width was 5.9 ± 0.5 nm. (Right) These CdSe nanorods were synthesized at 310 °C for 25 min using Cd(Ac)2 (0.15 mmol), CH3(CH2)16COOH (0.3 mmol), TOPO (2.5 g), ODA (2.5 g), Se (0.15 mmol) and TOP (1.79 mmol). The molar ratio

The growth of metals on semiconductors as a means of increasing functionalities is an im‐ portant subject in material science. A preparation of such hybrids usually involves the injec‐ tion of a metal salt into the semiconductor nanocrystal solution. Mokari *et al*. first achieved a position-selective growth of Au nanoparticles onto CdSe nanorods by gradually adding gold chloride (AuCl3) solution into CdSe nanorods solution at room temperature in the pres‐ ence of dodecyldimethylammonium bromide (DDAB, [CH3(CH2)11]2N(CH3)2Br,) and dode‐ cylamine (CH3(CH2)11NH2) [7]. They claimed that amine reduced AuCl3 to form CdSe/Au hybrid with Au nanoparticles deposited on the tips of CdSe nanorods. Furthermore, by con‐ trolling the reaction times, the sizes of Au nanoparticles could be tuned from 2 to 4 nm. The most compelling aspect of Mokari's work was the high selective growth onto the rod's tips.

of Se/Cd was 1.2. The length of the rod was 27.0 ± 14.4 nm, the width was 5.6 ± 0.9 nm.

**3. Synthesis of CdSe-Au Nanorod**

**3.1. Material Synthesis**

verify this growth model.

330 Nanowires - Recent Advances


Since in the ideal Au-based DMS, the growth of small Au nanoparticles has to achieve a uni‐ form distribution all over the surfaces of CdSe nanorods, not only on some particular sites on CdSe nanorods, an alternative method to achieve the uniform growth of Au nanoparti‐ cles all over the surfaces of CdSe nanorods is required.

The Au precursor stock solution was prepared by dissolving 12 mg (0.04 mmol) of AuCl3 (99.99%, Alfa Aesar), 15 mg (0.05 mmol) of 1-dodecyltrimethylammonium bromide (DTAB, C12H25(CH3)3NBr, 99%, Alfa Aesar) in 2 ml of chloroform at room temperature. The color of the AuCl3 solution changed to yellow from orange after 5 min sonication. 0.15 mmol of assynthesized CdSe nanorods with 30 mg TOPO (C24H51OP, 98%, Alfa Aesar) and 30 mg Octa‐ decylamine (ODA, CH3(CH2)17NH2, 90%, Acros) were dissolved in 12 ml of chloroform in a 50 ml three-neck round-bottom flask under vigorous stirring and argon protection. Au pre‐ cursor stock solution (0.167 ml) containing 1 mg (0.003 mmol) of AuCl3 was injected into the reaction flask within 10 seconds. The reaction was carried out under room temperature. The TEM samples of CdSe-Au hybrid nanorods were prepared by taking 0.3 ml reaction solution from the flask at several time intervals after the initial injection. After 5, 7, 10, 30 or 60 min, the samples were collected into a 14 ml centrifuge tube and centrifuged for 5 min to get the final products.

The galvanic replacement reaction usually involves a redox process. The driving force of this process is the reduction potential (E° ) difference between two reactants, with one reactant acting as the oxidizer and the other reactant as the reducer. Since E° (Au3+/Au0 ) has higher reduction potential than E° (Se2-/Se4+), the following reaction would happen in the solution:

$$2AuCl\_3 + CdSe \rightarrow \ 2Au + Cd^{2+} + Se^{4+} + 6Cl^-$$

After mixing with CdSe, Au3+ ions were reduced by Se2- to form Au atoms at the surfaces of CdSe nanorods. As reaction went on, Au nuclei formed on the surfaces of CdSe nanorods, as shown in Figure 4.

**Figure 4.** The galvanic replacement reactions between CdSe and AuCl3 and the formation of the Au nuclei.

#### **3.2. Time Effect**

In the syntheses of CdSe-Au hybrid nanorods, the reaction times varied from 5 to 60 min to control the sizes of Au nanoparticles (See selected TEM images of CdSe-Au synthesized at 5, 10, and 30 mins in Figure 5). The CdSe nanorods used for reaction were synthesized using a Se/Cd molar ratio of 1.7. The molar numbers of CdSe nanorods and AuCl3 was 0.15 mmol and 0.003 mmol. The growth of Au nanoparticle on CdSe nanorods with respect to reaction time was shown in Figure 6. From the curve we can see that in the first 10 min, the average size of Au nanoparticles increased sharply from 0 to 2.3 ± 0.7 nm; after 10 min reaction, Au nanoparticles exhibited a rather stagnant growth: sizes changed from 2.3 ± 0.7 nm at 10 min to 2.4 ± 0.9 nm at 60 min. The time-dependent growth of Au nanoparticles can be attributed to the change of concentration of AuCl3 in the solution at different reaction times. The for‐ mation of Au nanoparticles was triggered by the injection of AuCl3. At the initial time point (t = 0), there was no Au nanoparticle observed. Once the AuCl3 was injected into the solu‐ tion, the reaction between Au3+ and Se2- occurred subsequently at the surfaces of the CdSe nanorods. This reaction led to the formation of Au atoms on the surfaces of CdSe nanorods. In the first 10 min, the Au nanoparticles kept growing at a relatively rapid growth rate due to the abundant AuCl3 in the solution. As more Au nanoparticles formed in the solution, the concentration of Au3+ decreased until reaching the equilibrium. The growth rate of Au nano‐ particles slowed down and ultimately suppressed by the depletion of the Au3+, and the solu‐ tion was considered to approach the equilibrium state. At an equilibrium state, the growth of Au nanoparticles transitioned to ripening state.

Cadmium Selenide-Gold (CdSe-Au) Hybrid Nanowires: Synthesis and Study of Charge Transfer http://dx.doi.org/10.5772/52828 333

**Figure 4.** The galvanic replacement reactions between CdSe and AuCl3 and the formation of the Au nuclei.

In the syntheses of CdSe-Au hybrid nanorods, the reaction times varied from 5 to 60 min to control the sizes of Au nanoparticles (See selected TEM images of CdSe-Au synthesized at 5, 10, and 30 mins in Figure 5). The CdSe nanorods used for reaction were synthesized using a Se/Cd molar ratio of 1.7. The molar numbers of CdSe nanorods and AuCl3 was 0.15 mmol and 0.003 mmol. The growth of Au nanoparticle on CdSe nanorods with respect to reaction time was shown in Figure 6. From the curve we can see that in the first 10 min, the average size of Au nanoparticles increased sharply from 0 to 2.3 ± 0.7 nm; after 10 min reaction, Au nanoparticles exhibited a rather stagnant growth: sizes changed from 2.3 ± 0.7 nm at 10 min to 2.4 ± 0.9 nm at 60 min. The time-dependent growth of Au nanoparticles can be attributed to the change of concentration of AuCl3 in the solution at different reaction times. The for‐ mation of Au nanoparticles was triggered by the injection of AuCl3. At the initial time point (t = 0), there was no Au nanoparticle observed. Once the AuCl3 was injected into the solu‐ tion, the reaction between Au3+ and Se2- occurred subsequently at the surfaces of the CdSe nanorods. This reaction led to the formation of Au atoms on the surfaces of CdSe nanorods. In the first 10 min, the Au nanoparticles kept growing at a relatively rapid growth rate due to the abundant AuCl3 in the solution. As more Au nanoparticles formed in the solution, the concentration of Au3+ decreased until reaching the equilibrium. The growth rate of Au nano‐ particles slowed down and ultimately suppressed by the depletion of the Au3+, and the solu‐ tion was considered to approach the equilibrium state. At an equilibrium state, the growth

**3.2. Time Effect**

332 Nanowires - Recent Advances

of Au nanoparticles transitioned to ripening state.

**Figure 5.** Original and enlarged TEM images of CdSe-Au hybrid nanorods and Au sizing histogram after 5, 10 and 30 min reactions. The average diameters of Au nanoparticles were 1.9 ± 0.8 nm, 2.3 ± 0.7 nm and 2.3 ± 1.3 nm, respectively.

**Figure 6.** Changes of Au nanoparticles sizes as a function of reaction time

#### **3.3. The Ostwald Ripening Process in CdSe-Au Solution**

The Ostwald ripening process is the growth of large nanoparticles at the expense of disso‐ lution of small ones at an equilibrium state. It is a spontaneous process because large nano‐ particles are more thermodynamically stable than small ones. After the solution reached equilibrium state probably after 10 minutes reactions, large Au nanoparticles kept grow‐ ing while small ones disappeared. Mokari1 *et al*. first proposed a growth mechanism for this process that small Au nanoparticles first released electrons into solution and then trans‐ ferred these electrons to large Au nanoparticles through CdSe nanorods [8]. After giving out electrons, the small Au nanoparticles escaped from the surfaces of nanorods and be‐ came free Au ions in the solution. These Au ions re-deposited on large Au nanoparticles and were reduced to Au0 (Figure 7.). The result of this process was disappearance of small Au nanoparticles and growth of large ones. However, they did not provide direct evi‐ dence to prove that small Au particles can release the electrons automatically. Also, no observation of electric current in CdSe nanorod was reported. Here, we proposed an alter‐ native growth mechanism in which the growth of the Au nanoparticles is attributed to the Ostwald ripening process.

From a standpoint of kinetics, when the concentration of AuCl3 was high, formation of small Au nanoparticles was dominant. When Au atoms/clusters monomers were depleted, the whole system transited from a fast growth state (kinetics state) to a ripening state (thermo‐ dynamics state). At the ripening state, from a standpoint of thermodynamics, small Au nanoparticles on the surfaces of CdSe nanorods dissolved into solution in the form of Au atoms (monomers) and re-deposited on the surfaces of large Au nanoparticles. As a result of this ripening process, the small Au nanoparticles dissolved to favor the growth of large par‐ ticles on the CdSe nanorods. Figure 8 shows the Ostwald ripening process in CdSe-Au hy‐ brid nanorods solution at equilibrium state. As the reaction continued, we found that Au nanoparticles tend to migrate to the tips of CdSe nanorods. After 5 minutes reactions, (Fig‐ ure 6), small Au nanoparticles were found all over the surfaces of CdSe nanorods. After 30 and 60 minutes reactions, Au nanoparticles were found mostly at the tips of CdSe nanorods. This migration of Au nanoparticles can be attributed to higher surface energies on the tips of nanorod, resulting from more accessible sites on the tips where surfactant capping was weak. Therefore, the tips of nanorods are preferential growing sites for Au nanoparticles at long reaction times.

**Figure 6.** Changes of Au nanoparticles sizes as a function of reaction time

**3.3. The Ostwald Ripening Process in CdSe-Au Solution**

Ostwald ripening process.

334 Nanowires - Recent Advances

The Ostwald ripening process is the growth of large nanoparticles at the expense of disso‐ lution of small ones at an equilibrium state. It is a spontaneous process because large nano‐ particles are more thermodynamically stable than small ones. After the solution reached equilibrium state probably after 10 minutes reactions, large Au nanoparticles kept grow‐ ing while small ones disappeared. Mokari1 *et al*. first proposed a growth mechanism for this process that small Au nanoparticles first released electrons into solution and then trans‐ ferred these electrons to large Au nanoparticles through CdSe nanorods [8]. After giving out electrons, the small Au nanoparticles escaped from the surfaces of nanorods and be‐ came free Au ions in the solution. These Au ions re-deposited on large Au nanoparticles and were reduced to Au0 (Figure 7.). The result of this process was disappearance of small Au nanoparticles and growth of large ones. However, they did not provide direct evi‐ dence to prove that small Au particles can release the electrons automatically. Also, no observation of electric current in CdSe nanorod was reported. Here, we proposed an alter‐ native growth mechanism in which the growth of the Au nanoparticles is attributed to the

From a standpoint of kinetics, when the concentration of AuCl3 was high, formation of small Au nanoparticles was dominant. When Au atoms/clusters monomers were depleted, the whole system transited from a fast growth state (kinetics state) to a ripening state (thermo‐

**Figure 7.** The growth process proposed by Mokari1 et al. Small Au nanoparticles escape to the solution as free Au ions by release their electrons. The released electrons transfer to large Au nanoparticles to reduce Au ions in the solution and results in the growth of large Au nanoparticles.

**Figure 8.** The Ostwald ripening process in CdSe-Au solution. (A) dissolution of small Au particles; (B) growth of large Au particles; (C) CdSe-Au hybrid nanorods with large Au particles.

#### **3.4. In-Situ Growth of Au Nanoparticles under Electron Irradiation in TEM**

A unique growth of Au nanoparticles was observed under the irradiation of the electron beams in TEM. Figure 9 shows this process. Figure 9A is the original image of CdSe-Au nanorods after 7 min reactions exposed under electron beam of low intensity, Figure 9B is the same sample exposed under electron beam of high intensity for 1 min. From Figure 9A to B we can see that as the electron beam intensity increased, the average size of Au nano‐ particles experienced a clear growth from the initial 2.3 ± 0.7 nm to the final 5.0 ± 1.8 nm. This unique growth of Au nanoparticles under electron irradiation can be attributed to the size effect on the melting point of Au nanoparticles. It has been approved that the melting points of nanoparticles have great dependencies to their sizes. Typically, particles exhibit decreasing melting points with decreasing sizes. This size-dependent melting point of nano‐ particles can be calculated through the Gibbs-Thomson equation:

$$T\_{M}
\left(d
\right) = T\_{MB}
\left(1
\frac{4
\sigma\_{sl}}{
\rho\_{s}dH\_{sf}}
\right)$$

where *T MB* is bulk melting temperature, *σ sl* is solid liquid interface energy, *H <sup>f</sup>* is bulk heat of fusion, *ρ <sup>s</sup>* is density of solids and *d* is particle diameter. From this equation we can see that as particle diameter *d* increases, the particle melting temperature *T <sup>M</sup> (d)* decreases. The melt‐ ing point of bulk Au metal is 1336 k (1063 °C). If the size of Au nanoparticle is 1.6 nm (the smallest Au nanoparticles we observed), the corresponding melting point is only 554 K (281 °C) (The parameters for calculation are in Table 1). Also, the operation of TEM is under vac‐ uum, which further decreases the melting points of small Au nanoparticles. When Au nano‐ particles were exposed to the electron beam in TEM, small Au nanoparticles melted easily due to their lower melting points. In the meantime, these melted Au particles were migrat‐ ing towards large Au nanoparticles in order to achieve lower surface energies thermody‐ namically. The result of the process was the melting of small particles and the growth of large ones. The observed phenomenon also indicated that small Au nanoparticles distribut‐ ed all over the surface of nanorod in the original sample. Otherwise, there were not enough growth resources for this dramatic growth of Au nanoparticles.


**Table 1.** Parameters for Calculating the Melting Point of Au Nanoparticle

 

**3.4. In-Situ Growth of Au Nanoparticles under Electron Irradiation in TEM**

particles can be calculated through the Gibbs-Thomson equation:

growth resources for this dramatic growth of Au nanoparticles.

**Table 1.** Parameters for Calculating the Melting Point of Au Nanoparticle

 

where *T MB* is bulk melting temperature, *σ sl* is solid liquid interface energy, *H <sup>f</sup>*

*TM* (*d*) <sup>=</sup>*TMB*(1 - <sup>4</sup>*σsl*

336 Nanowires - Recent Advances

A unique growth of Au nanoparticles was observed under the irradiation of the electron beams in TEM. Figure 9 shows this process. Figure 9A is the original image of CdSe-Au nanorods after 7 min reactions exposed under electron beam of low intensity, Figure 9B is the same sample exposed under electron beam of high intensity for 1 min. From Figure 9A to B we can see that as the electron beam intensity increased, the average size of Au nano‐ particles experienced a clear growth from the initial 2.3 ± 0.7 nm to the final 5.0 ± 1.8 nm. This unique growth of Au nanoparticles under electron irradiation can be attributed to the size effect on the melting point of Au nanoparticles. It has been approved that the melting points of nanoparticles have great dependencies to their sizes. Typically, particles exhibit decreasing melting points with decreasing sizes. This size-dependent melting point of nano‐

> *ρsdH <sup>f</sup>* )

fusion, *ρ <sup>s</sup>* is density of solids and *d* is particle diameter. From this equation we can see that as particle diameter *d* increases, the particle melting temperature *T <sup>M</sup> (d)* decreases. The melt‐ ing point of bulk Au metal is 1336 k (1063 °C). If the size of Au nanoparticle is 1.6 nm (the smallest Au nanoparticles we observed), the corresponding melting point is only 554 K (281 °C) (The parameters for calculation are in Table 1). Also, the operation of TEM is under vac‐ uum, which further decreases the melting points of small Au nanoparticles. When Au nano‐ particles were exposed to the electron beam in TEM, small Au nanoparticles melted easily due to their lower melting points. In the meantime, these melted Au particles were migrat‐ ing towards large Au nanoparticles in order to achieve lower surface energies thermody‐ namically. The result of the process was the melting of small particles and the growth of large ones. The observed phenomenon also indicated that small Au nanoparticles distribut‐ ed all over the surface of nanorod in the original sample. Otherwise, there were not enough

*d***/m σ***sl* **/J m-2** *Hf***/J Kg-1 ρ***s***/Kg m-3** *TMB***/K** *TM(d)***/K** *TM(d)***/ °C** 1.6\*10-9 0.27 6.27\*104 1.84\*104 1336 554 281

is bulk heat of

**Figure 9.** TEM images of CdSe-Au synthesized using 0.003 mmol AuCl3 into 0.15 mmol CdSe nanorods: (A) before irra‐ diation (B) after irradiation. The diameters of Au nanoparticles were (A) 2.3 ± 0.7 nm and (B) 5.0 ± 1.8 nm.

## **4. Electronic Structure Characterized by X–ray Absorption Near Edge Spectroscopy (XANES)**

XANES is a powerful X–ray absorption spectroscopy technique for investigating charger transfer between distinctive elements as well as oxidation states of specific element within the nanomaterials [9]. By analyzing absorption signals in XANES, one can extract abundant chemical information of the studied material, like formal valence, coordination environ‐ ment, the unoccupied band structure of a material, and fingerprint a mixture of sites/ compounds (e.g. oxidation state determination).

Since XANES is sensitive to the local electronic structure, it has been used to study the elec‐ tronic redistribution or charge transfer upon alloying between various compounds. Obvi‐ ously, it would also be the desirable technique to study the electronic structure of those CdSe/Au hybrid nanowires. Despite the fact that Au metal with filled d10 orbital usually doesn't exhibit an intense white line (the first feature in the absorption edge) at L3 edge thresholds compared with unfilled d9 metals (e.g. Pt), the orbital s–p–d hybridization results in a 5d10–x6(sp)1+x electron configuration, leading to a potentially noticeable increase of white line intensity. The changes of the number of holes in the d band reflect charge transfer be‐ tween various CdSe-Au hybrid nanostructures with different sizes of Au components. Therefore, enhancement of the white line intensity from Au L3 edges can thus be used as a sensor, to detect and quantify the hybrid formation. Figure 10 shows normalized Au L3 edge XANES spectra of CdSe-Au hybrids and that of Au foil for reference. It can be seen that the strength of the Au white line feature in L3 edge is noticeably lower than that of pure Au foil as the size of Au components decreased. Since white line features in Au L3 edge arise from 2p3/2 to 5d5/2 dipole transitions, the observations indicate a decrease in the number of unoc‐ cupied states of d character at the Au site upon the decreasing the size of Au components. The data can be further interpreted in terms of gain of d charge in Au site. These findings dramatically highlight the intriguing structure–property relationship of nanoscaled electron‐ ic materials, and deserve further investigations.

**Figure 10.** Normalized XANES spectra of Au in the CdSe-Au hybrid nanorods.

#### **5. Conclusion**

The target of this work was to synthesize CdSe-Au hybrid nanorods. The first step was to synthesize CdSe nanorods. A simple mono-injection method was adopted in the syntheses. By reducing the molar ratios between Se and Cd, we have successfully achieved the shape control of CdSe nanocrystals from nanoparticles to nanorods. We found 5 to1 was a critical molar ratio between Se and Cd to form different CdSe nanostructures. When the molar ratio between Se and Cd was higher than 5 (6.7), CdSe nanoparticles were synthesized. When the molar ratio was lower than 5 (1.2 and 1.7), CdSe nanorods were synthesized. The growth models for different CdSe nanostructures were proposed and validated by the subsequent experimental results. In the model, the concentration of monomer plays an important role in the shape control of CdSe nanostructures. High concentration of monomer leads to the iso‐ tropic growth of CdSe nanoparticles. Low concentration of monomer leads to the anisotrop‐ ic growth of CdSe nanorods. Comparing with traditional approach adopted in the shape control of CdSe nanostructures, the synthesis approach used in this work is more easy to conduct; also a relatively moderate reaction temperature and more favorable Cd precursor make the synthesis approach safer.

The growth of Au nanoparticles on CdSe nanorods involves galvanic replacement reactions between AuCl3 and CdSe nanorods at room temperature. In the reaction, Au3+ was reduced to Au atoms while Se2- was oxidized to Se4+. By adjusting the reaction times, Au nanoparti‐ cles with diameter from 1.7 ± 0.6 nm to 2.4 ± 0.9 nm could be synthesized on the surfaces of CdSe nanorods. Before the reaction closing to equilibrium state, Au nanoparticles exhibited a fast growth rate due to a high concentration of AuCl3 in solution. When the concentration of AuCl3 approaching to equilibrium level, the reaction was closely related to the Ostwald ripening process in which the small Au nanoparticles dissolved into solution and the mass was transferred to increase the size of the large nanoparticle on the nanorods.

## **Acknowledgements**

the nanomaterials [9]. By analyzing absorption signals in XANES, one can extract abundant chemical information of the studied material, like formal valence, coordination environ‐ ment, the unoccupied band structure of a material, and fingerprint a mixture of sites/

Since XANES is sensitive to the local electronic structure, it has been used to study the elec‐ tronic redistribution or charge transfer upon alloying between various compounds. Obvi‐ ously, it would also be the desirable technique to study the electronic structure of those CdSe/Au hybrid nanowires. Despite the fact that Au metal with filled d10 orbital usually doesn't exhibit an intense white line (the first feature in the absorption edge) at L3 edge thresholds compared with unfilled d9 metals (e.g. Pt), the orbital s–p–d hybridization results in a 5d10–x6(sp)1+x electron configuration, leading to a potentially noticeable increase of white line intensity. The changes of the number of holes in the d band reflect charge transfer be‐ tween various CdSe-Au hybrid nanostructures with different sizes of Au components. Therefore, enhancement of the white line intensity from Au L3 edges can thus be used as a sensor, to detect and quantify the hybrid formation. Figure 10 shows normalized Au L3 edge XANES spectra of CdSe-Au hybrids and that of Au foil for reference. It can be seen that the strength of the Au white line feature in L3 edge is noticeably lower than that of pure Au foil as the size of Au components decreased. Since white line features in Au L3 edge arise from 2p3/2 to 5d5/2 dipole transitions, the observations indicate a decrease in the number of unoc‐ cupied states of d character at the Au site upon the decreasing the size of Au components. The data can be further interpreted in terms of gain of d charge in Au site. These findings dramatically highlight the intriguing structure–property relationship of nanoscaled electron‐

compounds (e.g. oxidation state determination).

338 Nanowires - Recent Advances

ic materials, and deserve further investigations.

**Figure 10.** Normalized XANES spectra of Au in the CdSe-Au hybrid nanorods.

The target of this work was to synthesize CdSe-Au hybrid nanorods. The first step was to synthesize CdSe nanorods. A simple mono-injection method was adopted in the syntheses.

**5. Conclusion**

This work is supported by the University of New Hampshire. Use of the National Synchro‐ tron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract DE-AC02-98CH10886 and beam lines X19A/X18B are partly supported by Synchrotron Catalysis Consortium under contract DE-FG02-05ER15688.

## **Author details**

Xiaowei Teng\* and Xiaoxiong Jia

\*Address all correspondence to: xw.teng@unh.edu

University of New Hampshire, USA

### **References**


## **A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric Nanowire MOSFETs**

Daniela Munteanu and Jean-Luc Autran

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52587

## **1. Introduction**

**References**

340 Nanowires - Recent Advances

[1] Saunders, A. E., Popov, I., & Banin, U. (2006). Synthesis of Hybrid CdS-Au Colloidal

[2] Crespo, P., Litran, R., Multigner, M., de la Fuente, J. M., Sanchez, J. C., Lopez, M. A., Garcia, C., Lopez, Cartes. A., Hernando, S., & Penades, A. Fernandez. (2004). Perma‐ nent magnetism, magnetic anisotropy, and hysteresis of thiol-capped gold nanoparti‐

[3] Zhong, C. J., & Maye, M. M. (2001). Core-shell assembled nanoparticles as catalysts.

[4] Murray, C. B., Noms, D. J., & Bawendi, M. G. (1993). Synthesis and characterization of nearly monodispersed CdE (E = S, Se, Te) semiconductor Nano crystallites. *J. Am.*

[5] Peng, X. G., Manna, L., Yang, W., Wickham, J., Escher, E., Kadavanich, A., & Alivisa‐

[6] Shieh, F., Saunders, A. E., & Korgel, B. A. (2005). General shape control of colloidal CdS, CdSe, CdTe quantum rods and quantum rod heterostructures,. *Phys. Rev. B.*

[7] Mokari, T., Rothenberg, E., Popov, I., Costi, R., & Banin, U. (2004). Selective Growth of Metal Tips onto Semiconductor Quantum Rods and Tetrapods,. Science. , 304,

[8] Mokari, T., Sztrum, C. G., Salant, A., Rabani, E., & Uri, Banin. (2005). Formation of asymmetric one-sided metal-tipped semiconductor nanocrystal dots and rods. *Nature*

[9] Frenkel, A. (2007). Solving the 3D structure of metal nanoparticles. *Zeitschrift Fur*

tos, A. P. (2000). Shape control of CdSe nanocrystals. *Nature*, 404, 59-61.

Nanostructures,. *J. Phys. Chem. B*, 110, 25421-25429.

cles,. Phys. ReV. Lett. , 93, 087204.

*Advanced Materials*, 13, 1507-1511.

*Chem. Soc.*, 115, 8706-8715.

*Lett*, 109, 8538-8542.

1787-1789.

*Mater*, 4, 855-863.

*Kristallographie*, 222, 605-611.

The phenomenal success of CMOS technology, and, then the progress of the information technology, can be attributed without any doubt to the scaling of the MOS transistor, which has been pushed during more than thirty years to increasingly levels of integration and per‐ formances. Then, MOSFETs have been fabricated always smaller, denser, faster and cheaper in order to provide ever more powerful products for digital electronics. Recently, the scaling rate has accelerated, and the MOSFET gate length is now less than 40 nm, with devices en‐ tering into the nanometer world [1]-[2]. The so-called "bulk" MOSFET is the basic and his‐ torical key-device of microelectronics: its dimensions have been reduced more than ~103 times during the three past decades. However, the bulk MOSFET scaling has recently en‐ countered significant limitations, mainly related to the gate oxide (SiO2) leakage currents [3]- [4], the large increase of parasitic short channel effects and the dramatic mobility reduction [5]-[6] due to highly doped Silicon substrates precisely used to reduce these short channel effects. Technological solutions have been proposed in order to continue to use the "bulk solution" until the 45 nm ITRS node. Most of these solutions envisage the introduction of high-permittivity gate dielectric stacks (to reduce the gate leakage, [4], [7]-[8]), midgap metal gate (to suppress the Silicon gate polydepletion-induced parasitic capacitances) and strained Silicon channel (to increase carrier mobility, [9]). However, in parallel to these efforts, alter‐ native solutions to replace the conventional bulk MOSFET architecture have been proposed and studied in the recent literature. These options are numerous and can be classified in general according to three main directions: (i) the use of new materials in the continuity of the "bulk solution", allowing increasingly MOSFET performances due to their dielectric properties (permittivity), electrostatic immunity (SOI materials), mechanical (strain), or

© 2012 Munteanu and Autran; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Munteanu and Autran; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

transport (mobility) properties; (ii) the complete change of the device architecture (e.g. mul‐ tiple-gate devices, Silicon nanowires MOSFET) allowing better electrostatic control, and, as a result, intrinsic channels with higher mobilities and currents; (iii) the exploitation of certain new physical phenomena that appear at the nanometer scale, such as quantum transport, substrate orientation or modification of the material band structure in devices/wires with nanometer dimensions [2], [10].

Nanowire MOSFETs with completely surrounding gate are presently considered as one of the possible solutions for replacing bulk (single-gate) devices and continuing MOSFET scal‐ ing in the nanometer scale [11]-[12]. The main advantage of this architecture is to offer a re‐ inforced electrostatic coupling between the conduction channel and the gate electrode, which considerably reduces short-channel effects (SCE) compared to conventional bulk de‐ vices [11]-[13]. Then, the constraints on channel doping levels can be relaxed and nanowire MOSFET devices can be designed with intrinsic (pure Si) channels. This offers considerable advantages, especially in terms of mobility and elimination of doping fluctuations. The in‐ trinsic channel can also be beneficial with regard to source-to-drain transport due to the high probability of ballistic transport.

Nanowire MOSFETs are generally designed with very thin silicon films in order to reinforce the electrostatic control of the gate over the channel. The ultra-thin silicon film creates a suffi‐ ciently narrow rectangular potential well for inducing the quantization of carrier energy. Car‐ riers are then confined in a rectangular quantum well having feature size close to the electron wavelength. This gives rise to a splitting of the energy levels into subbands (two-dimensional (2-D) density-of-states) [14]-[15], such that the lowest of the allowed energy levels for elec‐ trons (resp. for holes) in the well does not coincide with the bottom of the conduction band (resp. the top of the valence band). In nanowire devices the carrier energy is quantified in the two directions perpendicular to the source-to-drain axis (leading to one-dimensional (1-D) density-of-states). The total density of states in a 1-D system is less than that in a three-dimen‐ sional (3-D) (or classical) system, especially for low energies. Carriers occupying the lowest en‐ ergy levels behave like quantized carriers while those lying at higher energies, which are not as tightly confined in the potential well, can behave like classical (3-D) particles with three de‐ grees of freedom. As the surface electric field increases, the system becomes more quantized and more and more carriers become confined in the potential well. The quantum mechanical confinement considerably modifies the carrier distribution in the channel: the maximum of the inversion charge is shifted away from the interface into the Silicon film. Because of the smaller density of states in the 1-D system, the total population of carriers is smaller for the same Fer‐ mi level than in the corresponding 3-D (or classical) case. This phenomenon affect the net sheet charge of carriers in the inversion layer, thus requiring a larger gate voltage in order to popu‐ late a 1-D inversion layer to have the same number of carriers as the corresponding 3-D sys‐ tem. This leads to an increase of the threshold voltage of the MOSFET, which is an important issue, especially as the power supply voltages drop to lower levels. The gate capacitance and carrier mobility are also modified by quantum effects. These considerations indicate that the wave nature of carriers can no longer be neglected in very thin nanowire MOSFETs and has to be considered in simulation studies.

Modeling and simulation of nanowire MOSFETs devices are currently experiencing a grow‐ ing interest due to unique capabilities: (i) simulation provides useful insights into device op‐ eration since all internal physical quantities that cannot be measured on real devices are available as outputs in simulation; (ii) the predictive capability of simulation studies makes possible the reduction of systematical experimental investigation of these new ultra-scaled devices; (iii) simulation offers the possibility to test hypothetical devices which have not yet been manufactured. Since computers are today considerably cheaper resources, simulation is becoming an indispensable tool for the device engineer, not only for the device optimiza‐ tion, but also for specific studies such as the investigation of new physical phenomena spe‐ cific to the ultra-short channels (quantum confinement of carriers or short-channel electrostatic effects).

transport (mobility) properties; (ii) the complete change of the device architecture (e.g. mul‐ tiple-gate devices, Silicon nanowires MOSFET) allowing better electrostatic control, and, as a result, intrinsic channels with higher mobilities and currents; (iii) the exploitation of certain new physical phenomena that appear at the nanometer scale, such as quantum transport, substrate orientation or modification of the material band structure in devices/wires with

Nanowire MOSFETs with completely surrounding gate are presently considered as one of the possible solutions for replacing bulk (single-gate) devices and continuing MOSFET scal‐ ing in the nanometer scale [11]-[12]. The main advantage of this architecture is to offer a re‐ inforced electrostatic coupling between the conduction channel and the gate electrode, which considerably reduces short-channel effects (SCE) compared to conventional bulk de‐ vices [11]-[13]. Then, the constraints on channel doping levels can be relaxed and nanowire MOSFET devices can be designed with intrinsic (pure Si) channels. This offers considerable advantages, especially in terms of mobility and elimination of doping fluctuations. The in‐ trinsic channel can also be beneficial with regard to source-to-drain transport due to the

Nanowire MOSFETs are generally designed with very thin silicon films in order to reinforce the electrostatic control of the gate over the channel. The ultra-thin silicon film creates a suffi‐ ciently narrow rectangular potential well for inducing the quantization of carrier energy. Car‐ riers are then confined in a rectangular quantum well having feature size close to the electron wavelength. This gives rise to a splitting of the energy levels into subbands (two-dimensional (2-D) density-of-states) [14]-[15], such that the lowest of the allowed energy levels for elec‐ trons (resp. for holes) in the well does not coincide with the bottom of the conduction band (resp. the top of the valence band). In nanowire devices the carrier energy is quantified in the two directions perpendicular to the source-to-drain axis (leading to one-dimensional (1-D) density-of-states). The total density of states in a 1-D system is less than that in a three-dimen‐ sional (3-D) (or classical) system, especially for low energies. Carriers occupying the lowest en‐ ergy levels behave like quantized carriers while those lying at higher energies, which are not as tightly confined in the potential well, can behave like classical (3-D) particles with three de‐ grees of freedom. As the surface electric field increases, the system becomes more quantized and more and more carriers become confined in the potential well. The quantum mechanical confinement considerably modifies the carrier distribution in the channel: the maximum of the inversion charge is shifted away from the interface into the Silicon film. Because of the smaller density of states in the 1-D system, the total population of carriers is smaller for the same Fer‐ mi level than in the corresponding 3-D (or classical) case. This phenomenon affect the net sheet charge of carriers in the inversion layer, thus requiring a larger gate voltage in order to popu‐ late a 1-D inversion layer to have the same number of carriers as the corresponding 3-D sys‐ tem. This leads to an increase of the threshold voltage of the MOSFET, which is an important issue, especially as the power supply voltages drop to lower levels. The gate capacitance and carrier mobility are also modified by quantum effects. These considerations indicate that the wave nature of carriers can no longer be neglected in very thin nanowire MOSFETs and has to

nanometer dimensions [2], [10].

342 Nanowires - Recent Advances

high probability of ballistic transport.

be considered in simulation studies.

This chapter presents a simulation study of electrostatics and electronic transport in radial‐ ly-symmetric nanowire MOSFETs by quantum drift-diffusion numerical simulation. In this chapter, we will investigate the operation of radially-symmetric nanowires using a 2-D/3-D Schrödinger/Poisson solver, called Cylmos. Various methods have been suggested to model quantum confinement effects. Among the approaches that are compatible with classical de‐ vice simulators based on the drift-diffusion approach, the physically most accurate method is to include the Schrödinger equation into the self-consistent computation of the device characteristics [15]. Then, the solver described in this work provides self-consistent solutions of the 2-D Schrödinger equation and the 3-D Poisson equation in cylindrical coordinates [16]-[18], coupled with the drift-diffusion transport equation; this approach is commonly called in literature "quantum drift-diffusion". Simulated drain current versus gate voltage characteristics have been compared to data obtained from simulation with commercial code with an excellent agreement. Cylmos also provides a lot of additional information and val‐ uable physical insights (such as the 3-D profile of electrostatic potential, classical and quan‐ tum carrier densities in the channel, the energy levels and total inversion charge) used to investigate the influence of short-channel and quantum-mechanical effects. Finally, the quantum drift-diffusion code will be used to analyze electrical parameters such as threshold voltage, drain-induced barrier lowering (DIBL) effect and subthreshold swing in radiallysymmetric nanowire MOSFETs. The difference between classical (i.e., without quantum con‐ finement) and quantum threshold voltage, as function of different film silicon thicknesses in nanowire MOSFETs will be also discussed.

The chapter is organized as follows. In section 2 we describe in detail the theoretical back‐ ground of our approach and the simulation code. After the description of the simulated de‐ vices, the Schrödinger, Poisson and current continuity solvers are presented. The complete discretization of these equations and the calculation algorithm are explicitly presented. In Section 3 we compare Cylmos to results obtained using a commercial simulator, particularly in terms of output parameters of the Schrödinger module and drain current. Finally, after this validation step, we use Cylmos to analyze and discuss the operation and performance of circular nanowire-based MOSFETs, including the off-state current, threshold voltage and short channel effects.

## **2. Theoretical background and description of the simulation code**

Cylmos is based on the numerical solving of the Poisson-Schrödinger system coupled with the drift-diffusion equation. The Poisson equation is solved on the entire 3-D structure. To solve the Schrödinger equation, the device is divided into parallel vertical slices (y-z plane) (one slice per mesh point in the x direction). The 2-D Schrödinger equation is solved in each slice to ob‐ tain the wave functions, the quantum energy levels and the charge density. The solution of the Schrödinger equation in 2-D Cartesian coordinates for the nanowires with square or circular section is straightforward to implement numerically, but it requires very extensive computa‐ tion time [19]-[22]. To simplify the calculations, in radially-symmetric nanowires it is possible to reduce the size of the two Poisson and Schrödinger equations. By expressing these equa‐ tions in cylindrical coordinates and using the property of cylindrical symmetry of the struc‐ ture (which implies a symmetry of the potential and wave function), the Poisson equation becomes a 2 dimensional equation and the Schrödinger equation becomes one dimensional. The Cartesian coordinates (x, y, z) are converted in cylindrical coordinates (x, r, θ), as shown in Fig.1, and considering these later, the structure is symmetrical with respect to the coordinate θ. Therefore, the coordinate θ will be ignored and the Poisson equation is solved in two dimen‐ sions, r and x, on the 2-D mesh shown in Fig. 2(a). In the same way, the circular symmetry al‐ lows us to simplify the Schrödinger equation, which will be solved on a 1-D mesh along the radial direction r, as shown in Fig. 2(b). The Schrödinger equation will be solved along vertical cut-lines, perpendicular to the x axis, in each mesh-point i of the x-axis. The advantage of this transformation is that, contrary to the 3-D-Poisson/2-D-Schrödinger equations, the 2-D-Pois‐ son/1-D-Schrödinger system is numerically less CPU consuming. Nevertheless, this system of equations is more difficult to implement due to the form of the Laplacian operator [16]. In the following, we detail the equations for a n-channel fully depleted nanowire MOSFET, but simi‐ lar equations can be derived for p-channel structures.

**Figure 1.** Schematic representation of the radially-symmetric nanowire MOS transistor simulated in this work. The main geometrical parameters are also defined.

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric Nanowire MOSFETs http://dx.doi.org/10.5772/52587 345

**2. Theoretical background and description of the simulation code**

lar equations can be derived for p-channel structures.

Gate

D

main geometrical parameters are also defined.

Metal Oxide

344 Nanowires - Recent Advances

Silicon

Source

L

Drain

**Figure 1.** Schematic representation of the radially-symmetric nanowire MOS transistor simulated in this work. The

x y

z r q

Cylmos is based on the numerical solving of the Poisson-Schrödinger system coupled with the drift-diffusion equation. The Poisson equation is solved on the entire 3-D structure. To solve the Schrödinger equation, the device is divided into parallel vertical slices (y-z plane) (one slice per mesh point in the x direction). The 2-D Schrödinger equation is solved in each slice to ob‐ tain the wave functions, the quantum energy levels and the charge density. The solution of the Schrödinger equation in 2-D Cartesian coordinates for the nanowires with square or circular section is straightforward to implement numerically, but it requires very extensive computa‐ tion time [19]-[22]. To simplify the calculations, in radially-symmetric nanowires it is possible to reduce the size of the two Poisson and Schrödinger equations. By expressing these equa‐ tions in cylindrical coordinates and using the property of cylindrical symmetry of the struc‐ ture (which implies a symmetry of the potential and wave function), the Poisson equation becomes a 2 dimensional equation and the Schrödinger equation becomes one dimensional. The Cartesian coordinates (x, y, z) are converted in cylindrical coordinates (x, r, θ), as shown in Fig.1, and considering these later, the structure is symmetrical with respect to the coordinate θ. Therefore, the coordinate θ will be ignored and the Poisson equation is solved in two dimen‐ sions, r and x, on the 2-D mesh shown in Fig. 2(a). In the same way, the circular symmetry al‐ lows us to simplify the Schrödinger equation, which will be solved on a 1-D mesh along the radial direction r, as shown in Fig. 2(b). The Schrödinger equation will be solved along vertical cut-lines, perpendicular to the x axis, in each mesh-point i of the x-axis. The advantage of this transformation is that, contrary to the 3-D-Poisson/2-D-Schrödinger equations, the 2-D-Pois‐ son/1-D-Schrödinger system is numerically less CPU consuming. Nevertheless, this system of equations is more difficult to implement due to the form of the Laplacian operator [16]. In the following, we detail the equations for a n-channel fully depleted nanowire MOSFET, but simi‐

**Figure 2.** Schematics of the mesh used in Cylmos code and definition of mesh points in x-direction and r-direction.

The physical description of electrostatics of the nanowire device requires solving the coupled Poisson equation with the Schrödinger equation. The general Poisson equation is given by:

$$\nabla \left( \varepsilon \nabla V \right) = -\mathfrak{p} \tag{1}$$

where *ρ* is the charge density, *V*is the electrostatic potential and ε is the material-dependent permittivity. This is a 3-D equation which can be expressed either in Cartesian coordinates or in cylindrical coordinates. In this work we will use the cylindrical coordinates, as it will be shown in the following. The general independent of time Schrödinger equation, ex‐ pressed in terms of effective mass, is given by:

$$\left\{-\frac{\hbar^2}{2}\left(\nabla\_\mu \cdot \frac{\nabla\_\mu}{\hat{m}^\*}\right) + \mathcal{U}(\mu)\right\}\Psi = E\Psi\tag{2}$$

where ℏ is the reduced Planck constant, *Ψ* is the wave function and *E* the energy of the par‐ ticle, 1 / *m* ^ \* is the inverse effective mass tensor and *U(u)* is the potential energy. The potential energy *U(u)* is related to the electrostatic potential *V(u)* via the relation:

$$
\Delta L(\mu) = -qV(\mu) + \Delta E\_C(\mu) \tag{3}
$$

where *ΔEC(u)* represents the band offset at the interface between two materials (i.e. the band offset between the conduction band of the semiconductor and the conduction band of the oxide). Similarly to the Poisson equation, the general form of the Schrödinger equation (2) will be expressed in paragraph 2.3 in cylindrical coordinates. These two equations will be solved self-consistently and the solution of this system of equations will be coupled with the courant continuity equation which is written:

$$
\nabla \nabla f\_n = q \mathcal{R} + q \frac{\partial n}{\partial t} \tag{4}
$$

where *R* is the net rate of generation-recombination and *Jn*is the electrons current density given by the following equation in the case of the drift-diffusion model:

$$J\_u = -q\mu\_n n \nabla \phi + qD\_n \nabla n \tag{5}$$

where *μn* is where the electron mobility, *Dn* is the coefficient of thermal diffusion, and *n* is the electron density. In equation (5) the first term on the right corresponds to the drift com‐ ponent of the total current density and the second term to the diffusion component.

Poisson, Schrödinger and drift-diffusion equations are solved using a finite difference scheme with a uniform mesh on a domain including the channel, the source and drain re‐ gions, the gate-oxide layer and the gate electrodes (Fig. 2(a)). The electric field penetration in the source/drain and electron wave function penetration in the gate-oxide can be thus taken into account.

The general flowchart of our code is presented in Fig. 3. In a first step, the quantized energy levels as well as their associated electron wave functions and populations can be obtained from the self-consistent solving of the Schrödinger and Poisson equations. The next step concerns the continuity equation which is numerically solved to ensure a constant drain cur‐ rent along the device channel. The continuity equation gives the numerical solution of the quasi-Fermi level, used further in the calculation of the charge density in the Schrödinger-Poisson system. After reaching the convergence, the drain current density is finally evaluat‐ ed from the drift-diffusion formalism. In the following, we will explain in detail the three modules corresponding to the solving procedures for the Poisson, Schrödinger and current continuity equations. The discretization of theses equations and the simplifying assump‐ tions used in the calculations, assumptions mainly related to the cylindrical symmetry of the structure, will be systematically explained and justified.

#### **2.1. Description of the simulated devices**

The description of the 3-D architecture of the radially-symmetric nanowire MOSFETs con‐ sidered in the simulation is presented in Fig. 1. The definition of the main geometrical pa‐ rameters of this structure is also shown in Fig. 1. The structure is symmetric with intrinsic thin silicon film and highly doped source and drain (NSD = 3×1020 cm-3). A midgap metal gates (Φm=4.61 eV) and a 1.0-nm-thick gate-oxide have been also considered. The source is grounded, and the gate and the drain biased at VG and VD, respectively. In order to investi‐ gate the influence of short-channel effects and quantum mechanical confinement, different gate lengths and nanowire diameters will be simulated in section 4.

**Figure 3.** Flowchart of the numerical code illustrating the different problems to solve in the quantum drift-diffusion calculation of the output drain current.

#### **2.2. Poisson solver**

will be expressed in paragraph 2.3 in cylindrical coordinates. These two equations will be solved self-consistently and the solution of this system of equations will be coupled with the

where *R* is the net rate of generation-recombination and *Jn*is the electrons current density

where *μn* is where the electron mobility, *Dn* is the coefficient of thermal diffusion, and *n* is the electron density. In equation (5) the first term on the right corresponds to the drift com‐

Poisson, Schrödinger and drift-diffusion equations are solved using a finite difference scheme with a uniform mesh on a domain including the channel, the source and drain re‐ gions, the gate-oxide layer and the gate electrodes (Fig. 2(a)). The electric field penetration in the source/drain and electron wave function penetration in the gate-oxide can be thus taken

The general flowchart of our code is presented in Fig. 3. In a first step, the quantized energy levels as well as their associated electron wave functions and populations can be obtained from the self-consistent solving of the Schrödinger and Poisson equations. The next step concerns the continuity equation which is numerically solved to ensure a constant drain cur‐ rent along the device channel. The continuity equation gives the numerical solution of the quasi-Fermi level, used further in the calculation of the charge density in the Schrödinger-Poisson system. After reaching the convergence, the drain current density is finally evaluat‐ ed from the drift-diffusion formalism. In the following, we will explain in detail the three modules corresponding to the solving procedures for the Poisson, Schrödinger and current continuity equations. The discretization of theses equations and the simplifying assump‐ tions used in the calculations, assumptions mainly related to the cylindrical symmetry of the

The description of the 3-D architecture of the radially-symmetric nanowire MOSFETs con‐ sidered in the simulation is presented in Fig. 1. The definition of the main geometrical pa‐ rameters of this structure is also shown in Fig. 1. The structure is symmetric with intrinsic thin silicon film and highly doped source and drain (NSD = 3×1020 cm-3). A midgap metal gates (Φm=4.61 eV) and a 1.0-nm-thick gate-oxide have been also considered. The source is grounded, and the gate and the drain biased at VG and VD, respectively. In order to investi‐

ponent of the total current density and the second term to the diffusion component.

¶ (4)

*nn n J qμ n qD n* = - Ñf + Ñ (5)

*n <sup>n</sup> J qR q <sup>t</sup>* ¶ Ñ= +

given by the following equation in the case of the drift-diffusion model:

structure, will be systematically explained and justified.

**2.1. Description of the simulated devices**

courant continuity equation which is written:

346 Nanowires - Recent Advances

into account.

For a given channel geometry and bias conditions, the 2-D Poisson equation is solved (see flowchart in Fig. 3) considering a net charge density given by [23]:

$$\mathfrak{p} = q \Big[ \mathcal{N}\_{\text{DOPING}} - n \Big] \tag{6}$$

where *q* is the absolute value of the electron charge, *NDOPING* is the doping atom concentra‐ tions in the Si film (*NDOPING* = +*ND* in source and drain regions and *NDOPING* = -*NA* in the chan‐ nel) and *n(x, r)* is the electron density. This electron density is given by the Schrödinger module (see paragraph 2.3).

The Poisson equations expressed in cylindrical coordinates (x, r, θ) and ignoring the de‐ pendence in θ due to the circular symmetry of the electrostatic potential can be written un‐ der the form:

$$\frac{\partial^2 V}{\partial x^2} + \frac{1}{r} \frac{\partial V}{\partial r} + \frac{\partial^2 V}{\partial r^2} = -\frac{\rho(\mathbf{x}, r)}{\varepsilon} \tag{7}$$

This equation is discretized using a three-point finite difference scheme applied to the grid shown in Fig. 2, as follows:

$$\mathcal{V}\_{i,f} = \left[\frac{\mathbf{p}\_{i,f}}{\mathbf{x}\_f} + \mathbf{A}\_{i,f} + \mathbf{B}\_{i,f} \right] / \Sigma\_{i,f} \tag{8}$$

where *Ai,j*, *Bi,j* and *Σi,j* are given by:

$$A\_{i,j} = \frac{V\_{i-1,j} + V\_{i+1,j}}{\Delta x^2} + V\_{i,j-1} \left(\frac{1}{\Delta r^2} - \frac{1}{r\_j} \frac{1}{2\Delta r}\right) + V\_{i,j+1} \left(\frac{1}{\Delta r^2} + \frac{1}{r\_j} \frac{1}{2\Delta r}\right) \tag{9}$$

$$B\_{i,j} = \frac{\varepsilon\_{j+1} - \varepsilon\_{j-1}}{2\Delta r} \times \frac{V\_{i,j+1} - V\_{i,j-1}}{2\Delta r} \times \frac{1}{\varepsilon\_j} \tag{10}$$

$$
\Sigma\_{i\_{sj}} = \mathbf{2} \left( \frac{1}{\Delta \mathbf{x}^2} + \frac{1}{\Delta r^2} \right) \tag{11}
$$

where the index *i* identifies the mesh points in x-direction (*i* is an integer, *i*=*1*,..,*M*), j identi‐ fies the mesh points in r-direction (*j* is an integer, *j*=*1*,..,*N*), *Δx* and *Δr* are the mesh sizes be‐ tween two mesh points in the x-direction and r-direction, respectively (Fig. 2) and *r=j‧Δr*.*Δx* and *Δr* are considered constant in this work. The electrostatic potential calculated by eq. (8) is then used in the solving of the Schrödinger equation for the calculation of the quantum charge density (see below).

#### **2.3. Schrödinger solver**

The Schrödinger equation in cylindrical coordinates (x, r, θ) is given by:

$$-\frac{\hbar^2}{2} \left[ \frac{\partial}{\partial x} \left( \frac{1}{m^\star} \frac{\partial \Psi}{\partial x} \right) + \frac{\partial}{\partial r} \left( \frac{1}{m^\star} \frac{\partial \Psi}{\partial r} \right) + \frac{1}{r} \frac{1}{m^\star} \frac{\partial \Psi}{\partial r} + \frac{1}{r^2} \frac{\partial}{\partial \theta} \left( \frac{1}{m^\star} \frac{\partial \Psi}{\partial \theta} \right) \right] + lI\Psi = E\Psi\tag{12}$$

where *m\*(x, r, θ)* is a position dependent effective mass, *U(x, r, θ)* is the potential energy of the conduction-band edge. It is assumed that *m\** is constant within the silicon nanowire so that it is not dependent on *θ*. The effective mass is also considered energy-independent (non-parabolicity effects are thus neglected).

Due to the cylindrical symmetry, we may consider wave functions of the form:

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric Nanowire MOSFETs http://dx.doi.org/10.5772/52587 349

$$\Psi(\mathbf{x},r,\theta) = \Psi(\mathbf{x},r) \, e^{i\mathbf{n}\theta} \tag{13}$$

where *m*=0, ± 1, ± 2, ± 3,… is the angular momentum quantum number. This wave function is normalized according to:

$$2\pi \int\_0^R r \left| \Psi(\mathbf{x}, r) \right|^2 dr = 1 \tag{14}$$

where the factor *2π* is due to the integration with respect to *θ*.

Using relation (13), equation (12) becomes:

2 2 2 2 *V V V xr* 1 (,)

*x r r r*

, , ,, , / *i j i j ij ij ij j V AB* é ù <sup>r</sup> = ++ S ê ú ê ú <sup>e</sup> ë û

, 22 2 , 1 , 1

*AV V xr r r r r r*

2 2 *j j ij ij*

, 2 2 1 1 <sup>2</sup> *i j x r* æ ö S= + ç ÷

*<sup>B</sup> r r* +- + - e -e -


*V V*

where the index *i* identifies the mesh points in x-direction (*i* is an integer, *i*=*1*,..,*M*), j identi‐ fies the mesh points in r-direction (*j* is an integer, *j*=*1*,..,*N*), *Δx* and *Δr* are the mesh sizes be‐ tween two mesh points in the x-direction and r-direction, respectively (Fig. 2) and *r=j‧Δr*.*Δx* and *Δr* are considered constant in this work. The electrostatic potential calculated by eq. (8) is then used in the solving of the Schrödinger equation for the calculation of the quantum

1 1 ,1 ,1

*i j i j i j*

The Schrödinger equation in cylindrical coordinates (x, r, θ) is given by:

\* \* \* 2\* 1 1 11 1 1

Due to the cylindrical symmetry, we may consider wave functions of the form:

*U E x x r rr r m m m rm* é ù ¶ ¶Y ¶ ¶Y ¶Y ¶ ¶Y æ öæ ö æ ö - ê ú ç ÷ç ÷ ç ÷ <sup>+</sup> + + + Y= Y ë û ¶ ¶ ¶ ¶ ¶ ¶q ¶q è øè ø è ø

where *m\*(x, r, θ)* is a position dependent effective mass, *U(x, r, θ)* is the potential energy of the conduction-band edge. It is assumed that *m\** is constant within the silicon nanowire so that it is not dependent on *θ*. The effective mass is also considered energy-independent

<sup>h</sup> (12)

shown in Fig. 2, as follows:

348 Nanowires - Recent Advances

where *Ai,j*, *Bi,j* and *Σi,j* are given by:

charge density (see below).

2

2

(non-parabolicity effects are thus neglected).

**2.3. Schrödinger solver**

1, 1,

,

*i j*

*ij ij*


*V V*

This equation is discretized using a three-point finite difference scheme applied to the grid

1 11 1 11 2 2

*j j*

1

*j*

=´ ´ D De (10)

è ø D D (11)

¶ ¶¶ r + + =- ¶ ¶ ¶ e (7)

(8)

(9)

$$-\frac{\hbar^2}{2} \left[ \frac{\partial}{\partial \mathbf{x}} \left( \frac{1}{m^\*} \frac{\partial \Psi}{\partial \mathbf{x}} \right) + \frac{\partial}{\partial r} \left( \frac{1}{m^\*} \frac{\partial \Psi}{\partial r} \right) + \frac{1}{r} \frac{1}{m^\*} \frac{\partial \Psi}{\partial r} \right] + \left( \mathcal{U} + \frac{\hbar^2}{2} \frac{1}{m^\*} \frac{m^2}{r^2} \right) \Psi = E\Psi \tag{15}$$

We recall here that the Schrödinger equation is solved along vertical cut-lines (1-D mesh along the radial direction), perpendicular to the x axis, in each mesh-point *i* of the x-axis (Fig. 2(b)). The procedure for solving the Schrödinger equation will be the same for each mesh-point *i*. In the following, we set a fixed mesh point *i* (then the dependence on x in the equation (14) vanishes); but this procedure will be repeated *M* times for each mesh-point of the x-axis (this will be explained in detail at the end of this paragraph).

As the Poisson equation, Schrödinger equation (14) is discretized using finite differences scheme to obtain a standard matrix eigen-value problem:

$$H\Psi = E\Psi\tag{16}$$

where H is the general Hamiltonian matrix. In practice, equation (16) is solved for each val‐ ue of the angular momentum quantum number m separately. In the following, we will de‐ scribe in detail the Hamiltonian H for a given m value.

To simplify the equations, we consider firstly *m*=0, and we will address the other cases later in this paragraph. With this condition, the discrete form of the Schrödinger equation for the ladder of states with *m*=0 on a mesh point *j* along the radial direction is written as:

$$-\frac{\hbar^2}{2} \left[ \frac{\Psi\_{j+1} - \Psi\_j}{m^\* \,\Delta r^2} \right] + \frac{\hbar^2}{2} \left[ \frac{\Psi\_j - \Psi\_{j-1}}{m^\* \,\Delta r^2} \right] - \frac{\hbar^2}{2} \frac{1}{(j-1)\Delta r} \left[ \frac{\Psi\_{j+1} - \Psi\_{j-1}}{2m^\* \,\Delta r} \right] + \mathcal{U}\_{l,j} \Psi\_j = E \Psi\_j \tag{17}$$

where the effective mass has been considered independent on r (since we simulate here a fully silicon nanowire). In *Uij*, the index *i* corresponds to the cut-line on the x axis for which the Schrödinger equation is solved. The potential energy term *Uij* is obtained using eq. (3) which is discretized as:

$$
\Delta L\_{i,f} = -qV\_{i,f} + \Delta E\_{Cl,f} \tag{18}
$$

where *Vi,j* is the electrostatic potential obtained from the resolution of the Poisson equation as described in the previous paragraph (Poisson module).

We explain below how the effective masses are calculated in the cylindrical silicon nanowire considered here.The conduction band of silicon has 6 equivalent minima. The variation of the energy of the conduction band near the minimum is not isotropic and constant energy surfaces are ellipsoids around each of the axes (Fig. 4). In each minimum, the electrons prop‐ agating along the axis of the ellipsoid are characterized by a longitudinal effective mass, *ml* , and those that propagate in a direction perpendicular to this axis are characterized by a transverse effective mass, *mt* . In the circular nanowire simulated here, the two valleys along the x direction (transport direction) have a confinement mass equal to *mt* and a transport mass equal to ml . The other four valleys following the directions y and z are anisotropic.

**Figure 4.** Schematic illustration of electron confinement in a Silicon nanowire with circular cross-section.

In order to exploit the cylindrical symmetry, we use an approximate isotropic effective mass in the radial direction. Because of this assumption, now there are only two different types of conduction band valleys: the valley along the x direction which is doubly degenerate and the valley along the radial (r) direction whose degeneracy is equal to 4. The effective masses (mr, mx) along radial and axial directions for the two types of valleys are given in Table 1.


**Table 1.** Definition of the effective masses along the radial and axial directions in cylindrical nanowire (*mt=0.19×m0, ml =0.98×m0, mr2=0.35×m0, m0= 9.11×10-31Kg).*

It is important to note that in addition to the solving in one dimension of the Schrödinger equation, which drastically reduces the computation time compared to a 2-D solving, the computation time is further reduced because of the smaller number of valleys taken into ac‐ count in the calculations compared to the case where an anisotropic effective mass is consid‐ ered in the radial direction.The choice of an approximate isotropic effective mass for rdirection is justified by the fact that the eigenvectors and eigenenergies resulting from the cylindrical solver are very similar to those of an exact 2-D solver taking into account the ani‐ sotropy of the effective mass [24]. In the following, the effective mass *m\** used in the Schrö‐ dinger equation is equal to the radial effective mass *mr*, which may have two distinct values, *mr1* for the valleys 1 and 2 and *mr2* for the valleys 3, 4, 5 and 6, as shown in Table 1.

Let's go back now to the solving the Schrödinger equation. Using the discrete form given by eq. (17), the Schrödinger equation for *m*=0 may be set in the form of a matrix equation:

$$H^0 \Psi^0 = E^0 \Psi^0 \tag{19}$$

where H0 (0 stands for *m*=0 condition) is the N×N tridiagonal hamiltonian matrix of the de‐ vice (N is the total grid-point number in r-direction):

$$H^0 = \begin{bmatrix} \bullet & \times & 0 & \cdots & 0 \\ \circ & \bullet & \times & \cdot & \vdots \\ 0 & \circ & \bullet & \times & 0 \\ \vdots & \ddots & \circ & \bullet & \times \\ 0 & \cdots & 0 & \circ & \bullet \end{bmatrix} \tag{20}$$

where symbols are given by the following equations:

*U qV E i j i j Ci j* ,,, =- +D (18)

. In the circular nanowire simulated here, the two valleys along

,

and a transport

y

x

z

*mt*

**4**

*mt*

**5 6**

**3**

q

*ml*

where *Vi,j* is the electrostatic potential obtained from the resolution of the Poisson equation

We explain below how the effective masses are calculated in the cylindrical silicon nanowire considered here.The conduction band of silicon has 6 equivalent minima. The variation of the energy of the conduction band near the minimum is not isotropic and constant energy surfaces are ellipsoids around each of the axes (Fig. 4). In each minimum, the electrons prop‐ agating along the axis of the ellipsoid are characterized by a longitudinal effective mass, *ml*

and those that propagate in a direction perpendicular to this axis are characterized by a

*ml*

In order to exploit the cylindrical symmetry, we use an approximate isotropic effective mass in the radial direction. Because of this assumption, now there are only two different types of conduction band valleys: the valley along the x direction which is doubly degenerate and the valley along the radial (r) direction whose degeneracy is equal to 4. The effective masses (mr, mx) along radial and axial directions for the two types of valleys are given in Table 1.

**equivalent valleys Radial effective mass, mr (confinement) Axial effective mass, mx(transport)**

. The other four valleys following the directions y and z are anisotropic.

the x direction (transport direction) have a confinement mass equal to *mt*

*mt*

**6**

**Figure 4.** Schematic illustration of electron confinement in a Silicon nanowire with circular cross-section.

1-2 2 *mr1 = mt ml*

*mt* + *ml*

**Table 1.** Definition of the effective masses along the radial and axial directions in cylindrical nanowire (*mt=0.19×m0,*

**1 2**

**3**

as described in the previous paragraph (Poisson module).

transverse effective mass, *mt*

y

z

**Valley**

*ml*

x

**Number of**

*=0.98×m0, mr2=0.35×m0, m0= 9.11×10-31Kg).*

3-6 <sup>4</sup> *mr*<sup>2</sup> <sup>=</sup> <sup>2</sup> *mt ml*

**5**

**4**

mass equal to ml

350 Nanowires - Recent Advances

$$\rho \circ = -\frac{\hbar^2}{2m\_r \Delta r^2} + \frac{\hbar^2}{4m\_r \Delta r^2 (j-1)}\tag{21}$$

$$
\bullet = \frac{\hbar^2}{m\_r \Delta r^2} + \mathcal{U}\_{l,f} \tag{22}
$$

$$\times = -\frac{\hbar^2}{2m\_r \Delta r^2} - \frac{\hbar^2}{4m\_r \Delta r^2 (j-1)}\tag{23}$$

The matrix H0 defined in (20) is a non-symmetric tridiagonal matrix and the computation of eigenvalues and eigenvectors of such matrix is numerically not optimized. However, using the method proposed in [16] by Zervos and Feiner, we can convert *H0* into a symmetric ma‐ trix *S* using the relation:

$$S = TH^0\tag{24}$$

where *T* is an (NxN) diagonal matrix whose elements are determined as follows. Since *S* is a symmetric matrix, it must fill the condition:

$$\mathcal{S}\_{j,f+1} = \mathcal{S}\_{f+1,f} \tag{25}$$

Equation (24) can be rewritten in a discretized form as:

$$\mathcal{S}\_{f,f+1} = T\_{f,f} H^0\_{f,f+1} \tag{26}$$

Combining equations (25) and (26) we obtain the following relation for matrix *T*:

$$\,\_{1}T\_{j,j}H\_{j,j+1}^{0} = T\_{j+1,j+1}H\_{j+1,j}^{0} \tag{27}$$

Then we can obtain the elements of the matrix *T* [16]: the first element *T11* is set equal to 1 and the subsequent elements are obtained from (27) as:

$$T\_{j+1,j+1} = T\_{j,j} H^0\_{j,j+1} / H^0\_{j+1,j} \tag{28}$$

Matrix *S* is then symmetric and using equation (19) we obtain:

$$S\Psi^0 = TH^0\Psi^0 = E^0T\Psi^0\tag{29}$$

In equation (29) *Ψ<sup>0</sup>* and *E0* are the eigenvector and the eigenvalue of the real matrix *H0* , re‐ spectively. Then, equation (29) has to be further transformed for obtaining a standard matrix eigenvalue problem. As shown in [16], we define a diagonal matrix *L* whose elements are the square root of the corresponding elements of *T*:

$$T\_{j,j} = L\_{j,j}^2 \tag{30}$$

Multiplying eq. (29) from the left by *L-1* we obtain:

$$L^{-1}SL^{-1}L\Psi^0 = L^{-1}LLH^0\Psi^0 = E^0L^{-1}LL\Psi^0 = E^0L\Psi^0 \tag{31}$$

We define:

$$
\Delta H^N = L^{-1}SL^{-1} \tag{32}
$$

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric Nanowire MOSFETs http://dx.doi.org/10.5772/52587 353

and

<sup>0</sup> *S TH* = (24)

*jj j j* , 1 1, *S S* + + = (25)

*jj jj jj* ,1 , ,1 *S TH* + + = (26)

*jj jj j j j j* , , 1 1, 1 1, *TH T H* + ++ + = (27)

1, 1 , , 1 1, / *j j jj jj j j T TH H* + + + + = (28)

0 00 0 0 *S TH E T* Y= Y= Y (29)

*jj jj* , , *T L* = (30)

*<sup>N</sup>* 1 1 *H L SL* - - = (32)

, re‐

are the eigenvector and the eigenvalue of the real matrix *H0*

1 1 0 1 00 01 0 0 0 *L SL L L LLH E L LL E L* -- - - Y= Y= Y= Y (31)

where *T* is an (NxN) diagonal matrix whose elements are determined as follows. Since *S* is a

0

Combining equations (25) and (26) we obtain the following relation for matrix *T*:

0 0

Then we can obtain the elements of the matrix *T* [16]: the first element *T11* is set equal to 1

0 0

spectively. Then, equation (29) has to be further transformed for obtaining a standard matrix eigenvalue problem. As shown in [16], we define a diagonal matrix *L* whose elements are

2

symmetric matrix, it must fill the condition:

352 Nanowires - Recent Advances

Equation (24) can be rewritten in a discretized form as:

and the subsequent elements are obtained from (27) as:

and *E0*

the square root of the corresponding elements of *T*:

Multiplying eq. (29) from the left by *L-1* we obtain:

In equation (29) *Ψ<sup>0</sup>*

We define:

Matrix *S* is then symmetric and using equation (19) we obtain:

$$
\Delta \Psi^N = L \Psi^0 \tag{33}
$$

Then, eq. (29) becomes:

$$H^N \Psi^N = E^0 \Psi^N \tag{34}$$

which is the new eigenvalue problem. The new Hamiltonian *HN* is a symmetric tridiagonal ma‐ trix whose eigenvalues are *E0* and the eigenvectors are *Ψ<sup>N</sup>*. Then, we solve eq. (34) instead of eq. (19) to find eigenvalues and eigenvectors, and to finally calculate the quantum carrier density. The eigenenergies *E0* corresponding to the wave functions *Ψ<sup>N</sup>* are computed using a QL algo‐ rithm [25] with implicit shifts. Next, we obtain the original wave functions *Ψ<sup>0</sup>* from eq. (33) as:

$$
\Psi^{0} = L^{-1} \Psi^{N} \tag{35}
$$

So far, we explained the resolution of the Schrödinger equation for *m*=0, which has simpli‐ fied the equations. To take into account all values of m, eq. (15) must be solved following the same procedure as that used for *m*=0. The discretization of equation (15) leads to a matrix relation of the same form as the relation (17):

$$H^{\*m}\Psi^{\*m} = E^{\*m}\Psi^{\*m} \tag{36}$$

In this equation, the Hamiltonian *Hm* is always a tridiagonal matrix which has the same ele‐ ments "◦" and "×", respectively given by eqs. (21) and (23). The only term that changes is the diagonal term, "• ", which becomes:

$$\bullet = \frac{\hbar^2}{m\_r \Delta r^2} + \frac{\hbar^2}{2m\_r} \frac{m^2}{j^2 \Delta r^2} + \mathcal{U}\_{i,j} \tag{37}$$

Equation (36) is a new eigenvalue problem to be solved separately for each *m* to find the lad‐ der states for *m*=0, ± 1, ± 2, ± 3,… The eigenvalues and eigenvectors are determined using the same procedure as previously used for the matrix *H0* . For each m, a distinct set of wave func‐ tions and energy levels is found. The energy levels obtained in this way for all values of m will be sorted and classified in ascending order, from the lowest level and to the highest level. The wave function corresponding to each classified level will be also stored. The first lowest levels (and the corresponding wave functions), in a number equal to *Nlevel*, will be used next to calcu‐ late the electron density, *n(r)*. In the following, we call *Ek* (respectively *Ψk*), these first lowest levels (respectively their corresponding wave function), where *k* is the subband index (*k*=*1*,..,*Nlevel*) which implicitly includes the angular momentum quantum number *m*.

The procedure described before for the solving of the Schrödinger equation is performed for both *mr*=*mr1* and *mr*=*mr2*. Two distinct sets of energy levels and wave functions are then ob‐ tained and are used in the following for the calculation of the electron density, *n(r)*.

The 1-D electron gas density is given by:

$$m(r) = \sum\_{r1,r2} \mathbf{g}\_{r1,r2} \sum\_{k=1}^{N\_{\text{local}}} n\_k \left| \; \Psi\_k(r) \right|^2 \tag{38}$$

where *gr1,r2* is the valley degeneracy (2 for *mr*=*mr1* and 4 for *mr*=*mr2*) and *nk* is the electron den‐ sity per unit of length of the *k*th subband given by:

$$m\_k = N\_{1D} \Im\_{-1/2} \left( \frac{E\_{Fn} - E\_k}{k\_B T} \right) \tag{39}$$

where *N1D* is the effective density of states (DOS), *kB* is the Boltzmann constant, *T* is the tem‐ perature, *EFn* is the quasi-Fermi level which will be calculated in the current continuity mod‐ ule (by a self-consistent coupling of the solution of the Poisson-Schrödinger system to the current continuity equation) and ℑ−1/2 is the Fermi-Dirac integral of order -1/2 [26]. The 1-D density-of-states *N1D* is calculated as:

$$N\_{1D} = \frac{1}{\pi} \sqrt{\frac{2m\_{s1\_{F}2}^{1D} k\_{B}T}{\hbar^2}}\tag{40}$$

where *mr*1,*r*<sup>2</sup> <sup>1</sup>*D* is the 1-D density of states effective mass (*mr*<sup>1</sup> <sup>1</sup>*<sup>D</sup>* <sup>=</sup>*ml* for valleys 1-2 and *mr*<sup>2</sup> <sup>1</sup>*<sup>D</sup>* <sup>=</sup>*mt*for valleys 3-6). Similar to energy levels and wave functions, *N1D* and *nk* are calculated for the two types of valleys described in Table 1. Then, these values are used in eq. (38) for the first sum in the right side of the equation. Note that the unity of *nk* is m-1 which multiplied by the normal‐ ized probability density | *Ψ<sup>k</sup>* (*r*)| 2(eq. 38) gives the volume electron density *n(r)*.

As we said before, the 1-D Schrödinger equation is solved (using the procedure outlined above) in a vertical cut-line in each mesh point of the x axis (which means that it will be solved *M* times). For each cut-line in a given mesh-point *i*, the electron density *n(r)* is calcu‐ lated using eq. (38), and it will then be assigned to the mesh-point point *i*; this makes possi‐ ble to build step by step the total density *n(x, r)*. This density is injected into the module solving the 2-D Poisson equation (eqs. (6)and (8), paragraph 2.2), which will provide a new potential profile *V(x, r).* This new potential will be introduced later in the solving of the Schrödinger equation, which will give the new carrier density *n(x, r)* (in the same manner as described above), which in turn will be injected into Poisson's equation and so on. This pro‐ cedure allows us to self-consistently solve the two Poisson and Schrödinger equations. The loop stops when the convergence criterion is reached, as shown in Fig. 3.

#### **2.4. Current continuity module**

The procedure described before for the solving of the Schrödinger equation is performed for both *mr*=*mr1* and *mr*=*mr2*. Two distinct sets of energy levels and wave functions are then ob‐

where *gr1,r2* is the valley degeneracy (2 for *mr*=*mr1* and 4 for *mr*=*mr2*) and *nk* is the electron den‐

*Fn k*

*B*

where *N1D* is the effective density of states (DOS), *kB* is the Boltzmann constant, *T* is the tem‐ perature, *EFn* is the quasi-Fermi level which will be calculated in the current continuity mod‐ ule (by a self-consistent coupling of the solution of the Poisson-Schrödinger system to the current continuity equation) and ℑ−1/2 is the Fermi-Dirac integral of order -1/2 [26]. The 1-D

> 1 1, 2

*m kT*

valleys 3-6). Similar to energy levels and wave functions, *N1D* and *nk* are calculated for the two types of valleys described in Table 1. Then, these values are used in eq. (38) for the first sum in the right side of the equation. Note that the unity of *nk* is m-1 which multiplied by the normal‐

As we said before, the 1-D Schrödinger equation is solved (using the procedure outlined above) in a vertical cut-line in each mesh point of the x axis (which means that it will be solved *M* times). For each cut-line in a given mesh-point *i*, the electron density *n(r)* is calcu‐ lated using eq. (38), and it will then be assigned to the mesh-point point *i*; this makes possi‐ ble to build step by step the total density *n(x, r)*. This density is injected into the module solving the 2-D Poisson equation (eqs. (6)and (8), paragraph 2.2), which will provide a new potential profile *V(x, r).* This new potential will be introduced later in the solving of the Schrödinger equation, which will give the new carrier density *n(x, r)* (in the same manner as described above), which in turn will be injected into Poisson's equation and so on. This pro‐ cedure allows us to self-consistently solve the two Poisson and Schrödinger equations. The

<sup>1</sup>*<sup>D</sup>* <sup>=</sup>*ml*

1 2 1 2 *<sup>D</sup> rr B*

*<sup>N</sup>* <sup>=</sup> <sup>p</sup> <sup>h</sup>

ized probability density | *Ψ<sup>k</sup>* (*r*)| 2(eq. 38) gives the volume electron density *n(r)*.

loop stops when the convergence criterion is reached, as shown in Fig. 3.

*D*

<sup>1</sup>*D* is the 1-D density of states effective mass (*mr*<sup>1</sup>

2

<sup>=</sup> å å <sup>Y</sup> (38)

è ø (39)

for valleys 1-2 and *mr*<sup>2</sup>

(40)

<sup>1</sup>*<sup>D</sup>* <sup>=</sup>*mt*for

tained and are used in the following for the calculation of the electron density, *n(r)*.

1, 2 1, 2 1 ( ) ( ) *Nlevel rr k k*

1 1/2

*E E n N k T* æ ö - = Á ç ÷

*k D*

*rr k nr g n r* =

The 1-D electron gas density is given by:

354 Nanowires - Recent Advances

density-of-states *N1D* is calculated as:

where *mr*1,*r*<sup>2</sup>

sity per unit of length of the *k*th subband given by:

As we explained at the beginning of Section 2, the solution of the Schrödinger-Poisson sys‐ tem of equations is coupled with the current continuity equation given by relation (4). In eq. (4), the current density is defined by a model of charge transport, usually obtained in a semi-classical approximation or simplifying the Boltzmann transport equation (ETB). The simplifying assumptions made on this equation lead to several different semi-classical models such as the drift-diffusion model or the hydrodynamic model. Nevertheless, what‐ ever the model used for the calculation of the carrier density (drift-diffusion, hydrodynam‐ ic, etc), the condition of current continuity should be guaranteed. Indeed, the continuity equations describe the evolution of carriers in the silicon nanowire film (source-channeldrain) in order to maintain a constant current along the film in which there is no charge ac‐ cumulation.

In this work we consider the drift-diffusion model for which the current density is given by equation (5). The charge transport in the nanowire MOSFET simulated here involves only the minority carrier conduction. Therefore, the current density used herein concerns unique‐ ly the electrons (for an n-channel transistor, NMOSFET), but a similar equation can be used for holes in the case of a p-channel transistor (PMOSFET) and the procedure of solving the current continuity equation will be rigorously identical with that shown below for electrons. In addition, to simplify the equations, we consider a stationary regime of operation and we assume that no process of generation or recombination of carriers occurs in our simulations. With these assumptions, equation (4) becomes:

$$
\nabla I\_n = 0 \tag{41}
$$

The current density of electrons can be rewritten as a function of the quasi-Fermi level of electrons in the silicon nanowire, *ΦF*, as:

$$\mathcal{J}\_n = -q\mu\_n n \nabla \Phi + q\mathcal{D}\_n \nabla n = q\mu\_n n \nabla \Phi\_F \tag{42}$$

where the electron diffusion coefficient, *Dn*, is calculated from the Einstein's relation (for the non-degenerate case):

$$D\_n = \mu\_n \cdot \left(\frac{k\_B T}{q}\right) \tag{43}$$

In this work, the electron mobility *μn* is considered constant everywhere in the silicon nano‐ wire. Substituting equation (42) into equation (41) we obtain:

$$\nabla \left( -\mu\_n \mu \nabla V + D\_n \nabla n \right) = 0 \tag{44}$$

Using the properties of vector analysis and after various algebraic manipulations, equation (44) becomes:

$$
\Delta\Phi\_F + \oint \nabla \Phi\_F \,\nabla \left( V - \Phi\_F \right) = 0 \tag{45}
$$

where *β=kBT/q* and *V* is the electrostatic potential which comes from the solution of the coupled solving of the Poisson and Schrödinger equations. Equation (45) can be rewrit‐ ten:

$$
\Delta \Phi\_F(\mathbf{x}, r) = -G(\mathbf{x}, r) \tag{46}
$$

where

$$\mathbf{G} = \emptyset \,\nabla \Phi\_{\mathbf{F}} \,\nabla \left( V - \Phi\_{\mathbf{F}} \right) \tag{47}$$

In cylindrical coordinates, *ΔΦ<sup>F</sup>* (*x*, *r*) is given by the following expression:

$$
\Delta\Phi\_F(\mathbf{x}, r) = \frac{\partial^2 \Phi\_F}{\partial \mathbf{x}^2} + \frac{1}{r} \frac{\partial \Phi\_F}{\partial r} + \frac{\partial^2 \Phi\_F}{\partial r^2} \tag{48}
$$

and G is written as:

$$\mathbf{G} = \Re\left(\frac{\partial \Phi\_F}{\partial \mathbf{x}} + \frac{\partial \Phi\_F}{\partial r}\right) \left[\frac{\partial \left(V - \Phi\_F\right)}{\partial \mathbf{x}} + \frac{\partial \left(V - \Phi\_F\right)}{\partial r}\right] \tag{49}$$

Equation (46) is discretized using a three-point finite difference scheme applied to the mesh shown in Fig. 2, as follows:

$$\Phi\_{F\ell,j} = \frac{\mathbf{G}\_{\ell,j} + \mathbf{C}\_{\ell,j}}{\Sigma\_{\ell,j}} \tag{50}$$

where Σi,j is given by equation (11) and Gi,j and Ci,j are given by:

$$\mathbf{G}\_{i\_\*f} = \emptyset \,\, \mathbf{D}\_{i\_\*f} \,\, \mathbf{E}\_{i\_\*f} \tag{51}$$

$$\mathbf{C}\_{i,j} = \frac{\boldsymbol{\Phi}\_{\mathrm{F}\boldsymbol{i}-1,j} + \boldsymbol{\Phi}\_{\mathrm{F}\boldsymbol{i}+1,j}}{\Delta x^2} + \boldsymbol{\Phi}\_{\mathrm{F}\boldsymbol{i},j-1} \left(\frac{1}{\Delta r^2} - \frac{1}{r\_{\boldsymbol{j}}} \frac{1}{2\Delta r}\right) + \boldsymbol{\Phi}\_{\mathrm{F}\boldsymbol{i},j+1} \left(\frac{1}{\Delta r^2} + \frac{1}{r\_{\boldsymbol{j}}} \frac{1}{2\Delta r}\right) \tag{52}$$

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric Nanowire MOSFETs http://dx.doi.org/10.5772/52587 357

$$D\_{i,j} = \left[\frac{\Phi\_{Fi+1,j} - \Phi\_{Fi-1,j}}{2\Delta x} \cdot \frac{\left(V\_{i+1,j} - \Phi\_{Fi+1,j}\right) - \left(V\_{i-1,j} - \Phi\_{Fi-1,j}\right)}{2\Delta x}\right] \tag{53}$$

$$E\_{i,j} = \left[\frac{\boldsymbol{\Phi}\_{Fi,j+1} - \boldsymbol{\Phi}\_{Fi,j+1}}{2\Delta r} \cdot \frac{\left(V\_{i,j+1} - \boldsymbol{\Phi}\_{Fi,j+1}\right) - \left(V\_{i,j-1} - \boldsymbol{\Phi}\_{Fi,j-1}\right)}{2\Delta r}\right] \tag{54}$$

The quasi-Fermi level calculated by eq. (50) is then injected into the system of equations Poisson-Schrödinger, where it is used for the calculation of the carrier density. This density is then used to calculate a new potential, which is then reinserted into the module solving the continuity equation that gives a new quasi-Fermi level. In its turn, this new quasi-Fermi level will be injected into the Schrödinger-Poisson and so on. The loop will stop when the convergence criterion is reached. The final values of the electron density, *n(x,r)*, and the qua‐ si-Fermi level, *ΦF(x,r)*, will be used in the following for calculating the current density and the final drain current of the device.

#### **2.5. Drain current calculation**

Using the properties of vector analysis and after various algebraic manipulations, equation

where *β=kBT/q* and *V* is the electrostatic potential which comes from the solution of the coupled solving of the Poisson and Schrödinger equations. Equation (45) can be rewrit‐

> 2 2 2 2 <sup>1</sup> (,) *F FF <sup>F</sup> x r x r r r* ¶ F ¶F ¶ F DF = + +

Equation (46) is discretized using a three-point finite difference scheme applied to the mesh

, ,

, *ij ij*

*i j*

,

, 22 2 , 1 , 1

*<sup>C</sup> xr r r r r r*


*i j Fi j Fi j*

*G C*<sup>+</sup> F =

*Fi j*

where Σi,j is given by equation (11) and Gi,j and Ci,j are given by:

1, 1,

*Fi j Fi j*


( ) ( ) *F F V V F F <sup>G</sup> xr x r* æ ö ¶F ¶F é ù ¶ -F ¶ -F =b + ç ÷ ê ú <sup>+</sup> ¶¶ ¶ ¶ è ø ë û

In cylindrical coordinates, *ΔΦ<sup>F</sup>* (*x*, *r*) is given by the following expression:

( ) 0 DF + b ÑF Ñ - F = *FF F V* (45)

(,) (,) *<sup>F</sup>* DF = - *xr Gxr* (46)

( ) *G V F F* = b ÑF Ñ - F (47)

¶ ¶ ¶ (48)

<sup>S</sup> (50)

*G DE ij ij ij* , ,, = b (51)

1 11 1 11 2 2

*j j*

(49)

(52)

(44) becomes:

356 Nanowires - Recent Advances

ten:

where

and G is written as:

shown in Fig. 2, as follows:

The drain current density (per unit area) in the drift-diffusion formalism is finally evaluated from equation (42) which can be written as:

$$J\_n(\mathbf{x}, r) = q \mid \mu\_n \ n(\mathbf{x}, r) \; \nabla \Phi\_F(\mathbf{x}, r) \tag{55}$$

The total current drain as a function of the applied voltages is then calculated by summing the contribution of each current line in the silicon film.

Finally, the code allows us to store the main internal key-parameters such as the carrier den‐ sity, quantum energy levels, wave functions, potential, quasi-Fermi level, etc. Some of these parameters will be used for the code validation and for the analysis of the device performan‐ ces in terms of threshold voltage or short-channel effects.

#### **3. Simulation code validation**

Before using the code for analyzing the operation of nanowire MOSFETs, we compared the results of the Schrödinger-Poisson module to the same data obtained from a commercial simulator (Silvaco, [24]) which can simulate cylindrical nanowires. We therefore compare in the following the results issued from Cylmos and Silvacoin terms of quantum energy levels, wave functions and drain current. Each time, we will obtain a very good agreement between the two data sets, which demonstrates the validity of our code.

#### **3.1. Quantum energy levels and wave functions**

We begin this comparison by looking in detail the internal parameters of the Schröding‐ er module, namely the quantum energy levels and associated wave functions. We distin‐ guish here between the two radial effective masses, *mr1* and *mr2*. Thus, two sets of energy levels and wave function are compared, one set for each effective mass. The results shown in the following are obtained for a nanowire of diameter D=10 nm (i.e. R=5 nm). The simulated MOSFET is a long channel transistor, with 1-nm-thick gate oxide and an intrinsic channel. Other geometric parameters and doping levels are those described in paragraph 2.1.

Table 2 shows the comparison between the energy levels, *Ek*, calculated by Cylmos and those issued from Silvaco for the radial effective mass. The transistor is biased at VG = 0 V and a very low polarization is applied on the drain (VD=10 mV). As the channel is long and the drain bias is negligible, the potential does not vary much along the axial direction x. Therefore, almost identical energy levels will be obtained in any mesh point *i* of the intrinsic channel region (outside the source and drain regions). The results shown in Table 2 (as well as those reported in Figs. 5-6, Table 3 and Fig. 7) are calculated for a vertical cut-line in the middle of the channel. Table 2 shows that the energy levels for the radial effective mass cal‐ culated by Cylmos agree very well with those obtained using Silvaco.


**Table 2.** Energy levels, *Ek*, calculated by Cylmos and issued from Silvaco for the radial effective mass *mr1* in a circular nanowire of diameter D=10 nm. The transistor is biased at VG=0 V and VD=10 mV. The energy levels are calculated for a vertical cut-line in the middle of the channel.

The wave functions corresponding to the energy levels reported in Table 2 are plotted in Fig. 5 as a function of the position on the radial direction (for *k* from 1 to 6) and in Fig. 6 (for *k* from 7 to 10). The wave functions calculated by Cylmos correspond very well to those from Silvaco, particularly in the region corresponding to the silicon (0 to 5 nm). However, a rather clear difference is observed in the oxide region, due to the fact that Silvaco, contrary to Cy‐ lmos, does not take into account the penetration of wave functions in the oxide. This phe‐ nomenon is included in Cylmos, as previously stated, which leads to this discrepancy between the wave functions of Cylmos and Silvaco in the gate oxide.

The same analysis can be conducted for the energy levels and wave functions calculated for the second radial effective mass, *mr2*. These results are shown in Table 3 and Fig. 7 for the same transistor and the same polarization as those used above. A very nice agreement is found again between the data calculated by Cylmos and those obtained from Silvaco.

16 Nanowires - Recent Advances

Cylmos, does not take into account the penetration of wave functions in the oxide. This phenomenon is included in Cylmos, as previously stated, which leads to this discrepancy

between the wave functions of Cylmos and Silvaco in the gate oxide.

**3.1. Quantum energy levels and wave functions**

paragraph 2.1.

358 Nanowires - Recent Advances

We begin this comparison by looking in detail the internal parameters of the Schröding‐ er module, namely the quantum energy levels and associated wave functions. We distin‐ guish here between the two radial effective masses, *mr1* and *mr2*. Thus, two sets of energy levels and wave function are compared, one set for each effective mass. The results shown in the following are obtained for a nanowire of diameter D=10 nm (i.e. R=5 nm). The simulated MOSFET is a long channel transistor, with 1-nm-thick gate oxide and an intrinsic channel. Other geometric parameters and doping levels are those described in

Table 2 shows the comparison between the energy levels, *Ek*, calculated by Cylmos and those issued from Silvaco for the radial effective mass. The transistor is biased at VG = 0 V and a very low polarization is applied on the drain (VD=10 mV). As the channel is long and the drain bias is negligible, the potential does not vary much along the axial direction x. Therefore, almost identical energy levels will be obtained in any mesh point *i* of the intrinsic channel region (outside the source and drain regions). The results shown in Table 2 (as well as those reported in Figs. 5-6, Table 3 and Fig. 7) are calculated for a vertical cut-line in the middle of the channel. Table 2 shows that the energy levels for the radial effective mass cal‐

**E2-E1 E3-E1 E4-E1 E5-E1 E6-E1 E7-E1 E8-E1 E9-E1 E10-E1**

**Silvaco** 0.0406 0.0406 0.094 0.094 0.111 0.158 0.158 0.194 0.194

**Cylmos** 0.0409 0.0409 0.094 0.094 0.112 0.1601 0.1601 0.199 0.199

The wave functions corresponding to the energy levels reported in Table 2 are plotted in Fig. 5 as a function of the position on the radial direction (for *k* from 1 to 6) and in Fig. 6 (for *k* from 7 to 10). The wave functions calculated by Cylmos correspond very well to those from Silvaco, particularly in the region corresponding to the silicon (0 to 5 nm). However, a rather clear difference is observed in the oxide region, due to the fact that Silvaco, contrary to Cy‐ lmos, does not take into account the penetration of wave functions in the oxide. This phe‐ nomenon is included in Cylmos, as previously stated, which leads to this discrepancy

The same analysis can be conducted for the energy levels and wave functions calculated for the second radial effective mass, *mr2*. These results are shown in Table 3 and Fig. 7 for the same transistor and the same polarization as those used above. A very nice agreement is

found again between the data calculated by Cylmos and those obtained from Silvaco.

**Table 2.** Energy levels, *Ek*, calculated by Cylmos and issued from Silvaco for the radial effective mass *mr1* in a circular nanowire of diameter D=10 nm. The transistor is biased at VG=0 V and VD=10 mV. The energy levels are calculated for

culated by Cylmos agree very well with those obtained using Silvaco.

between the wave functions of Cylmos and Silvaco in the gate oxide.

a vertical cut-line in the middle of the channel.

effective mass *mr1* in a nanowire of diameter D=10 nm. (a) *k*=1,2,3; (b) *k*=4,5,6. Other parameters and biases are the same as reported in Table 2. **Figure 5.** Wave function as a function of r calculated by Cylmos and Silvaco for the radial effective mass *mr1* in a nano‐ wire of diameter D=10 nm. (a) *k*=1,2,3; (b) *k*=4,5,6. Other parameters and biases are the same as reported in Table 2.

**Figure 5.** Wave function as a function of r calculated by Cylmos and Silvaco for the radial

Nanowire MOSFETs

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric

17

**Figure 6.** Wave function as a function of *r* calculated by Cylmos and Silvaco for the radial effective mass *mr1* in a nanowire of diameter D=10 nm and k from 7 to 10. Other parameters **Figure 6.** Wave function as a function of *r* calculated by Cylmos and Silvaco for the radial effective mass *mr1* in a nano‐ wire of diameter D=10 nm and k from 7 to 10. Other parameters and biases are the same as reported in Table 2.

and biases are the same as reported in Table 2.


*E2-E1 E3-E1 E4-E1 E5-E1 E6-E1* **Table 3.** Energy levels, *Ek*, calculated by Cylmos and those issued from Silvaco for the radial effective mass *mr2* in a nanowire of diameter D=10 nm. Other parameters and biases are the same as reported in Table 2.

Silvaco 0.064 0.064 0.148 0.148 0.175 Cylmos 0.067 0.067 0.154 0.154 0.183 **Table 3.** Energy levels, *Ek*, calculated by Cylmos and those issued from Silvaco for the radial effective mass *mr2* in a nanowire of diameter D=10 nm. Other parameters and biases are the same as reported in Table 2. In this analysis we studied the energy levels *Ek* and the corresponding wave functions *k*, with *k* varying between 1 and 10. We recall here that the variation of *k* includes the variation of the angular momentum quantum number, *m*. This means that each value of *k* corresponds to a value of *m*, and when *k* varies from 1 to 10, *m* can have the following values: 0, ± 1, ± 2, ... We also recall that the energy levels are sorted in ascending order and In this analysis we studied the energy levels *Ek* and the corresponding wave functions *Ψk*, with *k* varying between 1 and 10. We recall here that the variation of *k* includes the variation of the angular momentum quantum number, *m*. This means that each value of *k* corresponds to a value of *m*, and when *k* varies from 1 to 10, *m* can have the following values: 0, ± 1, ± 2,... We also recall that the energy levels are sorted in ascending order and that only the first *Nlevel* (10 in the present study) lowest levels are stored. The level *k*=1 always corresponds to *m*=0 and all energy levels with non-zero *m* are doubly degenerate, which corresponds to + *m* and-*m* or to the clockwise and counter-clockwise polarization of the wave function. This can be seen, for example in Table 2, where *E2* and *E3* levels are twofold degenerated (as well as the pairs of levels *E4*-*E5*, *E7*-*E8* and *E9*-*E10*).

that only the first *Nlevel* (10 in the present study) lowest levels are stored. The level *k*=1 always corresponds to *m*=0 and all energy levels with non-zero *m* are doubly degenerate, which corresponds to + *m* and -*m* or to the clockwise and counter-clockwise polarization of the wave function. This can be seen, for example in Table 2, where *E2* and *E3* levels are twofold degenerated (as well as the pairs of levels *E4*-*E5*, *E7*-*E8* and *E9*-*E10*). Another remark concerns the variation of the wave functions with the radial position of the axis. For *m*=0 (which corresponds to *k*=1 or *k*=6 for the case studied here), the wave function is always non zero and has zero first derivative at *r*=0; this is valid for both radial effective masses. This remark can be verified in the results presented in Figs. 5 and 7. For *m*=1, the

wave function is always zero and has zero second derivative at *r*=0. For *m*>1, the wave func‐ tion is always zero and has zero first derivative at *r*=0. function is always non zero and has zero first derivative at *r*=0; this is valid for both radial effective masses. This remark can be verified in the results presented in Figs. 5 and 7. For *m*=1, the wave function is always zero and has zero second derivative at *r*=0. For *m*>1, the wave function is always zero and has zero first derivative at *r*=0.

18 Nanowires - Recent Advances

**Figure 7.** Wave functions as a function of *r* calculated by Cylmos and Silvaco for the radial effective mass *mr2* in a nanowire of diameter D=10 nm. (a) *k*=1,2,3; (b) *k*=4,5,6. Other **Figure 7.** Wave functions as a function of *r* calculated by Cylmos and Silvaco for the radial effective mass*mr2* in a nano‐ wire of diameter D=10 nm. (a) *k*=1,2,3; (b) *k*=4,5,6. Other parameters and biases are the same as reported in Table 2.

parameters and biases are the same as reported in Table 2.

#### **3.2. Drain current**

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric

k=7,8

**Figure 6.** Wave function as a function of *r* calculated by Cylmos and Silvaco for the radial effective mass *mr1* in a nanowire of diameter D=10 nm and k from 7 to 10. Other parameters

**Figure 6.** Wave function as a function of *r* calculated by Cylmos and Silvaco for the radial effective mass *mr1* in a nano‐ wire of diameter D=10 nm and k from 7 to 10. Other parameters and biases are the same as reported in Table 2.

0123456

k=9,10

Silvaco Cylmos

Radial effective mass *mr1*

VG=0 V, R=5 nm

r (nm)

**E2-E1 E3-E1 E4-E1 E5-E1 E6-E1**

The same analysis can be conducted for the energy levels and wave functions calculated for the second radial effective mass, *mr2*. These results are shown in Table 3 and Fig. 7 for the same transistor and the same polarization as those used above. A very nice agreement is found again between the data calculated by Cylmos and those obtained from Silvaco.

**Silvaco** 0.064 0.064 0.148 0.148 0.175

**Cylmos** 0.067 0.067 0.154 0.154 0.183

Silvaco 0.064 0.064 0.148 0.148 0.175

In this analysis we studied the energy levels *Ek* and the corresponding wave functions *Ψk*, with *k* varying between 1 and 10. We recall here that the variation of *k* includes the variation of the angular momentum quantum number, *m*. This means that each value of *k* corresponds to a value of *m*, and when *k* varies from 1 to 10, *m* can have the following values: 0, ± 1, ± 2,... We also recall that the energy levels are sorted in ascending order and that only the first *Nlevel* (10 in the present study) lowest levels are stored. The level *k*=1 always corresponds to *m*=0 and all energy levels with non-zero *m* are doubly degenerate, which corresponds to + *m* and-*m* or to the clockwise and counter-clockwise polarization of the wave function. This can be seen, for example in Table 2, where *E2* and *E3* levels are twofold degenerated (as well as

**Table 3.** Energy levels, *Ek*, calculated by Cylmos and those issued from Silvaco for the radial effective mass *mr2* in a

nanowire of diameter D=10 nm. Other parameters and biases are the same as reported in Table 2.

Cylmos 0.067 0.067 0.154 0.154 0.183

**Table 3.** Energy levels, *Ek*, calculated by Cylmos and those issued from Silvaco for the radial effective mass *mr2* in a nanowire of diameter D=10 nm. Other parameters and biases

with *k* varying between 1 and 10. We recall here that the variation of *k* includes the variation of the angular momentum quantum number, *m*. This means that each value of *k* corresponds to a value of *m*, and when *k* varies from 1 to 10, *m* can have the following values: 0, ± 1, ± 2, ... We also recall that the energy levels are sorted in ascending order and that only the first *Nlevel* (10 in the present study) lowest levels are stored. The level *k*=1 always corresponds to *m*=0 and all energy levels with non-zero *m* are doubly degenerate, which corresponds to + *m* and -*m* or to the clockwise and counter-clockwise polarization of the wave function. This can be seen, for example in Table 2, where *E2* and *E3* levels are

Another remark concerns the variation of the wave functions with the radial position of the axis. For *m*=0 (which corresponds to *k*=1 or *k*=6 for the case studied here), the wave function is always non zero and has zero first derivative at *r*=0; this is valid for both radial effective masses. This remark can be verified in the results presented in Figs. 5 and 7. For *m*=1, the

In this analysis we studied the energy levels *Ek* and the corresponding wave functions

twofold degenerated (as well as the pairs of levels *E4*-*E5*, *E7*-*E8* and *E9*-*E10*).

*E2-E1 E3-E1 E4-E1 E5-E1 E6-E1*

and biases are the same as reported in Table 2.

0



3.0×108

4.0×108

1.0×108

2.0×108

are the same as reported in Table 2.

the pairs of levels *E4*-*E5*, *E7*-*E8* and *E9*-*E10*).

17

*k*,

Nanowire MOSFETs

360 Nanowires - Recent Advances

Wave function (nm


We also compare our code Cylmos with Silvaco in terms of drain current. Unfortunately, Silva‐ co does not provide the drain current, because the coupling between the Schrödinger-Poisson system of equations and the continuity equation is not yet implemented. Nevertheless, Silvaco couples the Poisson equation directly to the continuity equation (i.e. the Schrödinger equation is ignored), which makes possible the calculation of the drain current in the so-called "classi‐ cal" case (this means that the quantum confinement effects are not taken into account in this ap‐ proach). To facilitate comparison with Silvaco, we have implemented the same procedure in Cylmos. Thus, in our code it is possible to calculate, in addition to the "quantum" drain current (issued from the self-consistently solving of the Poisson, Schrödinger and continuity equa‐ tions), the "classical" drain current (by directly coupling the Poisson equation to the continuity equation). In this "classical" case, the electron density is reevaluated using the well-known Fermi-Dirac or Boltzmann statistics: A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric Nanowire MOSFETs 19 3.2. Drain current We also compare our code Cylmos with Silvaco in terms of drain current. Unfortunately, Silvaco does not provide the drain current, because the coupling between the Schrödinger-Poisson system of equations and the continuity equation is not yet implemented. Nevertheless, Silvaco couples the Poisson equation directly to the continuity equation (i.e. the Schrödinger equation is ignored), which makes possible the calculation of the drain current in the so-called "classical" case (this means that the quantum confinement effects

$$m(\mathbf{x}, r) = \mathcal{N}\_{\mathbb{C}} \exp\left[ -\mathfrak{J}\_{\mathbb{T}} \left( \mathbb{E}\_{\mathbb{C}}(\mathbf{x}, r) - \Psi\_{\mathbb{F}}(\mathbf{x}, r) \right) \right] \tag{56}$$

where *EC(x,r)* is the minimum of the conduction band and *NC* is the effective 3-D density-ofstates for the conduction band [27]. coupling the Poisson equation to the continuity equation). In this "classical" case, the electron density is reevaluated using the well-known Fermi-Dirac or Boltzmann statistics: *expN)r,x(n* ( *)r,x()r,x(E* ) *<sup>C</sup> CT <sup>F</sup>* (56)

are not taken into account in this approach). To facilitate comparison with Silvaco, we have

Poisson, Schrödinger and continuity equations), the "classical" drain current (by directly

The solving procedure is the same as explained in Section 2. The discretized Poisson equa‐ tion (eq. (8)) is solved taking into account this time no more the electron density issued from the Schrödinger equation, but the "classical" electron density given by eq. (56). The potential is then calculated and is afterward injected directly into the module solving the continuity equation. The output of this module is the new quasi-Fermi level which is subsequently used in eq. (56) to calculate a new carrier density to be used in the Poisson module. In this way, a new potential is found and will be injected into the continuity equation and so on. The loop stops when the convergence criterion is reached. where *EC(x,r)* is the minimum of the conduction band and *NC* is the effective 3-D density-ofstates for the conduction band [27]. The solving procedure is the same as explained in Section 2. The discretized Poisson equation (eq. (8)) is solved taking into account this time no more the electron density issued from the Schrödinger equation, but the "classical" electron density given by eq. (56). The potential is then calculated and is afterward injected directly into the module solving the continuity equation. The output of this module is the new quasi-Fermi level which is subsequently used in eq. (56) to calculate a new carrier density to be used in the Poisson module. In this way, a new potential is found and will be injected into the continuity equation

and so on. The loop stops when the convergence criterion is reached.

**Figure 8.** Drain current versus gate voltage calculated by Cylmos and Silvaco in the "classical" case (i.e., without quantum effects) for a nanowire MOSFET with L=40 nm channel length and a silicon nanowire of diameter D=3 nm. **Figure 8.** Drain current versus gate voltage calculated by Cylmos and Silvaco in the "classical" case (i.e., without quan‐ tum effects) for a nanowire MOSFET with L=40 nm channel length and a silicon nanowire of diameter D=3 nm.

Figure 8 shows the drain current, simulated in this way, for a long channel MOSFET (L=40 nm) with a circular nanowire with a diameter D=3 nm. The drain current is plotted as a function of the gate voltage for VD=50 mV. This figure illustrates that a good agreement is obtained be‐ tween the results of Cylmos ("classical" drain current) and those provided by Silvaco. Figure 8 shows the drain current, simulated in this way, for a long channel MOSFET (L=40 nm) with a circular nanowire with a diameter D=3 nm. The drain current is plotted as a function of the gate voltage for VD=50 mV. This figure illustrates that a good agreement is obtained between the results of Cylmos ("classical" drain current) and those provided by

20 Nanowires - Recent Advances

couples the Poisson equation directly to the continuity equation (i.e. the Schrödinger equation is ignored), which makes possible the calculation of the drain current in the so-called "classi‐ cal" case (this means that the quantum confinement effects are not taken into account in this ap‐ proach). To facilitate comparison with Silvaco, we have implemented the same procedure in Cylmos. Thus, in our code it is possible to calculate, in addition to the "quantum" drain current (issued from the self-consistently solving of the Poisson, Schrödinger and continuity equa‐ tions), the "classical" drain current (by directly coupling the Poisson equation to the continuity equation). In this "classical" case, the electron density is reevaluated using the well-known

We also compare our code Cylmos with Silvaco in terms of drain current. Unfortunately, Silvaco does not provide the drain current, because the coupling between the Schrödinger-Poisson system of equations and the continuity equation is not yet implemented. Nevertheless, Silvaco couples the Poisson equation directly to the continuity equation (i.e. the Schrödinger equation is ignored), which makes possible the calculation of the drain current in the so-called "classical" case (this means that the quantum confinement effects are not taken into account in this approach). To facilitate comparison with Silvaco, we have implemented the same procedure in Cylmos. Thus, in our code it is possible to calculate, in addition to the "quantum" drain current (issued from the self-consistently solving of the Poisson, Schrödinger and continuity equations), the "classical" drain current (by directly coupling the Poisson equation to the continuity equation). In this "classical" case, the electron density is reevaluated using the well-known Fermi-Dirac or Boltzmann statistics:

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric

where *EC(x,r)* is the minimum of the conduction band and *NC* is the effective 3-D density-of-

The solving procedure is the same as explained in Section 2. The discretized Poisson equa‐ tion (eq. (8)) is solved taking into account this time no more the electron density issued from the Schrödinger equation, but the "classical" electron density given by eq. (56). The potential is then calculated and is afterward injected directly into the module solving the continuity equation. The output of this module is the new quasi-Fermi level which is subsequently used in eq. (56) to calculate a new carrier density to be used in the Poisson module. In this way, a new potential is found and will be injected into the continuity equation and so on.

where *EC(x,r)* is the minimum of the conduction band and *NC* is the effective 3-D density-of-

The solving procedure is the same as explained in Section 2. The discretized Poisson equation (eq. (8)) is solved taking into account this time no more the electron density issued from the Schrödinger equation, but the "classical" electron density given by eq. (56). The potential is then calculated and is afterward injected directly into the module solving the continuity equation. The output of this module is the new quasi-Fermi level which is subsequently used in eq. (56) to calculate a new carrier density to be used in the Poisson module. In this way, a new potential is found and will be injected into the continuity equation

**Figure 8.** Drain current versus gate voltage calculated by Cylmos and Silvaco in the "classical" case (i.e., without quantum effects) for a nanowire MOSFET with L=40 nm

**Figure 8.** Drain current versus gate voltage calculated by Cylmos and Silvaco in the "classical" case (i.e., without quan‐ tum effects) for a nanowire MOSFET with L=40 nm channel length and a silicon nanowire of diameter D=3 nm.

0 0.2 0.4 0.6 0.8 1

Gate voltage VG (V)

*CT*

*expN)r,x(n* ( *)r,x()r,x(E* ) *<sup>C</sup>* 

*nxr N E xr xr* ( , ) exp ( , ) ( , ) = -b - Y *C TC F* é ù ( ) ë û (56)

*<sup>F</sup>* (56)

Drain current ID

 (A)

10-4

10-6

10-8

10-10

10-12

10-14

10-16

10-18

19

Silvaco.

L=D=8 nm.

Fermi-Dirac or Boltzmann statistics:

Nanowire MOSFETs

362 Nanowires - Recent Advances

3.2. Drain current

states for the conduction band [27].

states for the conduction band [27].

Drain current ID

2.0×10-6

0

4.0×10-6 6.0×10-6 8.0×10-6 1.0×10-5

1.2×10-5 1.4×10-5 1.6×10-5

 (A)

The loop stops when the convergence criterion is reached.

channel length and a silicon nanowire of diameter D=3 nm.

and so on. The loop stops when the convergence criterion is reached.

L=40 nm D=3 nm VD=50 mV

Silvaco Cylmos

**Figure 9.** Drain current versus gate voltage calculated by Cylmos and Silvaco in the "classical" case (i.e. without quantum effects) for a short-channel nanowire MOSFET with **Figure 9.** Drain current versus gate voltage calculated by Cylmos and Silvaco in the "classical" case (i.e. without quan‐ tum effects) for a short-channel nanowire MOSFET with L=D=8 nm.

We also tested a short-channel transistor, for which the gate length is equal to the diameter of the nanowire (L=D=8 nm). This transistor is probably provided with quite important shortchannel electrostatic effects and it is necessary to verify that our code correctly reproduces We also tested a short-channel transistor, for which the gate length is equal to the diameter of the nanowire (L=D=8 nm). This transistor is probably provided with quite important shortchannel electrostatic effects and it is necessary to verify that our code correctly reproduces these effects. In addition, the short-channel effects are considerably enhanced when the drain voltage is high, therefore it is important to test the current characteristics for several VD. Fig‐ ure 9 shows such a verification on both characteristics plotted in linear scale (for a better view on the regime above the threshold) and in logarithmic scale (to visualize the subthreshold re‐ gime). These figures show once again an excellent agreement between Cylmos and Silvaco, which means that the electrostatic effects are properly taken into account in Cylmos.

## **4. Simulation results**

After validating Cylmos by comparison with a commercial simulator, we use in the follow‐ ing our code to analyze the operation and performances of circular nanowire-based MOS‐ FETs subjected to quantum confinement. We will particularly focus on the variation of the drain current as a function of the nanowire diameter and of the gate length and we will dis‐ cuss the device performances in terms of off-state current, threshold voltage and parasitic short-channel effects. These key parameters will also be compared between the classical and the quantum-mechanical case.

#### **4.1. Drain current characteristics**

We start with the variation of the drain current as a function of the nanowire diameter. Fig‐ ure 10(a) plots the drain current characteristics as a function of the gate voltage, as calculat‐ ed by Cylmos, for two diameters D=5 nm and D=8 nm. Data obtained in both classical and quantum cases are shown in this figure. A first remark concerns the off-state current (i.e., value of the drain current for VG=0 V): for D=5 nm this current is lower than for D=8 nm, whatever the computation method, classical or quantum. Similarly, the current in the sub‐ threshold regime is lower for the smallest diameter. This is due to the so-called "volume in‐ version" phenomenon which takes place in all multiple-gate devices in the subthreshold regime [28]-[29]. Thus, we logically find this phenomenon here in the circular nanowire, for which the gate completely surrounds the silicon film. Because of the volume inversion, the subthreshold current is directly proportional to the nanowire diameter, which implies that the off-state current decreases when the nanowire is thinned.

A second remark concerns the reduction of the drain current in the quantum case compared to the classical case, for both nanowire diameters considered here. As we explained earlier in this chapter, the existence of a strong quantum confinement of carriers leads to the reduction of the charge inversion in the channel (compared to the classic case where this effect is not taken into account). Therefore, for the same gate voltage VG, the quantum drain current is necessarily smaller than the classical drain current, and this for all nanowire diameters. This behavior is similar to that of all other multiple-gate devices where strong quantum confinement effects (1- D or 2-D) occur [30], mainly due to the geometrical construction of these devices.

Another point concerns the difference between the current calculated in the classical case and the current calculated in the quantum case. This difference is larger for D=5 nm than for D=8 nm, as can be seen in Fig. 10(a), particularly in the subthreshold regime. The explanation comes from the large variation of quantum confinement effects with the nanowire diameter. When the nano‐ wire is thinned, the carrier confinement is more pronounced (which also implies an increase of the allowed energy levels) and the inversion charge density in the channel reduces. This increas‐ es strongly the threshold voltage and leads to a significant decrease of the drain current.

We also tested a short-channel transistor, for which the gate length is equal to the diameter of the nanowire (L=D=8 nm). This transistor is probably provided with quite important shortchannel electrostatic effects and it is necessary to verify that our code correctly reproduces these effects. In addition, the short-channel effects are considerably enhanced when the drain voltage is high, therefore it is important to test the current characteristics for several VD. Fig‐ ure 9 shows such a verification on both characteristics plotted in linear scale (for a better view on the regime above the threshold) and in logarithmic scale (to visualize the subthreshold re‐ gime). These figures show once again an excellent agreement between Cylmos and Silvaco,

which means that the electrostatic effects are properly taken into account in Cylmos.

After validating Cylmos by comparison with a commercial simulator, we use in the follow‐ ing our code to analyze the operation and performances of circular nanowire-based MOS‐ FETs subjected to quantum confinement. We will particularly focus on the variation of the drain current as a function of the nanowire diameter and of the gate length and we will dis‐ cuss the device performances in terms of off-state current, threshold voltage and parasitic short-channel effects. These key parameters will also be compared between the classical and

We start with the variation of the drain current as a function of the nanowire diameter. Fig‐ ure 10(a) plots the drain current characteristics as a function of the gate voltage, as calculat‐ ed by Cylmos, for two diameters D=5 nm and D=8 nm. Data obtained in both classical and quantum cases are shown in this figure. A first remark concerns the off-state current (i.e., value of the drain current for VG=0 V): for D=5 nm this current is lower than for D=8 nm, whatever the computation method, classical or quantum. Similarly, the current in the sub‐ threshold regime is lower for the smallest diameter. This is due to the so-called "volume in‐ version" phenomenon which takes place in all multiple-gate devices in the subthreshold regime [28]-[29]. Thus, we logically find this phenomenon here in the circular nanowire, for which the gate completely surrounds the silicon film. Because of the volume inversion, the subthreshold current is directly proportional to the nanowire diameter, which implies that

A second remark concerns the reduction of the drain current in the quantum case compared to the classical case, for both nanowire diameters considered here. As we explained earlier in this chapter, the existence of a strong quantum confinement of carriers leads to the reduction of the charge inversion in the channel (compared to the classic case where this effect is not taken into account). Therefore, for the same gate voltage VG, the quantum drain current is necessarily smaller than the classical drain current, and this for all nanowire diameters. This behavior is similar to that of all other multiple-gate devices where strong quantum confinement effects (1-

D or 2-D) occur [30], mainly due to the geometrical construction of these devices.

**4. Simulation results**

364 Nanowires - Recent Advances

the quantum-mechanical case.

**4.1. Drain current characteristics**

the off-state current decreases when the nanowire is thinned.

**Figure 10.** Drain current characteristics versus the gate voltage calculated by Cylmos: (a) L=40 nm and two nanowire diameters D=8 nm and D=5 nm; (b) D=5 nm and two channel lengths L=40 nm and L=5 nm. The drain current is calcu‐ lated in both quantum and classical cases.

Finally, Fig. 10(b) shows the variation of the drain current as a function of the gate length for D=5 nm. In this figure we can observe that when the gate length is reduced (for the same nanowire diameter) the off-state current strongly increases and the subthreshold slope is much higher than in the case of a long channel (L=40 nm), reflecting a significant degradation of the device performances. This is due to the huge increase of short-channel effects, because in this ultra-short channel having a gate length equal to the nanowire di‐ ameter, the gate control on the electrostatic potential of the channel is much less effective. Thus, parasitic electrostatic effects dramatically increase and lead to this large increase in off-state current and subthreshold slope. This behavior appears for both classical or quan‐ tum cases.

#### **4.2. Threshold voltage and short channel effects**

For a more detailed analysis of these parasitic short-channel effects, we will now focus on the variations of the threshold voltage as a function of the gate length and of the nano‐ wire diameter. The threshold voltage is extracted using the constant current method [31] from the drain current characteristics as a function of the gate voltage. In Fig. 11 we plot the variation of this threshold voltage as function of the nanowire diameter in the case of a long-channel transistor and in the case of a transistor with a channel length equal to the nanowire diameter (L=D). The results reported in this figure come from a quantum-me‐ chanical calculation. We note that the threshold voltage of the long-channel transistor is higher than that of the short-channel transistor with L=D. This is an expected result be‐ cause the short channel effects that occur in the transistor with L=D greatly reduce the threshold voltage. One can also note that when the diameter of the nanowire is reduced, the threshold voltage remains constant over a certain range of values and next begins to increase with decreasing the nanowire diameter (in both cases, long channel and short channel with L=D). This increase is due, of course, to the fact that the quantum confine‐ ment of carriers is more significant when the diameter is reduced, which induces an in‐ crease in the threshold voltage (as explained earlier). This increase is higher in the case of long channel than in the case of a short-channel. This indicates that the strong short-chan‐ nel effects that exist in the channel with L=D affect the quantum confinement, which be‐ comes weaker than in a long-channel transistor.

Another way to highlight this phenomenon is to calculate the impact of short channel ef‐ fects, reflected in the SCE metric, which is equal to the difference between the threshold voltage of the short-channel transistor with L=D and the threshold voltage of the long-chan‐ nel transistor (for a given diameter D). SCE is then plotted as a function of the nanowire di‐ ameter in Fig. 12. This figure shows that, after an increase and a maximum at D=8 nm, the SCE starts to decrease as the diameter reduces below 8 nm. This behavior is explained by the fact that the strong quantum confinement that occurs in these ultrafine nanowires reduces short-channel effects, which will probably lead to increased performance of these ultra-short and ultra-thin devices.

A 2-D/3-D Schrödinger-Poisson Drift-Diffusion Numerical Simulation of Radially-Symmetric Nanowire MOSFETs http://dx.doi.org/10.5772/52587 367

Finally, Fig. 10(b) shows the variation of the drain current as a function of the gate length for D=5 nm. In this figure we can observe that when the gate length is reduced (for the same nanowire diameter) the off-state current strongly increases and the subthreshold slope is much higher than in the case of a long channel (L=40 nm), reflecting a significant degradation of the device performances. This is due to the huge increase of short-channel effects, because in this ultra-short channel having a gate length equal to the nanowire di‐ ameter, the gate control on the electrostatic potential of the channel is much less effective. Thus, parasitic electrostatic effects dramatically increase and lead to this large increase in off-state current and subthreshold slope. This behavior appears for both classical or quan‐

For a more detailed analysis of these parasitic short-channel effects, we will now focus on the variations of the threshold voltage as a function of the gate length and of the nano‐ wire diameter. The threshold voltage is extracted using the constant current method [31] from the drain current characteristics as a function of the gate voltage. In Fig. 11 we plot the variation of this threshold voltage as function of the nanowire diameter in the case of a long-channel transistor and in the case of a transistor with a channel length equal to the nanowire diameter (L=D). The results reported in this figure come from a quantum-me‐ chanical calculation. We note that the threshold voltage of the long-channel transistor is higher than that of the short-channel transistor with L=D. This is an expected result be‐ cause the short channel effects that occur in the transistor with L=D greatly reduce the threshold voltage. One can also note that when the diameter of the nanowire is reduced, the threshold voltage remains constant over a certain range of values and next begins to increase with decreasing the nanowire diameter (in both cases, long channel and short channel with L=D). This increase is due, of course, to the fact that the quantum confine‐ ment of carriers is more significant when the diameter is reduced, which induces an in‐ crease in the threshold voltage (as explained earlier). This increase is higher in the case of long channel than in the case of a short-channel. This indicates that the strong short-chan‐ nel effects that exist in the channel with L=D affect the quantum confinement, which be‐

Another way to highlight this phenomenon is to calculate the impact of short channel ef‐ fects, reflected in the SCE metric, which is equal to the difference between the threshold voltage of the short-channel transistor with L=D and the threshold voltage of the long-chan‐ nel transistor (for a given diameter D). SCE is then plotted as a function of the nanowire di‐ ameter in Fig. 12. This figure shows that, after an increase and a maximum at D=8 nm, the SCE starts to decrease as the diameter reduces below 8 nm. This behavior is explained by the fact that the strong quantum confinement that occurs in these ultrafine nanowires reduces short-channel effects, which will probably lead to increased performance of these ultra-short

tum cases.

366 Nanowires - Recent Advances

**4.2. Threshold voltage and short channel effects**

comes weaker than in a long-channel transistor.

and ultra-thin devices.

**Figure 11.** Threshold voltage versus the nanowire diameter extracted from quantum calculation of the drain current characteristics.

**Figure 12.** SCE metric versus the nanowire diameter for short-channel devices with L=D.

### **5. Conclusion**

In conclusion, we described in this chapter a numerical solver, called Cylmos, that provides self-consistent solving of the 2-D Schrödinger equation and the 3-D Poisson equation in cy‐ lindrical coordinates, coupled with the drift-diffusion transport equation. We presented in details the discretization of these equations and the overall solving algorithm used to obtain the final drain current of the device. We also paid a particular attention to justify the approx‐ imations and simplifications used in the solving of these equations in cylindrical coordi‐ nates. A meticulous validation step was performed by comparing the results of our code with those issued from a commercial code. We demonstrated that our code properly takes into account particular phenomena specific to MOSFET devices based on circular nanowires with surrounding gate, such as volume inversion, short channel and quantum-mechanical effects. Simulated drain current versus gate voltage characteristics have also been success‐ fully compared to data obtained from simulation with this commercial code. Our code was finally used to analyze electrical parameters such as drain current, off-state current, thresh‐ old voltage and short-channel effects in ultra-short radially-symmetric nanowire MOSFETs. We have shown that quantum-mechanical confinement is very important in circular nano‐ wires, but the presence of this phenomenon reduces the impact of parasitic short-channel ef‐ fects. This is a very encouraging result for the operation of future integrated circuits based on ultra-short and ultra-thin circular nanowires.

## **Author details**

Daniela Munteanu and Jean-Luc Autran

CNRS &Aix-Marseille University, Marseille, France

### **References**


[5] A. Lochtefeld and D. A. Antoniadis, "On Experimental Determination of Carrier Ve‐ locity in Deeply Scaled NMOS: How Close to the Thermal Limit?", IEEE Electron De‐ vice Lett., vol. 22, pp. 95–97, 2001.

**5. Conclusion**

368 Nanowires - Recent Advances

**Author details**

**References**

2006.

on ultra-short and ultra-thin circular nanowires.

CNRS &Aix-Marseille University, Marseille, France

regime", Proc. IEEE, vol. 85, pp. 486–504, 1997.

op., vol. 50, no. 4/5, pp. 387–410, 2006.

Daniela Munteanu and Jean-Luc Autran

http://public.itrs.net.

In conclusion, we described in this chapter a numerical solver, called Cylmos, that provides self-consistent solving of the 2-D Schrödinger equation and the 3-D Poisson equation in cy‐ lindrical coordinates, coupled with the drift-diffusion transport equation. We presented in details the discretization of these equations and the overall solving algorithm used to obtain the final drain current of the device. We also paid a particular attention to justify the approx‐ imations and simplifications used in the solving of these equations in cylindrical coordi‐ nates. A meticulous validation step was performed by comparing the results of our code with those issued from a commercial code. We demonstrated that our code properly takes into account particular phenomena specific to MOSFET devices based on circular nanowires with surrounding gate, such as volume inversion, short channel and quantum-mechanical effects. Simulated drain current versus gate voltage characteristics have also been success‐ fully compared to data obtained from simulation with this commercial code. Our code was finally used to analyze electrical parameters such as drain current, off-state current, thresh‐ old voltage and short-channel effects in ultra-short radially-symmetric nanowire MOSFETs. We have shown that quantum-mechanical confinement is very important in circular nano‐ wires, but the presence of this phenomenon reduces the impact of parasitic short-channel ef‐ fects. This is a very encouraging result for the operation of future integrated circuits based

[1] 2012 International Technology Roadmap for Semiconductors. Available online:

[2] T. Hiramoto, M. Saitoh, G. Tsutsui, "Emerging nanoscale Silicon devices taking ad‐ vantage of nanostructure physics", IBM J. Res.Develop., vol. 50, no. 4/5, pp. 411-418,

[3] Y. Taur, D. Buchanan, W. Chen, D. Frank, K. Ismail, S.-H. Lo, G. Sai-Halasz, R. Vis‐ wanathan, H.-J. C. Wann, S. Wind, H.-S. Wong, "CMOS scaling into the nanometer

[4] E. P. Gusev, V. Narayanan, M. M. Frank, "Advanced high-k dielectric stacks with polySi and metal gates: Recent progress and current challenges", IBM J. Res. Devel‐


## **Recent Advances in Mechanical Properties of Nanowires**

Hui Li and Fengwei Sun

[18] J. L. Autran and D. Munteanu, "Simulation of electron transport in nanoscale inde‐ pendent-gate DG devices using a full 2D Green's function approach", Journal of

[19] J. Martorell, H. Wu, and D. W. L. Sprung, "Systematic trends in self-consistent calcu‐ lations of linear quantum wires", Phys. Rev. B, vol. 50, no. 23, pp. 17298-17308, 1994.

[20] T. Kerkhoven, A.T.Galick, U. Ravaioli, J.H. Arends, Y. Saad,"Efficient numerical sim‐ ulation of electron states in quantum wires", Journal of Applied Physics, vol. 68, no.

[21] G. L. Snider, I. H. Tan, and E. L. Hu, "Electron states in mesa‐etched one‐dimensional quantum well wires", Journal of Applied Physics, vol. 68, no. 6, pp. 2849 – 2853, 1990.

[22] T. Kerkhoven, M. Raschke, and U. Ravaioli, "Self‐consistent simulation of quantum wires in periodic heterojunction structures", Journal of Applied Physics,vol. 74, no. 2,

[23] Y. Taur and T. H. Ning, Fundamentals of Modern VLSI Devices, Cambridge Univer‐

[25] C. Raynaud, J.-L. Autran, P. Masson, M. Bidaud and A. Poncet,"Analysis of MOS de‐ vices capacitance-voltage characteristics based on the self-consistent solution of the Schröndinger and Poisson equation", Mat. Res. Soc. Symp. Proceedings, vol. 592, pp.

[26] J. S. Blakemore, "Approximations for Fermi–Dirac integrals, especially the function F1/2(ηF ) used to describe electron density in a semiconductor," Solid State Electron‐

[27] M. Moreau, D. Munteanu and J.L. Autran, "Simulation Analysis of Quantum Con‐ finement and Short-Channel Effects in Independent Double-Gate Metal-Oxide-Semi‐ conductor Field-Effect Transistors ", Japanese Journal of Applied Physics, vol.47, no.

[28] Y. Taur, X. Liang, W. Wang, H. Lu, " A continuous, analytic Drain-Current model for DG MOSFETs", IEEE Electron Device Letters, vol. 25, no. 2, pp. 107-109, 2004.

[29] D. Munteanu, J.L. Autran, X. Loussier, S. Harrison, R. Cerutti, T. Skotnicki, "Quan‐ tum Short-Channel Compact Modeling of Drain-Current in Double-Gate MOSFET",

[30] D. Munteanu, J.L. Autran, V. Ferlet-Cavrois, P. Paillet, J. Baggio, K. Castellani, "3D Quan‐ tum Numerical Simulation of Single-Event Transients in Multiple-Gate Nanowire MOS‐

[31] D. Munteanu, J.L. Autran, S. Harrison, K. Nehari, O. Tintori, T. Skotnicki, "Compact Model of the Quantum Short-Channel Threshold Voltage in Symmetric Double-Gate

FETs", IEEE Transactions on Nuclear Science, vol. 54, no. 4, pp. 994-1001, 2007.

MOSFET", Molecular Simulation, vol. 31, no. 12, pp. 831–837, 2005.

[24] Atlas User's Manual, Silvaco International, Santa Clara, CA, 2010.

Computational and Theoretical Nanoscience, vol. 5, pp. 1120–1127, 2008.

7,pp. 3461-3469,1990.

370 Nanowires - Recent Advances

pp. 1199 – 1204,1993.

9, pp. 7013-7018, 2008.

159, 2000.

sity Press, New York, 1998, p. 25.

ics, vol. 25, no. 11, pp. 1067–1076, Nov. 1982.

Solid State Electronics, vol. 50, no. 4, pp. 630-636, 2006.

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/ 52592

## **1. Introduction**

Over the recent decade, nanowires (NWs) have attracted great attention due to their small size and thus large surface area to volume ratio which results in interesting surface effect. As the devices become smaller and smaller, NWs serve as basic building blocks for future elec‐ tronic and electromechanical systems, such as performing sensitive mass and force detec‐ tion, acting as high frequency resonators and so on. The potential of NWs in future application has led to significant interest in experimental and theoretical characterization of the size-dependent properties of NWs. Constructing and determining NWs with better me‐ chanical properties than the corresponding bulk materials are a challenge even though the technical equipments are improving. Meantime, another important thing is how to integrate NWs with the existing technology. Material simulations of investigating mechanical proper‐ ties of NWs are still an important technique until now. This is essential to understand the inherent mechanism of the NWs' deformation. The deformation mechanism depends on several aspects including the intrinsic material properties, crystal structures, surface geome‐ try, applied stress state, axial and surface orientation and exposed transverse surfaces. Until now many experimental techniques have been utilized to investigate mechanical properties. Scanning tunneling microscope (STM) was employed to study atomic contact of NWs and their conductance [1-3] in early experiments. Since a novel nanobridge structure of Au NWs is generated by electron-beam irradiation on a thin Au film in an ultrahigh vacuum trans‐ mission electron microscope (TEM) [4], many researches have been carried out to study the properties of this stable structure. A specific developed STM supplemented with a force sen‐ sor is used in order to reveal the process of forming thin Au NWs [5]. Atomic force micro‐ scope(AFM) has also attracted particular attention due to its high spatial resolution and force-sensing capabilities[6] and was used for bending tests of single crystal, micromachined

© 2012 Li and Sun; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Li and Sun; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

silicon beams[7]. Recently developed in situ TEM technique can also be applied to perform bending deformation of single SiC NWs[8]. Theoretically, some new simulation methods are developed and widely used to study nanoscale materials. For FCC crystal metal NW, most simulations were performed with embedded atom method (EAM) [9-17], which reproduce exactly the experimental second-order and third-order elastic moduli as well as the phase stability. Furthermore, the modified EAM with the developed potential [18-20] is utilized to study the phase transformation of Au NWs. While for the semiconductor NWs, Tersoff many-body bond-order reactive potential and Stillinger-Weber many-body potential [21, 22] are appropriate to describe interatomic interactions[23] in Si and SiC NW. Tight-binding molecular dynamics simulations [24-26], which lies in between first principles and empirical methods is also used to study the atomic structure of Au NW. It is more accurate than em‐ pirical potential methods because it explicitly includes the electronic structure, and is much faster than first principles methods.

This review is organized as follows. In section 2, we discuss some novel mechanical proper‐ ties of single crystal metal NWs caused by surface effect, strain rate, temperature and so on. In section 3, we consider the mechanical properties of semiconductor NWs such as Si, Si compound and ZnO et. al. In section 4, some other irregular structures of NWs are investi‐ gated to compare the mechanical properties with bulk and normal crystal NWs. In the last section, we discuss the enhancement of NWs which are encapsulated in carbon nanotubes during compression.

## **2. Single crystal metal nanowires**

#### **2.1. Side surface effect on axial deformation mode**

Several researchers have demonstrated that the structures and properties of nanomaterials can be quite different from those of bulk materials by experiments or molecular dynamics (MD) simulations due to surface effect which results in different deformation modes.

In compression of the <110> NW, low energy {111} side surfaces prevented free-surface initi‐ ated slip. On the contrary, slip is observed for higher energy {100} and {110} side surfaces. For the <100> wires, distributed stacking faults appearing in the <100>/{110} wires result from simple and important rotation of the crystal orientation along its longitudinal axis, which is different from the deformation mode of twinning for the <100>/{100} wires. Some other important findings of fundamental deformation mechanism are revealed in Ref.[27].

#### **2.2. Size effect on Young's modulus**

Mechanical properties based on Au NW bending under the lateral load from an AFM tip are performed by Wu et.al [6]. Their experiments reveals that Young's modulus is essentially in‐ dependent of diameter, whereas the yield strength is largest for the smallest diameter wires, with strengths up to 100 times that of bulk materials, and substantially larger than that re‐ ported for bulk nanocrystalline metals. MD simulations demonstrate that Young's modulus is independent of strain rate and cross sectional size of specimen. Neither geometry nor sur‐ face structure has an appreciable effect on the modulus[28]. However, in some other molec‐ ular statics simulations[29], the modulus values for NWs with a <110> axial orientation are observed to increase with decreasing cross-sectional thickness, while for NWs with a <100> axial orientation the modulus values decrease with decreasing cross-sectional thickness. This is attributed to the nonlinear elasticity of the NW core and dissimilar surface and core elastic modulus values[12]. The size-dependence modulus[28] is consistent with predictions based on core nonlinear elasticity, but it is difficult to separate the surface elasticity and core nonlinearity effects because of a lack of strong trends with NW size or surface orienta‐ tion[29]. Meantime, deformation induced by surface stresses might also has a significant im‐ pact on the Young's modulus in bending simulations, where a strong boundary-condition dependence is also found [30]. Therefore, a nonlinear, finite deformation formulation [30] that captures changes in both bulk and surface elastic properties resulting from surface stress-induced deformation is critical to reproducing the experimentally observed boun‐ dary-condition dependence in Young's modulus of metal NWs.

#### **2.3. Surface stress effect**

silicon beams[7]. Recently developed in situ TEM technique can also be applied to perform bending deformation of single SiC NWs[8]. Theoretically, some new simulation methods are developed and widely used to study nanoscale materials. For FCC crystal metal NW, most simulations were performed with embedded atom method (EAM) [9-17], which reproduce exactly the experimental second-order and third-order elastic moduli as well as the phase stability. Furthermore, the modified EAM with the developed potential [18-20] is utilized to study the phase transformation of Au NWs. While for the semiconductor NWs, Tersoff many-body bond-order reactive potential and Stillinger-Weber many-body potential [21, 22] are appropriate to describe interatomic interactions[23] in Si and SiC NW. Tight-binding molecular dynamics simulations [24-26], which lies in between first principles and empirical methods is also used to study the atomic structure of Au NW. It is more accurate than em‐ pirical potential methods because it explicitly includes the electronic structure, and is much

This review is organized as follows. In section 2, we discuss some novel mechanical proper‐ ties of single crystal metal NWs caused by surface effect, strain rate, temperature and so on. In section 3, we consider the mechanical properties of semiconductor NWs such as Si, Si compound and ZnO et. al. In section 4, some other irregular structures of NWs are investi‐ gated to compare the mechanical properties with bulk and normal crystal NWs. In the last section, we discuss the enhancement of NWs which are encapsulated in carbon nanotubes

Several researchers have demonstrated that the structures and properties of nanomaterials can be quite different from those of bulk materials by experiments or molecular dynamics

In compression of the <110> NW, low energy {111} side surfaces prevented free-surface initi‐ ated slip. On the contrary, slip is observed for higher energy {100} and {110} side surfaces. For the <100> wires, distributed stacking faults appearing in the <100>/{110} wires result from simple and important rotation of the crystal orientation along its longitudinal axis, which is different from the deformation mode of twinning for the <100>/{100} wires. Some other important findings of fundamental deformation mechanism are revealed in Ref.[27].

Mechanical properties based on Au NW bending under the lateral load from an AFM tip are performed by Wu et.al [6]. Their experiments reveals that Young's modulus is essentially in‐ dependent of diameter, whereas the yield strength is largest for the smallest diameter wires, with strengths up to 100 times that of bulk materials, and substantially larger than that re‐ ported for bulk nanocrystalline metals. MD simulations demonstrate that Young's modulus

(MD) simulations due to surface effect which results in different deformation modes.

faster than first principles methods.

**2. Single crystal metal nanowires**

**2.2. Size effect on Young's modulus**

**2.1. Side surface effect on axial deformation mode**

during compression.

372 Nanowires - Recent Advances

#### *2.3.1. Surface stress effects on the resonant properties*

Boundary conditions determine how the NWs deform in response to surface stresses. Fixed/ free NWs (cantilever) are able to contract, thus relieving the tensile surface stresses, and leading to a state of compression in the NW core. Fixed/fixed NWs are constrained such that they cannot deform axially, causing them to exist in a state of tensile stress.

The reduction in resonant frequencies for the fixed/free boundary condition and increase in resonant frequencies for the fixed/fixed boundary condition [31] indicate that those NWs are expected to be elastically softer, and stiffer, respectively, than the corresponding bulk mate‐ rial, agreeing with recent experimental results for both free-standing[32], and fixed/fixed NWs [33-35]. The variation in NW resonant frequencies due to surface stresses was found to be dependent on the NW aspect ratio. In contrast, no dependence of the resonant frequen‐ cies on the surface area to volume ratio was found, again agreeing with recent experimental data [36]. The variation in resonant frequency for the fixed/free case occurs due to the fact the bulk <100> material softens under compression. Therefore, the resonant frequencies of fixed/free NWs with other orientations that stiffen under compression, such as <110>, are ex‐ pected to increase rather than decrease as compared to the corresponding bulk material. In the higher order resonant modes (second bending, stretch, twist), surface stresses cause the largest variation in the second bending frequencies, while the twist and stretch frequencies show little variation due to surface stresses. All higher order resonant frequencies decrease with respect to the fundamental bending frequency with increasing aspect ratio for fixed/ fixed NWs due to surface stresses, while the higher order resonant frequencies increase with respect to the fundamental bending frequency for fixed/free NWs[31].

#### *2.3.2. Surface stress induced phase transformation*

Surface-stress-induced phase transformation in Au NWs is an interesting phenomenon. The emergence of the transformation is dominated by wire size, initial orientation, boundary conditions, temperature and initial cross sectional shape. For a crystal with <100> axial ori‐ entation, surface stresses alone cause Au NWs with cross-sectional area below 4 nm2 to transform from a FCC structure to a body centered-tetragonal (BCT) structure [20]. Not only has this transformation appeared in Au NWs, but also in alloy NWs. Park has demonstrated that B2 NiAl NWs undergo a stress-induced martensitic phase transformation to a BCT phase as well by the propagation and annihilation of {101} twinning planes [37]. While for Au NWs formed along the [110] direction, quantum mechanical molecular dynamics shows that it reconstructs upon stress to form helical NWs [38]. The different final structures are probably due to temperature effect as well. The helical NW is obtained by annealing the crystal structure from 600K, while the BCT structure is relaxed to a minimum energy state at 0K with conjugate gradient method.

Some further simulations have also been performed [39] to investigate the core/shell struc‐ ture in which the crystalline core is encapsulated by a glassy shell. Such core/shell hetero‐ structures usually possess particularly attractive applications in the nanoelectromechanical system, owing to their high complexity and functionality. The elongation of the metallic NW is realized by moving the rigid atoms at one end along the axial axis. The rigid atoms are stretched 0.001nm every 1 ps. This strain rate is sufficient to dissipate excess energy to main‐ tain a constant temperature. The Berendsen method [40] is used to control the system tem‐ perature. The motion equations are solved by a fourth-order Nordsieck-Gear predictor corrector method[41] using a time step of 1 fs. The average of the virial stress component along the axial axis during deformation is taken as the macroscopic stress of the wire. This stress measure totally differs from either engineering stress or true stress, which has been used widely in molecular simulations [42-44]. Here the low temperature (T=50K) was used to highlight the material responses upon mechanical activation, as the simulated quenching glass structure after rapid cooling is usually more prone to thermal activations than the ex‐ perimental one. Continuous deformation with noticeable necking occurs to the core/shell NW instead of discrete displacement as shown in Fig.1.

The core/shell structure, which can even be spun into a single atomic chain, tends to display much larger plasticity than the crystalline one. Most importantly, during elongation, local amorphization happens in the high strained region. Generally, it is thought that the amorph‐ ization behavior in nanoscale materials is induced by the high strain rate or the entropy at a high temperature. In terms of our simulation, the strain rate of 0.006% ps−1 for a 16.3nm long NW, which is equivalent to an approximate stretch velocity of 1.0ms−1, belongs to the range of low stretch velocity compared with previous simulation results. Therefore, we suggest that the surface amorphous layer would induce the disorder of the crystalline core during the elongating process.

*2.3.2. Surface stress induced phase transformation*

374 Nanowires - Recent Advances

0K with conjugate gradient method.

NW instead of discrete displacement as shown in Fig.1.

the elongating process.

Surface-stress-induced phase transformation in Au NWs is an interesting phenomenon. The emergence of the transformation is dominated by wire size, initial orientation, boundary conditions, temperature and initial cross sectional shape. For a crystal with <100> axial ori‐ entation, surface stresses alone cause Au NWs with cross-sectional area below 4 nm2

transform from a FCC structure to a body centered-tetragonal (BCT) structure [20]. Not only has this transformation appeared in Au NWs, but also in alloy NWs. Park has demonstrated that B2 NiAl NWs undergo a stress-induced martensitic phase transformation to a BCT phase as well by the propagation and annihilation of {101} twinning planes [37]. While for Au NWs formed along the [110] direction, quantum mechanical molecular dynamics shows that it reconstructs upon stress to form helical NWs [38]. The different final structures are probably due to temperature effect as well. The helical NW is obtained by annealing the crystal structure from 600K, while the BCT structure is relaxed to a minimum energy state at

Some further simulations have also been performed [39] to investigate the core/shell struc‐ ture in which the crystalline core is encapsulated by a glassy shell. Such core/shell hetero‐ structures usually possess particularly attractive applications in the nanoelectromechanical system, owing to their high complexity and functionality. The elongation of the metallic NW is realized by moving the rigid atoms at one end along the axial axis. The rigid atoms are stretched 0.001nm every 1 ps. This strain rate is sufficient to dissipate excess energy to main‐ tain a constant temperature. The Berendsen method [40] is used to control the system tem‐ perature. The motion equations are solved by a fourth-order Nordsieck-Gear predictor corrector method[41] using a time step of 1 fs. The average of the virial stress component along the axial axis during deformation is taken as the macroscopic stress of the wire. This stress measure totally differs from either engineering stress or true stress, which has been used widely in molecular simulations [42-44]. Here the low temperature (T=50K) was used to highlight the material responses upon mechanical activation, as the simulated quenching glass structure after rapid cooling is usually more prone to thermal activations than the ex‐ perimental one. Continuous deformation with noticeable necking occurs to the core/shell

The core/shell structure, which can even be spun into a single atomic chain, tends to display much larger plasticity than the crystalline one. Most importantly, during elongation, local amorphization happens in the high strained region. Generally, it is thought that the amorph‐ ization behavior in nanoscale materials is induced by the high strain rate or the entropy at a high temperature. In terms of our simulation, the strain rate of 0.006% ps−1 for a 16.3nm long NW, which is equivalent to an approximate stretch velocity of 1.0ms−1, belongs to the range of low stretch velocity compared with previous simulation results. Therefore, we suggest that the surface amorphous layer would induce the disorder of the crystalline core during

to

**Figure 1.** The stretching process of core-shelled CuZr NW, the color bar is to illustrate the contour plots of the peratom potential energy.

To shed light on the elastic and plastic deformation mechanism in the core/shell NW, evolu‐ tions of energy and structure in three key deformation stages are provided in Fig. 2. Here, Eper is the total energy per atom at different stretching moment. For the sake of simplicity, an original B2 unit cell taken from the denoted region is used to track the structural transforma‐ tion. Figure 2(a) shows an initial cross-section image along the axial direction. Figure 2(b) shows an incipient elongating stage. Initially, the original B2 lattice in crystalline core is cu‐ bic with a lattice parameter of 0.361 nm. When the stress is applied, a typical martensitic transformation from B2 to BCT phase is observed. That is, a set of new lattice parameters of a=0.289nm and c=0.393nm forms in the crystalline-core zone. When further elongated as shown in Fig. 2(c), the lattice in the deformed crystalline core exhibits increasingly serious distortion. Before fracture happens, atoms with high stress concentration and potential ener‐ gy get rid of the constraint of the crystalline lattice and turn to be dominant. After that, the crystalline phase in the core converts into the disordered state.

**Figure 2.** The structural evolution of composite nanowire in stretching, which varies from (a) the B2 state, then (b) the BCT state, to (c) amorphous state.

Only a global description of the structural evolution in the unique core/shell NW during elongation has been provided above, yet extensive analyses on those amazing phenomena are still in great need. To address this issue, three core/shell NW systems containing the same-sized crystalline core (the diameter of the crystalline core is kept at 2 nm) but different size of amorphous surface layer (the crystalline-amorphous ratios are 1 : 1, 1 : 2 and 1 : 3) are analyzed in Fig. 3. As shown in Fig. 3(a), the degree of amorphization has been employed to describe the percentage of the atoms which experiences an order-disorder transformation of the crystalline core inside the system. It is worth noting that the degree of amorphization in the core correlates closely to the crystalline-amorphous ratio. The degree of amorphization in the crystalline core rises from 18% to 40% when the crystalline-amorphous ratio changes from 1 : 1 into 1 : 3, indicating that the degree of amorphization in the core strongly depends on the crystalline-amorphous ratio. Nevertheless, the response between the degree of amorphization in the core and the crystalline-amorphous ratio is nonlinear, which should be discussed in further investigation. The corresponding total strain-stress curves with differ‐ ent crystalline-amorphous ratios are also plotted. As shown in Fig. 3(b), these curves indi‐ cate that the ductility of the core/shell NW is proportional to the thickness of the amorphous surface layer whereas the breaking strength keeps declining. In addition, the response of strain energy in crystalline core with strain increasing is also plotted in Fig. 3(c) which can quantitatively reflect the stress concentration exerted by the amorphous surface layer. With the amorphous surface layer thickening, the total strain energy in the crystalline core be‐ comes higher, indicating that a thicker amorphous surface layer could introduce more stress concentration into the crystalline core and thereby convert this zone into a disordered state more easily.

**Figure 3.** When increasing the surface amorphous thickness, the change curve of (a) the decrystallized degree of the internal crystalline core; (b) the total stress-strain responses; (c) and the strain energy-strain responses of the internal crystalline core. The red and blue curves in (b) and (c) are offset to a strain of 10% and 20%, respectively.

For a better understanding of the corresponding characteristics in the core/shell NW at an atomic level, real-time Honeycutt and Anderson (HA) bond-type index is introduced, which can describe and help to discern the concrete relationships between an atom and its nearest neighbors. The atoms at the high strained region in the crystalline core (ranged from 8 nm to 12 nm) are considered. Figure 4 illustrates the evolution of the BCC bond pairs, as well as the proportion of icosahedral bond pairs in the selected area of crystalline corewith strain increasing, which corresponds to the elongating process shown in Fig. 1. It is noted that when the strain increases, the total amount of BCC type bond pairs (1441 and 1661 bond pairs) remains approximately constant at first. And then both of them experience a sharp de‐ cline from 40% to about 5%. At the same time, the amount of 1541 bond pair which is related to the characteristic of defective icosahedral ordering initially keeps steady and then increas‐ es sharply. Moreover, there exists an obvious turning point in both curves which clearly re‐ flects the BCT-disorder transformation corresponding to Fig. 2(c). That is, the BCC bond pairs may firstly convert into the 1541 bond pairs. It should be noted that the increase of the 1551 bond pairs in Fig. 4(b) happens slightly later to that of the 1541 ones. The possible rea‐ sons for this special phenomenon will be further discussed in the following.

Only a global description of the structural evolution in the unique core/shell NW during elongation has been provided above, yet extensive analyses on those amazing phenomena are still in great need. To address this issue, three core/shell NW systems containing the same-sized crystalline core (the diameter of the crystalline core is kept at 2 nm) but different size of amorphous surface layer (the crystalline-amorphous ratios are 1 : 1, 1 : 2 and 1 : 3) are analyzed in Fig. 3. As shown in Fig. 3(a), the degree of amorphization has been employed to describe the percentage of the atoms which experiences an order-disorder transformation of the crystalline core inside the system. It is worth noting that the degree of amorphization in the core correlates closely to the crystalline-amorphous ratio. The degree of amorphization in the crystalline core rises from 18% to 40% when the crystalline-amorphous ratio changes from 1 : 1 into 1 : 3, indicating that the degree of amorphization in the core strongly depends on the crystalline-amorphous ratio. Nevertheless, the response between the degree of amorphization in the core and the crystalline-amorphous ratio is nonlinear, which should be discussed in further investigation. The corresponding total strain-stress curves with differ‐ ent crystalline-amorphous ratios are also plotted. As shown in Fig. 3(b), these curves indi‐ cate that the ductility of the core/shell NW is proportional to the thickness of the amorphous surface layer whereas the breaking strength keeps declining. In addition, the response of strain energy in crystalline core with strain increasing is also plotted in Fig. 3(c) which can quantitatively reflect the stress concentration exerted by the amorphous surface layer. With the amorphous surface layer thickening, the total strain energy in the crystalline core be‐ comes higher, indicating that a thicker amorphous surface layer could introduce more stress concentration into the crystalline core and thereby convert this zone into a disordered state

**Figure 3.** When increasing the surface amorphous thickness, the change curve of (a) the decrystallized degree of the internal crystalline core; (b) the total stress-strain responses; (c) and the strain energy-strain responses of the internal

For a better understanding of the corresponding characteristics in the core/shell NW at an atomic level, real-time Honeycutt and Anderson (HA) bond-type index is introduced, which can describe and help to discern the concrete relationships between an atom and its nearest neighbors. The atoms at the high strained region in the crystalline core (ranged from 8 nm to 12 nm) are considered. Figure 4 illustrates the evolution of the BCC bond pairs, as well as the proportion of icosahedral bond pairs in the selected area of crystalline corewith strain increasing, which corresponds to the elongating process shown in Fig. 1. It is noted that when the strain increases, the total amount of BCC type bond pairs (1441 and 1661 bond

crystalline core. The red and blue curves in (b) and (c) are offset to a strain of 10% and 20%, respectively.

more easily.

376 Nanowires - Recent Advances

Further studies focus on the transformation of bond-pairs at the crystalline-amorphous in‐ terface and investigate how the local amorphization happen. Figure 5 presents a superclus‐ ter consisting of over one hundred atoms taken from the core/shell NW as shown in Fig. 2(b). The red-centered cluster represents the typical BCC lattice, whereas the brown one sig‐ nifies the icosahedra cluster. At the initial stage, the B2 crystalline core can convert into the BCT structure. Meanwhile, many crystal defects, such as vacancies, can be found near the crystalline-amorphous interface. With the strain increasing, partial atoms near the interface which possess a high potential energy can insert into the crystal defects and consequently result in local amorphization.

**Figure 4.** Variation of the main fractions of HA indices in the selected region during stretch process: (a) the decreasing bond pairs; (b) the increasing bond pairs. The insets show the typical B2 and icosahedron structures.

For example, the atom labeled 10 in Fig. 5 can insert into a BCC lattice (the corresponding atoms labels 5, 6, 8 and 9) and form 1541 bond pair. With further elongation, more and more atoms will insert into the crystalline core and the total amount of the 1541 bond pairs be‐ comes larger. It is known that the 1551 polyhedra have a higher resistance to plastic defor‐ mation than the 1541 one owing to their higher symmetry. Thus, part of the 1541 bond pairs transforms into the 1551 ones, and as a result, the increase of the 1551 bond pairs is slightly later to that of 1541 ones as shown in fig. 4(b).

**Figure 5.** The evolution of bond pairs in the crystalline-amorphous interfaces. For the sake of analysis, atoms in the nearly same plane are shown.

#### **2.4. Strain rate effect on structure of nanowire**

Using molecular dynamics simulations with a many-body force field, Ikeda[45] studied the deformation of single crystal Ni and NiCu alloy NWs subjected to uniform strain rates at room temperature. For all strain rates, the Ni NWs is elastic up to 7.5% strain with a yield stress of 5.5 GPa. The crystalline phase transforms continuously to an amorphous phase at high strain rates, exhibiting a dramatic change in atomic short-range order and a near van‐ ishing of the tetragonal shear elastic constant perpendicular to the tensile direction. This phase transformation exhibits a new mode of amorphization[45] or even a super plastic be‐ havior [11]. Higher strain rate may result in deformation twins during plastic deformation as well. As the strain rate decreased to about 0.1% ps-1, a transition of deformation mecha‐ nism from combined twining and slip to sequential propagation of slip along well defined and favorably oriented slip plane is observed [11].

#### **2.5. Temperature effect**

The mechanical behavior depends highly on the temperature of the system. Several interest‐ ing features were observed when the NW was deformed in [001] direction at different tem‐ peratures. Due to the higher crystal stability at a lower temperature, the deformation behavior of the NW was characterized by brittle slips, rupture and very low ductility [46]. At the higher temperature of 300 K, the crystal structure became less stable due to higher amplitude of vibration of the atoms about their atomic positions. A relatively stable singlewalled helical substructure was formed due to the higher local vibration amplitude of the platinum atoms which enhances the ductility of the NW [46]. The calculated Young's modu‐ lus of the NW at T=300K was only about 50% that of bulk platinum, and was significantly smaller than that at T=50K.

#### **2.6. Shape memory**

Shape memory is an important property for some engineering materials at large scales. Some NWs also have characteristics of shape memory effect due to structural reorientations which are controlled by a combination of size, thermal energy, and the type of defects formed during inelastic deformation. Certain FCC NWs exhibit both shape memory and pseudoelastic behaviors in atomistic simulations. The formation of defect-free twins and surface stresses is the mechanism that controls the ability of FCC NWs of different materials to show a reversible transition between two crystal orientations during loading and thus shape memory and pseudoelasticity[14].

#### **2.7. Superplasticity**

**Figure 5.** The evolution of bond pairs in the crystalline-amorphous interfaces. For the sake of analysis, atoms in the

Using molecular dynamics simulations with a many-body force field, Ikeda[45] studied the deformation of single crystal Ni and NiCu alloy NWs subjected to uniform strain rates at room temperature. For all strain rates, the Ni NWs is elastic up to 7.5% strain with a yield stress of 5.5 GPa. The crystalline phase transforms continuously to an amorphous phase at high strain rates, exhibiting a dramatic change in atomic short-range order and a near van‐ ishing of the tetragonal shear elastic constant perpendicular to the tensile direction. This phase transformation exhibits a new mode of amorphization[45] or even a super plastic be‐ havior [11]. Higher strain rate may result in deformation twins during plastic deformation as well. As the strain rate decreased to about 0.1% ps-1, a transition of deformation mecha‐ nism from combined twining and slip to sequential propagation of slip along well defined

The mechanical behavior depends highly on the temperature of the system. Several interest‐ ing features were observed when the NW was deformed in [001] direction at different tem‐ peratures. Due to the higher crystal stability at a lower temperature, the deformation behavior of the NW was characterized by brittle slips, rupture and very low ductility [46]. At the higher temperature of 300 K, the crystal structure became less stable due to higher amplitude of vibration of the atoms about their atomic positions. A relatively stable singlewalled helical substructure was formed due to the higher local vibration amplitude of the platinum atoms which enhances the ductility of the NW [46]. The calculated Young's modu‐ lus of the NW at T=300K was only about 50% that of bulk platinum, and was significantly

nearly same plane are shown.

378 Nanowires - Recent Advances

**2.5. Temperature effect**

smaller than that at T=50K.

**2.4. Strain rate effect on structure of nanowire**

and favorably oriented slip plane is observed [11].

Defect-free Au NWs show superplasticity on tensile deformation [47]. Evidences from high resolution electron microscopes indicated that the plastic deformation proceeds layer-bylayer in an atomically coherent fashion to a long distance. The stress-strain curve provides full interpretation of the deformation. After initial super-elastic deformation, the NW shows superplastic deformation induced by coherent twin propagation, completely reorientating the crystal from <110> to <100>. Uniquely well disciplined and long-propagating atomic movements deduced here are ascribed to the superb crystallinity as well as the radial con‐ finement of the Au NWs. Other mechanical measurements in humid atmospheres suggest that salt NWs also form in ambient environments [48]. Superplastic NWs were formed by touching the NaCl(100) surface with a Au tip in a TEM. The final elongation strain of NWs reached 280%, when showered with the electron beam. More surprisingly, no dislocations were observable during the elongation due to fast diffusion.

#### **2.8. Cold welding**

General welding techniques require precise control of the heating mechanism and introduce the possibility of damage. The welding of nano-materials is likely to play an important role in the fabrication of electromechanical nano-devices. The cold welding (a process without heating) on a micrometer scale has been demonstrated, but only at macroscopic length scales and under large applied pressures. Lu et.al [49] demonstrate that single crystalline Au NWs with diameters between 3 and 10 nm can be welded together within seconds by me‐ chanical contact alone, and under relatively low applied pressures. The nearly perfect weld‐ ing structure possesses the same crystal orientation, strength and electrical conductivity as the rest of the NW. The high quality of the welds is attributed to the nanoscale sample di‐ mensions, oriented-attachment mechanisms and mechanically assisted fast surface-atom dif‐ fusion. The cold weld technique is also applied between other kinds of metals such as gold and silver, and silver and silver.

## **3. Crystalline semiconductor nanowires**

#### **3.1. Silicon and silicon compound nanowires**

It is generally accepted that Si NWs exhibit linear elastic behavior until fracture without any appreciable plastic deformation. However, the plasticity of Si NWs can be triggered under low strain rate inside the TEM. An electron-beam illuminating with a low current intensity would result in the bond re-forming processes, achieving the plastic deformation with a bent strain over 40% in Si NWs near the room temperature [50]. In some other tensile experi‐ ments, the Si NW also showed ultrahigh flexibility and strong toughness [51]. Large strain plasticity (LSP) of single-crystalline Si NWs, which resulted in a fourfold reduction in NW diameter before fracture occurred, was directly observed in an ultrahigh-resolution electron microscope at room temperature. The tensile plasticity was initiated by the emergence of dislocations, followed by the development of a continuous disordered lattice by the emis‐ sion of dislocations and formation of disordered crystalline structures. The size-dependent fracture mechanism is attributed to the scale-related dislocation activities [52]. For ceramic materials especially at low temperature (~300 K) large strain plasticity of ceramic SiC NWs at nearly room temperature was directly observed in situ by a high-resolution TEM as di‐ mensionality decreases. The continuous plasticity of the SiC NWs[8] is accompanied by a process of increased dislocation density at an early stage, followed by an obvious lattice dis‐ tortion, which is similar with stretching Si NW.

Individual single-crystalline Si NWs were bent by forming loops or arcs with different radi‐ us with high-resolution electron microscopy. Bending-induced ripple buckling was ob‐ served and a significant strain variation along the axial direction of the compressive region was revealed. The tensile surface atomic steps and the compressive buckling are the physi‐ cal origin of the asymmetric tensile-compressive properties of postelastic instabilities and the incipient plasticity. Both of the tensile surface atomic-steps and the compressive buck‐ ling initiated versatile ductile plastic dislocation events [53]. Sequential AFM manipulationscanning protocol is used to observe large bending stress states of [112] Si NWs. It is possible to observe large bending stress states of Si NWs as their radius of curvature is pro‐ gressively reduced. A slight increase in the fracture strength of Si NWs from 12 to 18 GPa as their radius decreased from 60 to 20 nm was observed in these measurements [54].

The fracture behavior of the Si NWs depends on both temperature and diameter. For NWs of diameter larger than 4 nm, cleavage fracture on the transverse (110) plane favors below the temperature 1000 K. As the temperature increases, the same NWs shear extensively on inclined {111} planes resulting in a brittle-to-ductile transition. As the diameter decreases to less than 4 nm, it fails by shear regardless of temperature. The reason is that the cleavage fracture is initiated by the nucleation of a crack, while shear failure is initiated by the nuclea‐ tion of a dislocation, both from the surface [55].

Mechanical properties are extremely different at nanometer scale from those at macroscopic scale. Few crystalline nanowires have strength of over 10 GPa, however, Brambilla et al. [56] have manufactured glass silica nanowires which are characterized by ultimate strength over 10 GPa with a top-down fabrication technique.

#### **3.2. ZnO nanowires**

**3. Crystalline semiconductor nanowires**

It is generally accepted that Si NWs exhibit linear elastic behavior until fracture without any appreciable plastic deformation. However, the plasticity of Si NWs can be triggered under low strain rate inside the TEM. An electron-beam illuminating with a low current intensity would result in the bond re-forming processes, achieving the plastic deformation with a bent strain over 40% in Si NWs near the room temperature [50]. In some other tensile experi‐ ments, the Si NW also showed ultrahigh flexibility and strong toughness [51]. Large strain plasticity (LSP) of single-crystalline Si NWs, which resulted in a fourfold reduction in NW diameter before fracture occurred, was directly observed in an ultrahigh-resolution electron microscope at room temperature. The tensile plasticity was initiated by the emergence of dislocations, followed by the development of a continuous disordered lattice by the emis‐ sion of dislocations and formation of disordered crystalline structures. The size-dependent fracture mechanism is attributed to the scale-related dislocation activities [52]. For ceramic materials especially at low temperature (~300 K) large strain plasticity of ceramic SiC NWs at nearly room temperature was directly observed in situ by a high-resolution TEM as di‐ mensionality decreases. The continuous plasticity of the SiC NWs[8] is accompanied by a process of increased dislocation density at an early stage, followed by an obvious lattice dis‐

Individual single-crystalline Si NWs were bent by forming loops or arcs with different radi‐ us with high-resolution electron microscopy. Bending-induced ripple buckling was ob‐ served and a significant strain variation along the axial direction of the compressive region was revealed. The tensile surface atomic steps and the compressive buckling are the physi‐ cal origin of the asymmetric tensile-compressive properties of postelastic instabilities and the incipient plasticity. Both of the tensile surface atomic-steps and the compressive buck‐ ling initiated versatile ductile plastic dislocation events [53]. Sequential AFM manipulationscanning protocol is used to observe large bending stress states of [112] Si NWs. It is possible to observe large bending stress states of Si NWs as their radius of curvature is pro‐ gressively reduced. A slight increase in the fracture strength of Si NWs from 12 to 18 GPa as

their radius decreased from 60 to 20 nm was observed in these measurements [54].

The fracture behavior of the Si NWs depends on both temperature and diameter. For NWs of diameter larger than 4 nm, cleavage fracture on the transverse (110) plane favors below the temperature 1000 K. As the temperature increases, the same NWs shear extensively on inclined {111} planes resulting in a brittle-to-ductile transition. As the diameter decreases to less than 4 nm, it fails by shear regardless of temperature. The reason is that the cleavage fracture is initiated by the nucleation of a crack, while shear failure is initiated by the nuclea‐

Mechanical properties are extremely different at nanometer scale from those at macroscopic scale. Few crystalline nanowires have strength of over 10 GPa, however, Brambilla et al. [56]

**3.1. Silicon and silicon compound nanowires**

380 Nanowires - Recent Advances

tortion, which is similar with stretching Si NW.

tion of a dislocation, both from the surface [55].

The mechanical properties of ZnO are of considerable interest due to the potential applica‐ tions in electromechanical devices. These mechanical properties are essential for designing, manufacturing, and operating such devices. ZnO NWs represent excellent model systems to investigate this size dependence, the ability to tune the radius over a continuous range and the manner in which their properties approach those of the bulk as a function of shape and size [57]. Young's modulus of ZnO NWs is essentially independent of diameter and close to the bulk value, whereas the ultimate strength increases for small diameter NWs with values up to 40 times that of bulk [58].

#### *3.2.1. Size effect*

A size dependence of Young's modulus in [0001] oriented ZnO NWs (NWs) with diameters ranging from 17 to 550 nm is revealed in Ref. [57]. The measured modulus for NWs with diameters smaller than about 120 nm is increasing dramatically with the decreasing diame‐ ters, and is significantly higher than that of the larger ones whose modulus tends to that of bulk ZnO. The radial elastic moduli of [0001]ZnO NWs are determined by contact resonance atomic force microscopy (CR-AFM). The radial indentation modulus and the tangential shear modulus from friction-type measurements are obtained by using realistic tip-nano‐ wire contact geometry. Both moduli increase when the wire diameter is reduced below 80 nm. The size dependence of the elastic properties can be explained by a core-shell model that accounts for a bulk-like core and an elastically stiffer surface layer [59].

Secondly, friction type measurements are useful to directly probe the effects of surface stiff‐ ening on the shear response of the NW. Both the lateral shear modulus and radial indenta‐ tion modulus increase significantly with decreasing NW diameter, which is due to a surface stiffening effect. The results could be understood by the core-shell model of the NW men‐ tioned above, in which the elastic properties of the shell becomes predominant in the limit of small ZnO NW diameters [59].

## **4. Other novel structures of nanowires**

#### **4.1. Ultra thin nanowires**

Computer simulations suggest that ultrathin metal wires should develop exotic, noncrystal‐ line stable atomic structures once their diameter decreases below a critical size of the or‐ der of a few atomic spacings. The new structures, e.g. helical, spiral-structured wires, whose details depend upon the material and the wire thickness, may be dominated by icosahe‐ dral packings. The phenomenon, analogous to the appearance of icosahedral and other noncrystalline shapes in small clusters, can be rationalized in terms of surface energy ani‐ sotropy and optimal packing [60]. In some experiments, suspended Au NWs have recent‐ ly been made in an ultrahigh vacuum and were imaged by the TEM [4]. Mechanisms of formation, evolution, and breaking of these atomically thin Au NWs under stress have attracted great interest.

In some MD simulations, a single chain forms and it is in agreement with experimental results. It shows how defects induce the formation of constrictions that eventually will form the one-atom chains [25]. While in the *Ab initio* calculations, the bond strength of the NW is about twice that of a bulk metallic bond. The importance of knowing bond and atom arrangement is that the total effective stiffness of the nanostructure is strongly affect‐ ed by them [5].

#### **4.2. Twinned nanowires**

Simulation results reveal that with decreasing twin boundary spacing there is a transition from softening to strengthening due to a change in the dominant mechanism of plasticity deformation. Moreover, the value of critical twin boundary spacing is independent of the length of NW but depend on the diameter of the NW. The source controlled deformation mechanism in nanostructured materials will be helpful in understanding the plastic behav‐ ior of twinned NWs[61]. Some other simulations reveal the existence of a fundamental limit dividing the mechanical behavior of twinned Au NWs into either strain softening or strain hardening regimes, as a function of the NW diameter and the number of twins. In some oth‐ er experiments, in-situ tensile testing of Ag NWs with diameters between 34 and 130 nm was carried out inside a SEM. Pronounced strain hardening was observed for most NWs. The strain hardening is mainly attributed to the presence of internal twin boundaries [62]. Since there are size-dependent mechanical properties in NWs, it is necessary for one to de‐ velop strategies to control the strength of these materials. The mechanical properties of Ag NWs with a unique fivefold twin structure using a lateral force AFM method were reported by Wu et.al [63]. It exhibits super elastic behavior followed by unexpected brittle failure without significant plastic deformation. However, thermal annealing resulted in a gradual transition to weaker, more ductile materials associated with the elimination of the twinned boundary structure [63].

### **5. Enhancement of nanowires in nano hybrid systems**

The optimized Cu NWs are constrained within the hollow CNT as shown in Fig. 6. It shows the front and side views of the initial configuration of Cu NW encapsulated into CNT(6,6) and CNT(8,8). Bulk copper has a FCC structure. But at nanometer scale, randomly and fully dispersed Cu atoms form a helical or multi-shell equilibrium structure after relaxation to reach the minimum energy. This phenomenon is mainly due to the constraint effect of the template CNT and its special rolled up hexagonal network structure. Cu NW filled in CNT(6,6) (Fig.6(a)) is three-strand NW twisting with each other.

sotropy and optimal packing [60]. In some experiments, suspended Au NWs have recent‐ ly been made in an ultrahigh vacuum and were imaged by the TEM [4]. Mechanisms of formation, evolution, and breaking of these atomically thin Au NWs under stress have

In some MD simulations, a single chain forms and it is in agreement with experimental results. It shows how defects induce the formation of constrictions that eventually will form the one-atom chains [25]. While in the *Ab initio* calculations, the bond strength of the NW is about twice that of a bulk metallic bond. The importance of knowing bond and atom arrangement is that the total effective stiffness of the nanostructure is strongly affect‐

Simulation results reveal that with decreasing twin boundary spacing there is a transition from softening to strengthening due to a change in the dominant mechanism of plasticity deformation. Moreover, the value of critical twin boundary spacing is independent of the length of NW but depend on the diameter of the NW. The source controlled deformation mechanism in nanostructured materials will be helpful in understanding the plastic behav‐ ior of twinned NWs[61]. Some other simulations reveal the existence of a fundamental limit dividing the mechanical behavior of twinned Au NWs into either strain softening or strain hardening regimes, as a function of the NW diameter and the number of twins. In some oth‐ er experiments, in-situ tensile testing of Ag NWs with diameters between 34 and 130 nm was carried out inside a SEM. Pronounced strain hardening was observed for most NWs. The strain hardening is mainly attributed to the presence of internal twin boundaries [62]. Since there are size-dependent mechanical properties in NWs, it is necessary for one to de‐ velop strategies to control the strength of these materials. The mechanical properties of Ag NWs with a unique fivefold twin structure using a lateral force AFM method were reported by Wu et.al [63]. It exhibits super elastic behavior followed by unexpected brittle failure without significant plastic deformation. However, thermal annealing resulted in a gradual transition to weaker, more ductile materials associated with the elimination of the twinned

attracted great interest.

382 Nanowires - Recent Advances

ed by them [5].

**4.2. Twinned nanowires**

boundary structure [63].

**5. Enhancement of nanowires in nano hybrid systems**

CNT(6,6) (Fig.6(a)) is three-strand NW twisting with each other.

The optimized Cu NWs are constrained within the hollow CNT as shown in Fig. 6. It shows the front and side views of the initial configuration of Cu NW encapsulated into CNT(6,6) and CNT(8,8). Bulk copper has a FCC structure. But at nanometer scale, randomly and fully dispersed Cu atoms form a helical or multi-shell equilibrium structure after relaxation to reach the minimum energy. This phenomenon is mainly due to the constraint effect of the template CNT and its special rolled up hexagonal network structure. Cu NW filled in

**Figure 6.** Front and side views of the initial configuration of Cu NW encapsulated in CNT. (a) NW@CNT(6,6), (b)NW@CNT(8,8).

In addition, another Cu NW encapsulated into CNT(8,8) has multi-shell packs that are com‐ posed of coaxial cylindrical shells (Fig.6(b)). In this case, Cu NW has a single strand chain at the center. Outside shell is formed by rows of atoms that are helically wound upwards side by side. A displacement increment of 0.0004 nm in axial direction is applied at both ends in each load step. The simulation is carried out at approximately 1 K to avoid thermal effects. Each time step used in the simulation is 1 fs. At the same time, the atoms at both ends of the composite are kept transparent to inter-atomic forces [64].

Figure 7 shows the function of the strain energy of Cu NW@CNT(16,0) with the length of 34.08Å as the time step increases. The inset is a compressive schematic, in which the widths of both fixed ends are 5Å. The constant velocity referred above is imposed on both fixed end atoms which consist of carbon and copper atoms. The strain energy per atom is determined as the difference in total energy per atom of the strained and unstrained NW@CNT. In addi‐ tion, we use the strain energy of the pristine hollow CNT(16,0) under compressive load for comparison. The strain energy of CNT is increasing non-linearly without any drop before buckling. However, the strain energy of the NW@CNT undergoes a small drop atε=5.2% (Fig.8(b)) and ε=6.6% (Fig. 8(c)) because of slipping of copper atoms during the compressive process. As the strain reaches a critical point at ε=10.4% (t=3000fs), the composite undergoes abrupt buckling deformation resulting in an only 2.7% spontaneous drop which is lower than that of a pristine CNT (33.5%). This means that the composite is more stable than the corresponding pristine CNT when subject to abrupt buckling. Wang et al. [65] have investi‐ gated the stability of Ni metal atoms fully encapsulated in CNT. They found that the stabili‐ ty of metal filled in CNT depends much upon the amount of filling metal. When the metal atom is fully filled into CNT, it is more stable than a pristine CNT. This result is similar to some reported in our study. In our calculations, we propose that the mechanical properties of NW@CNTs depend upon the diameter, length, and carbon nanotube's chirality.

**Figure 7.** Average strain energy per atom for Cu NW@CNT(16,0) as a function of time step. The red line represents the strain energy of a CNT(16,0) with the same length for comparison.

**Figure 8.** Morphological changes of a Cu NW@CNT(16,0) at different strains corresponding to Fig. 7. (a) Initial config‐ uration. (b) At t=1500fs. (c) At=1900fs. (d) The buckling occurs at t=3000fs.

Table 1 illustrates the relationship between the diameter and buckling loadPcr of a series of Cu NW encapsulated into armchair CNT and zigzag CNT. The lengths of the series of arm‐ chair CNTs are approximately 36.9Å, while those of zigzag CNTs are about 34Å.


**Table 1.** buckling loads Pcr for selected Cu NW@armchair CNT and NW@zigzag CNT

**Figure 7.** Average strain energy per atom for Cu NW@CNT(16,0) as a function of time step. The red line represents

**Figure 8.** Morphological changes of a Cu NW@CNT(16,0) at different strains corresponding to Fig. 7. (a) Initial config‐

Table 1 illustrates the relationship between the diameter and buckling loadPcr of a series of Cu NW encapsulated into armchair CNT and zigzag CNT. The lengths of the series of arm‐

chair CNTs are approximately 36.9Å, while those of zigzag CNTs are about 34Å.

the strain energy of a CNT(16,0) with the same length for comparison.

384 Nanowires - Recent Advances

uration. (b) At t=1500fs. (c) At=1900fs. (d) The buckling occurs at t=3000fs.

When the diameter increases from 8.14 to 27.12 Å (i.e. aspect ratio decreases from 4.53 to 1.33), the buckling load Pcr of NW@armchair CNTs increases from 0.93 to 2.29 x10-7N. More‐ over, the critical buckling load Pcr of NW@zigzag CNTs increases from 1.12 to 1.73 x10-7 N as its diameter increases. In order to confirm how the insertion of Cu NW affects the pris‐ tine CNT, we calculate the buckling load Pcr of NW@CNT in comparison with that of the corresponding pristine CNT. The percentage increase is defined as [(critical buckling load Pcr of NW@CNT-Pcr of CNT)/Pcr of CNT]. As the diameter increases, the percentage in‐ crease of both NW@armchair CNTs and NW@zigzag CNTs rises in the same way. Especial‐ ly for the NW@CNT (20,20) and NW@CNT(24,0), the percentage increase has reached 81.7% and 44.2%, respectively. So the insertion of Cu NW into a thicker CNT reinforces CNT more than the case with a thinner one. Another interesting point we should notice is that when the aspect ratio is 4.53 for NW@CNT(6,6), or 4.35 for NW@CNT(10,0), the percentage in‐ crease is minus.

**Figure 9.** The percentage increase for the buckling loads Pcr of the composite material (NW@CNT) compared with the corresponding hollow nanotube.

To further investigate the relationship between length and reinforcing effect of NW@CNT on the critical load, we illustrate the percentage increase of NW@armchair and NW@zigzag CNT as a function of length in Fig. 9. For NW@armchair CNT, the percentage increase de‐ creases as the length increases. There is an equilibrium value for each NW@CNT, after which the percentage increase will be minus. It means that if the length of the NW@CNT is larger than the certain value, the insertion of NW will weaken the strength of CNT. From another point of view, as the length of the composite increases, it can be viewed as one-di‐ mension long column, which means that the stability of the solid column is not stronger than the hollow structure during compression. In continuum mechanics, the Euler formula is employed to determine the critical strain of the long beamlike buckling mode *<sup>ε</sup>cr* <sup>=</sup> *<sup>π</sup>* <sup>2</sup> (*KL* /*r*)<sup>2</sup> , where *L* is the length of the column; *<sup>r</sup>* <sup>=</sup> *<sup>I</sup>* / *<sup>A</sup>*is the gyration radius, where *<sup>I</sup>* and *A* are the moment of inertia and the area of cross section; and *K* is a constant in lieu of the effects of boundary conditions on the Euler formula. The greatest difference between CNT and NW@CNT is the area of cross section. According to the foregoing formula, the crit‐ ical strain of a solid composite will be smaller than that of the hollow CNT. Furthermore, the buckling load of composite would be smaller than that of CNT.

In other words, when the length of the composite is long enough, Cu NW filled in the CNT buckles earlier than the pristine CNT at a certain strain. There will be some Cu atoms ap‐ proaching the inside wall of the CNT, resulting in strong interactions along the plane per‐ pendicular to the loading direction. As the compressive strain continues, along with the effect of van der Waals forces included, the interactions between the two different kinds of atoms will initiate the buckling of the whole composite earlier.

**Figure 10.** Critical buckling loads (black squares) and the quantity percentage increase of the composite (red circles) with approximately the same diameter and length as function of tube chirality for CNT encapsulating NWs.

Figure 10 shows both the critical loads Pcr and the percentage increase as functions of the nanotube's chiral angle. The chirality dependent compressive response of these NWs@CNTs with nearly the same aspect ratio (length=42 Å) is observed. The critical load decreases steplike as the chiral angle increases to about 15° and then increases to 1.36x10-7N when the chi‐ ral angle reaches 30 degrees (armchair CNT). The critical load of zigzag CNT encapsulating NWs is the largest of all these types of composites, because there are substantial bonds which are parallel to the loading direction and they can withstand larger compressive load‐ ing before buckling. Most chiral CNTs encapsulating NWs are more sensitive than NWs@zigzag CNT in terms of percentage increase. It is seen that copper NWs can increase the critical loading of CNT (13, 7) by about 25%. The insertion of copper NWs into other CNTs with different chiral angles can also enhance the strength of the corresponding pure CNTs. So this enhancement of CNT with different chiral angles makes this kind of compo‐ site a promising material to act as building blocks.

In order to further determine the relationship between buckling load Pcr and composite's length, the relationship between them is plotted in Fig. 11. Euler formula for the critical buckling load is *Pcr* =4*<sup>π</sup>* <sup>2</sup> *EI* / *L* <sup>2</sup> , where L is the length. As for NW@armchair CNT with a length ranging from 35 to 65 Å (shown in Fig. 11a), the buckling loads Pcr for NW@CNT (8,8) and NW@CNT(10,10) fit nearly linearly with the value 1/L2 . In addition, the similar trend is also found for NW@CNT (14,0) and NW@CNT(16,0) as shown in Fig. 11 (b).The axial Young's modulus can be defined from the continuum theory in the form of*<sup>Y</sup>* <sup>=</sup> *<sup>d</sup>* <sup>2</sup> *E Vd<sup>ε</sup>* <sup>2</sup> , in which E is the strain energy, ε is the axial strain and V is the overall volume of the

whole composite.

**Figure 9.** The percentage increase for the buckling loads Pcr of the composite material (NW@CNT) compared with the

To further investigate the relationship between length and reinforcing effect of NW@CNT on the critical load, we illustrate the percentage increase of NW@armchair and NW@zigzag CNT as a function of length in Fig. 9. For NW@armchair CNT, the percentage increase de‐ creases as the length increases. There is an equilibrium value for each NW@CNT, after which the percentage increase will be minus. It means that if the length of the NW@CNT is larger than the certain value, the insertion of NW will weaken the strength of CNT. From another point of view, as the length of the composite increases, it can be viewed as one-di‐ mension long column, which means that the stability of the solid column is not stronger than the hollow structure during compression. In continuum mechanics, the Euler formula is employed to determine the critical strain of the long beamlike buckling mode

(*KL* /*r*)<sup>2</sup> , where *L* is the length of the column; *<sup>r</sup>* <sup>=</sup> *<sup>I</sup>* / *<sup>A</sup>*is the gyration radius, where *<sup>I</sup>* and *A* are the moment of inertia and the area of cross section; and *K* is a constant in lieu of the effects of boundary conditions on the Euler formula. The greatest difference between CNT and NW@CNT is the area of cross section. According to the foregoing formula, the crit‐ ical strain of a solid composite will be smaller than that of the hollow CNT. Furthermore, the

In other words, when the length of the composite is long enough, Cu NW filled in the CNT buckles earlier than the pristine CNT at a certain strain. There will be some Cu atoms ap‐ proaching the inside wall of the CNT, resulting in strong interactions along the plane per‐ pendicular to the loading direction. As the compressive strain continues, along with the effect of van der Waals forces included, the interactions between the two different kinds of

buckling load of composite would be smaller than that of CNT.

atoms will initiate the buckling of the whole composite earlier.

corresponding hollow nanotube.

386 Nanowires - Recent Advances

*<sup>ε</sup>cr* <sup>=</sup> *<sup>π</sup>* <sup>2</sup>

**Figure 11.** a) Buckling loads Pcr of Cu NW@CNT (8,8) and NW@CNT(10,10) verse 1/L2 (where L represents the length of NW@CNT), (b) Pcr of Cu NW@CNT(14,0) and NW@CNT(16,0) versus 1/L2.

**Figure 12.** a) Plots depicting the effect of NW@CNT length on the Young's modulus. (b) Plots depicting the effect of NW@CNT radius on the Young's modulus.

As plotted in Fig. 12(a), an increment in length decreases the Young's modulus slightly. A decline of only 4.7% is observed in the decrement of Young's modulus of NW@CNT (8,8) as its length increases from 25 to 65 Å. And there is a similarly small decline of 5.3% in the dec‐ rement of Young's modulus of NW@CNT (14, 0) as its length increases from 35 to 70 Å. The general trend is that the Young's modulus decreases as the length of NW@CNT increases. Although there are some points that do not obey this rule, it is mainly due to the different distribution of Cu atoms filled in CNTs. As shown in Fig. 12(b), Young's modulus is sensi‐ tive to the radius of NW@CNT ranging from 8 to 14 Å. For example, it drops about 28.7% (i.e. from 0.645 to 0.46Tpa) when radius of the NW@armchair CNT increases from 4.07 to 10.85 Å. In addition, there is a 19.8% drop (from 0.551 to 0.442 TPa) for Young's modulus of NW@zigzag CNT. When the radius is sufficiently large, Young's modulus would reach a constant value 0.46TPa for NW@armchair CNT and 0.44 TPa for NW@zigzag CNT.

## **6. Conclusions**

**Figure 11.** a) Buckling loads Pcr of Cu NW@CNT (8,8) and NW@CNT(10,10) verse 1/L2 (where L represents the length

**Figure 12.** a) Plots depicting the effect of NW@CNT length on the Young's modulus. (b) Plots depicting the effect of

As plotted in Fig. 12(a), an increment in length decreases the Young's modulus slightly. A decline of only 4.7% is observed in the decrement of Young's modulus of NW@CNT (8,8) as its length increases from 25 to 65 Å. And there is a similarly small decline of 5.3% in the dec‐ rement of Young's modulus of NW@CNT (14, 0) as its length increases from 35 to 70 Å. The general trend is that the Young's modulus decreases as the length of NW@CNT increases. Although there are some points that do not obey this rule, it is mainly due to the different distribution of Cu atoms filled in CNTs. As shown in Fig. 12(b), Young's modulus is sensi‐ tive to the radius of NW@CNT ranging from 8 to 14 Å. For example, it drops about 28.7% (i.e. from 0.645 to 0.46Tpa) when radius of the NW@armchair CNT increases from 4.07 to 10.85 Å. In addition, there is a 19.8% drop (from 0.551 to 0.442 TPa) for Young's modulus of NW@zigzag CNT. When the radius is sufficiently large, Young's modulus would reach a

constant value 0.46TPa for NW@armchair CNT and 0.44 TPa for NW@zigzag CNT.

of NW@CNT), (b) Pcr of Cu NW@CNT(14,0) and NW@CNT(16,0) versus 1/L2.

NW@CNT radius on the Young's modulus.

388 Nanowires - Recent Advances

This paper reviews some recent development on the mechanical properties and mainly fo‐ cuses on the elongation of a metallic nanowire (NW) with a core/shell structure and the en‐ hancement of helical NWs in a hybrid system.


## **Acknowledgements**

We would like to acknowledge the support from the National Basic Research Program of China (Grant No.2012CB825702). This work is also supported by the National Natural Sci‐ ence Foundation of China (Grant No. 50971081, 51271100). This work is also funded by the grants from Shandong Province in China (Nos. 50625101, ZR2009FM043). We also thank the support from Shandong University (No2009JQ014).

## **Author details**

Hui Li1\* and Fengwei Sun2

\*Address all correspondence to: lihuilmy@hotmail.com

1 Key Laboratory for Liquid-Solid Structural Evolution and Processing of Materials, Minis‐ try of Education, Shandong University, China

2 Department of Materials Science and Engineering, Delft University of Technology, The Netherlands

## **References**


[12] Liang, H., Upmanyu, M., & Huang, H. (2005). Size-dependent elasticity of nanowires: Nonlinear effects. *Physical Review B R*, 71, 24, 241403(R).

1 Key Laboratory for Liquid-Solid Structural Evolution and Processing of Materials, Minis‐

2 Department of Materials Science and Engineering, Delft University of Technology, The

[1] Pascual, J. I., Mendez, J., Gomezherrero, J., Baro, A. M., Garcia, N., & Binh, V. T. (1993). Quantum Contact in Gold Nanostructures by Scanning-tunneling-microsco‐

[2] Agrait, N., Rodrigo, J. G., & Vieira, S. (1993). Conductance Steps and Quantization in

[3] Olesen, L., Laegsgaard, E., Stensgaard, I., Besenbacher, F., Schiotz, J., Stoltze, P., et al. (1994). Quantized Conductance in an Atom-sized Point-contact. *Physical Review Let‐*

[4] Kondo, Y., & Takayanagi, K. (1997). Gold Nanobridge Stabilized by Surface Struc‐

[5] Rubio-Bollinger, G., Bahn, S., Agraït, N., Jacobsen, K., & Vieira, S. (2001). Mechanical Properties and Formation Mechanisms of a Wire of Single Gold Atoms. *Physical Re‐*

[6] Wu, B., Heidelberg, A., & Boland, J. J. (2005). Mechanical properties of ultrahigh-

[7] Hoffmann, S., Utke, l., Moser, B., Michler, J., Christiansen, S. H., Schmidt, V., et al. (2006). Measurement of the Bending Strength of Vapor-Liquid-Solid Grown Silicon

[8] Han, X. D., Zhang, Y. F., Zheng, K., Zhang, X. N., Zhang, Z., Hao, Y. J., et al. (2007). Low-Temperature in Situ Large Strain Plasticity of Ceramic SiC Nanowires and Its

[9] Chantasiriwan, S., & Milstein, F. (1998). Embedded-atom models of 12 cubic metals incorporating second- and third-order elastic-moduli data. *Physical Review B*, 58(10),

[10] Branício, P. S., & Rino, P. J. (2000). Large deformation and amorphization of Ni nano‐ wires under uniaxial strain:A molecular dynamics study. *Physical Review B*, 62,

[11] Liang, W., & Zhou, M. (2003). Size and Strain Rate Effects in Tensile Deformation of

try of Education, Shandong University, China

py. *Physical Review Letters*, 71(12), 1852-5.

ture. *Physical Review Letters*, 79, 3455-8.

*ters*, 72(14), 2251-4.

*view Letters*, 87(2), 026101.

5996-6005.

16950-5.

Nanowires. *Nano Letters*, 6, 622-5.

Atomic-size Contacts. *Physical Review B*, 47(18), 12345-8.

strength gold nanowires. *Nature Materials*, 4(7), 525-9.

Atomic-Scale Mechanism. *Nano Letters*, 7, 452-7.

Cu Nanowires. *Nanotechnology*, 2, 452-5.

Netherlands

390 Nanowires - Recent Advances

**References**


[42] Wei, Y., Bower, A. F., & Gao, H. (2010). Analytical model and molecular dynamics simulations of the size dependence of flow stress in amorphous intermetallic nano‐ wires at temperatures near the glass transition. *Physical Review B*, 81(12), 125402.

[28] McDowell, M. T., Leach, A. M., & Gall, K. (2008). On The Elastic Modulus of Metallic

[29] McDowell, M. T., Leach, A. M., & Gall, K. (2008). Bending and tensile deformation of metallic nanowires. *Modelling and Simulation in Materials Science and Engineering*,

[30] Yun, G., & Park, H. (2009). Surface stress effects on the bending properties of fcc met‐

[31] Park, H., & Klein, P. (2008). Surface stress effects on the resonant properties of metal nanowires: The importance of finite deformation kinematics and the impact of the re‐ sidual surface stress. *Journal of the Mechanics and Physics of Solids*, 56(11), 3144-66.

[32] Petrova, H., Perez-Juste, J., Zhang, Z. Y., Zhang, J., Kosel, T., & Hartland, G. V. (2006). Crystal structure dependence of the elastic constants of gold nanorods. *J Ma‐*

[33] Cuenot, S., Fretigny, C., Demoustier-Champagne, S., & Nysten, B. (2004). Surface ten‐ sion effect on the mechanical properties of nanomaterials measured by atomic force

[34] Jing, G. Y., Duan, H. L., Sun, X. M., Zhang, Z. S., Xu, J., Li, Y. D., et al. (2006). Surface effects on elastic properties of silver nanowires: Contact atomic-force microscopy.

[35] Husain, A., Hone, J., Postma, H. W. C., Huang, X. M. H., Drake, T., Barbic, M., et al. (2003). Nanowire-based very-high-frequency electromechanical resonator. *Applied*

[36] Verbridge, S. S., Shapiro, D. F., Craighead, H. G., & Parpia, J. M. (2007). Macroscopic tuning of nanomechanics: Substrate bending for reversible control of frequency and

[37] Park, H. S. (2006). Stress-Induced Martensitic Phase Transformation in Intermetallic

[38] Amorim, E., & da, Silva. E. (2008). Helical [110] Gold Nanowires Make Longer Linear

[39] Zhang, K., Si, P. C., Li, H., Li, Y. F., Yu, H. Q., & Jiang, Y. Y. (2012). Elongation behav‐ ior and local amorphization of metallic nanowire with glassy shell and crystalline

[40] Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F., Di Nola, A., & Haak, J. R. (1984). Molecular dynamics with coupling to an external bath. *Journal of Chemical*

[41] Allen, M. P., & Tildesley, L. J. (1987). Computer Simulation of Liquids. *(Oxford Uni‐*

quality factor of nanostring resonators. *Nano Letters*, 7(6), 1728-35.

Nickel Aluminum Nanowires. *Nano Letters*, 6(5), 958-62.

Atomic Chains. *Physical Review Letters*, 101(12), 125502.

core. *EPL (Europhysics Letters)*, 97(2), 26005.

Nanowires. *Nano Letters*, 8, 3613-8.

al nanowires. *Physical Review B*, 79(19), 195421.

microscopy. *Physical Review B*, 69(16), 165410.

16(4), 045003.

392 Nanowires - Recent Advances

*ter Chem*, 16(40), 3957-63.

*Physical Review B*, 73(23), 235409.

*Physics Letters*, 83(6), 1240-2.

*Physics*, 81, 3684.

*versity Press, Oxford)*.


## **How and Why Alumina Matrix Architecture Influence the Shape and Composition of Nanowires Grown by AC Deposition?**

Arūnas Jagminas

[56] Brambilla, G., & Payne, D. N. (2009). The Ultimate Strength of Glass Silica Nano‐

[57] Chen, C., Shi, Y., Zhang, Y., Zhu, J., & Yan, Y. (2006). Size Dependence of Young's

[58] Wen, B., Sader, J., & Boland, J. (2008). Mechanical Properties of ZnO Nanowires.

[59] Stan, G., Ciobanu, C. V., Parthangal, P. M., & Cook, R. F. (2007). Diameter-Dependent Radial and Tangential Elastic Moduli of ZnO Nanowires. *Nano Letters*, 7, 3691-7.

[60] Gülseren, O., Ercolessi, F., & Tosatti, E. (1998). Noncrystalline Structures of Ultrathin

[61] Guo, X., & Xia, Y. (2011). Repulsive force vs. source number: Competing mechanisms in the yield of twinned gold nanowires of finite length. *Acta Materialia*, 59(6), 2350-7.

[62] Zhu, Y., Qin, Q., Xu, F., Fan, F., Ding, Y., Zhang, T., et al. (2012). Size effects on elas‐ ticity, yielding, and fracture of silver nanowires: In situ experiments. *Physical Review*

[63] Wu, B., Heidelberg, A., Boland, J. J., Sader, J. E., Sun, X., & Li, Y. (2006). Microstruc‐

[64] Sun, F. W., Li, H., & Liew, K. M. (2010). Compressive mechanical properties of car‐ bon nanotubes encapsulating helical copper nanowires. *Carbon*, 48(5), 1586-91.

[65] Wang, L., Zhang, H. W., Zhang, Z. Q., Zheng, Y. G., & Wang, J. B. (2007). Buckling behaviors of single-walled carbon nanotubes filled with metal atoms. *Applied Physics*

Modulus in ZnO Nanowires. *Physical Review Letters*, 96(7), 075505.

Unsupported Nanowires. *Physical Review Letters*, 80, 3775-8.

ture-Hardened Silver Nanowires. *Nano Letters*, 6, 468-72.

wires. *Nano Letters*, 9, 831-5.

394 Nanowires - Recent Advances

*B*, 85(4), 045443.

*Letters*, 91(5), 051122.

*Physical Review Letters*, 101(17), 175502.

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/52589

## **1. Introduction**

According to description, nanowires are one-dimensional materials with size ≤100 nm in two out of directions from which various architectures can be organized for recent devices offering new and sometimes unique oppurtunities. Among numerous methods can be ap‐ plied to date for densely packed nanowire (*nw*) arrays formation the template-assisted elec‐ trochemical deposition is attributed to the most widespread allowing simple control on the geometrical, morphological and crystallographic properties of various nanowire arrays in an independent manner. Note that in case of formation extremely thin and densely packed nanowires, demonstrating a significant improvement of their thermoelectric, photovoltaic, catalytic and optical properties, due to huge surface size and quantum-size effects, this pos‐ sibility becomes crucial (Bejenari et al.2011). Besides uniformity in wire diameter, morpholo‐ gy and composition, the crystallinity and crystallographic orientation also strongly influence the properties of metallic and semiconductor nanowires and their arrays (Lan et al.2009; Yan et al. 2010). However, most works to date have reported the growth of metallic and semicon‐ ductor nanowire arrays inside the alumina and polycarbonate (PC) templates with pores larger than 40 nm, especially in case of polymeric templates, and only few address the im‐ pact of pore diameter and deposition regime on the peculiarities of extremely thin nano‐ wires growth. For example, variables of the morphology, surface roughness and crystalline orientation of the Bi2Te3 nanowires with the PC membrane channel diameter decrease from 150 nm down to ~10 nm have been right now reported (Pitcht et al. 2012) demonstrating the possibility of obtaining the nanowire arrays with preferential growth of either {205}, {015}, or {110} planes perpendicular to the nanowire axis from a same composition of deposition sol‐ ution enabling us to tune their figure of merit and to improve the device performance.

© 2012 Jagminas; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Jagminas; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

In this chapter, the influence of alumina template barrier layer thickness on the electrochem‐ ical growth of copper and cobalt nanowires is demonstrated. Our results obtained investi‐ gating the peculiarities of bismuth selenide electrodeposition by alternating current (*AC*) treatment in alumina templates varied in pore diameter within 10 to 100 nm range are pre‐ sented in subsection 5.1 showing, for the first time, a strong dependency of formed *nws* com‐ position, morphology and their optical properties on the diameter of pores (*Ø*pore).

#### **2. Filling of porous alumina templates**

Porous oxide films (alumina) prepared via anodization of high pure and smooth aluminum surface in an aqueous solution of sulfuric, oxalic, and o-phosphoric acid at proper concen‐ tration, temperature, and voltage are typical templates for fabrication arrays of various nanowires in quite uniform diameter and spacing with well-defined product dimensions at packing density of 109 - 1011 species/cm2 (Masuda & Fukuda 1995; Jassensky et al. 1998; Li et al. 1998; Nielsch et al. 2002). To date high-ordered structure alumina with broad range of pore diameters as large as 300 nm (Quin et al. 2005) and as small as 5 nm (AlMawlawi et al. 1991) can be fabricated in unlimited size area. The pore diameter, cell size, and barrier-layer thickness positioned at the bottom of alumina pores (Fig. 1) are all linearly dependent on the anodizing voltage (O'Sullivan & Wood 1970), while the depth of pores can simple be con‐ trolled by anodizing time (Metzger et al. 2000). According to the final applications, the thick‐ ness of alumina templates is usually limited to 20-30 μm, but thicker templates (Römer & Steinem 2004), as well as very thin (Kokonou et al. 2007), are sometimes required for uni‐ form filling with various *nw* arrays. Till now, direct current (*DC*), alternating current (*AC*) and pulsed current depositions have been developed for filling of alumina pores by various materials. For *DC* depositions, enabling substantial control over composition and crystallini‐ ty of *nws*, the detachment of alumina film from the substrate, opening the pore bottoms via voltage decrease and chemical etching and conductive layer formation are usually required. To remove the barrier-layer only at the bottom of alumina nanochannels or perforate it, vari‐ ous etching solutions (Zheng et al. 2002) and different methods (Mardilovich et al. 1995) have been proposed during last decade. Also, the sputtering of gold (Yoo & Lee 2004), silver (Sun et al. 2005) and platinum (Panet al. 2005) layer onto the back-side of perforated tem‐ plate as a conducting and well-adhesive layer have been applied. On the other hand, there always exists the possibility of filling alumina pores by *AC* modes, leaving intact the barrierlayer at the bottom of pores because anodic aluminum oxides conduct preferentially in cathodic direction. According to some opinions (Goad & Moskovits 1978; Clebny et al. 1993), *AC* electrolysis is an ideal method for deposition of metals and semiconductors, starting from the bottom of pores. Moreover, *AC* depositions through the rectifying barrier-layer re‐ quire fewer processing steps and are more amenable to scall-up but currently provide far less control over the structure and the material deposited (Gerein & Haler 2005). As a result, different groups using this method (Preston & Moskovits 1993; Sheasby & Cook 1974) have observed interrupted growth of various polycrystalline materials and just partial deposi‐ tions, namely only in a fraction of pores. Note that for alumina templates formed at higher anodizing voltages and consequently having a thicker barrier-layer, some degree of this lay‐ er thinning is essential to enable deposition even under *AC* treatment. Furthermore, the chemical composition of alumina films differs depending on the composition of anodizing solution as well as anodizing conditions due to incorporation of acid anions and water mol‐ ecules into the outer part of alumina cells. For example, the sulfuric acid alumina films con‐ tained high amount of anion species (12-14 wt.% sulfate), while the phosphate and oxalate contents in corresponding alumina films are respectively 6-8 and 2-4 wt.% (Thompson 1997). Note that incorporated anion species produce a negative surface charge of the pore walls in‐ fluencing the pore filling process by one or another material. Therefore, the hydrophobic/ hydrophilic pore wall properties could play a significant role for nanowires growth in the precursor solution.

**Figure 1.** Typical top-side SEM view and a schematic diagram of porous alumina template.

## **3. The influence of porous alumina barrier-layer**

In this chapter, the influence of alumina template barrier layer thickness on the electrochem‐ ical growth of copper and cobalt nanowires is demonstrated. Our results obtained investi‐ gating the peculiarities of bismuth selenide electrodeposition by alternating current (*AC*) treatment in alumina templates varied in pore diameter within 10 to 100 nm range are pre‐ sented in subsection 5.1 showing, for the first time, a strong dependency of formed *nws* com‐

Porous oxide films (alumina) prepared via anodization of high pure and smooth aluminum surface in an aqueous solution of sulfuric, oxalic, and o-phosphoric acid at proper concen‐ tration, temperature, and voltage are typical templates for fabrication arrays of various nanowires in quite uniform diameter and spacing with well-defined product dimensions at

al. 1998; Nielsch et al. 2002). To date high-ordered structure alumina with broad range of pore diameters as large as 300 nm (Quin et al. 2005) and as small as 5 nm (AlMawlawi et al. 1991) can be fabricated in unlimited size area. The pore diameter, cell size, and barrier-layer thickness positioned at the bottom of alumina pores (Fig. 1) are all linearly dependent on the anodizing voltage (O'Sullivan & Wood 1970), while the depth of pores can simple be con‐ trolled by anodizing time (Metzger et al. 2000). According to the final applications, the thick‐ ness of alumina templates is usually limited to 20-30 μm, but thicker templates (Römer & Steinem 2004), as well as very thin (Kokonou et al. 2007), are sometimes required for uni‐ form filling with various *nw* arrays. Till now, direct current (*DC*), alternating current (*AC*) and pulsed current depositions have been developed for filling of alumina pores by various materials. For *DC* depositions, enabling substantial control over composition and crystallini‐ ty of *nws*, the detachment of alumina film from the substrate, opening the pore bottoms via voltage decrease and chemical etching and conductive layer formation are usually required. To remove the barrier-layer only at the bottom of alumina nanochannels or perforate it, vari‐ ous etching solutions (Zheng et al. 2002) and different methods (Mardilovich et al. 1995) have been proposed during last decade. Also, the sputtering of gold (Yoo & Lee 2004), silver (Sun et al. 2005) and platinum (Panet al. 2005) layer onto the back-side of perforated tem‐ plate as a conducting and well-adhesive layer have been applied. On the other hand, there always exists the possibility of filling alumina pores by *AC* modes, leaving intact the barrierlayer at the bottom of pores because anodic aluminum oxides conduct preferentially in cathodic direction. According to some opinions (Goad & Moskovits 1978; Clebny et al. 1993), *AC* electrolysis is an ideal method for deposition of metals and semiconductors, starting from the bottom of pores. Moreover, *AC* depositions through the rectifying barrier-layer re‐ quire fewer processing steps and are more amenable to scall-up but currently provide far less control over the structure and the material deposited (Gerein & Haler 2005). As a result, different groups using this method (Preston & Moskovits 1993; Sheasby & Cook 1974) have observed interrupted growth of various polycrystalline materials and just partial deposi‐ tions, namely only in a fraction of pores. Note that for alumina templates formed at higher

(Masuda & Fukuda 1995; Jassensky et al. 1998; Li et

position, morphology and their optical properties on the diameter of pores (*Ø*pore).

**2. Filling of porous alumina templates**


packing density of 109

396 Nanowires - Recent Advances

Potentially nanowires in the alumina template pores can be synthesized by adsorbing and decomposing precursor species, high-pressure injection of a melt, electro less and electro deposition routes. However, the dominant synthesis technique in this area remains*AC* depo‐ sition of metals and semiconductors copying exactly the pores configuration. However, the optimal *AC* electrolysis conditions differ for various solutions and various templates where‐ as under others the following phenomena as the alumina barrier-layer spalling (Sheasby & Cooke 1974), pitting corrosion (Routkevich et al. 1996) and the template peeling off from the substrate (Doughty et al. 1975) take place. Besides, the filling of alumina templates grown at higher voltages in the phosphoric or oxalic acid baths is more problematic, since at higher *AC* voltages, required in this case for metal ions discharge, the alumina barrier-layer break‐ down is inevitable. A few investigations devoted to uniform growth of metallic nanowires by means of pulse or *AC* electrolysis deal in this case with requirement to decrease the thick‐ ness of the alumina barrier-layer (*d*), (Forrer et al. 2000; Xu et al. 2002; Sauer et al. 2002). However, there is much uncertainty about the optimal *d* for uniform filling of alumina pores with metal nanowires. It is still far from clear which barrier-layer thickness and *AC* voltage value will be optimal for the uniform filling of one or another alumina template for given material. Reported in Paulus et al. study (2001) *d* values for deposition of Fe, Ni, and Co nanowires range from 10 to 20 nm. According to recent study (Sausa et al. 2011), the optimal barrier layer thickness of alumina templates for homogenous and complete filling of all pores with Ni by *AC* treatment approximates 10 nm. It is likely, however, that highly uni‐ form deposition of copper into oxalic acid grown pores by *AC* treatment is significantly more challenging than deposition into sulfuric acid grown template pores because of the dif‐ ferent chemistry and structure of their barrier-layers (Gerein & Haber 2005). Therefore, the influence of *d*is still an open issue since the composition and properties of various alumina barrier-layers are complex and ill defined, especially after thinning through voltage decrease and chemical etching.

To shed light on this problem, in this study we focused on the use of the dependency the plots of the quantity of copper deposited within the template pores on the alumina growth and subsequent *AC* deposition conditions expressed as *AC* electrolysis and the template forming voltages ratio.

The amount of copper was determined after the complete dissolution of Cu0 nanowires from a 4.5 cm2 template surface in HNO3: H2O (1:2) solution (2 cm3 ) for about 3 min. All solutions used for nanowires dissolution, sample rinsing and further double soaking in triply distilled water for 3 min were mixed together, diluted to constant volume and analyzed quantitative‐ ly using a Perkin Elmer Lambda 35 UV/Vis spectrometer. To increase absorbance detection sensitivity of copper analyte at 450 nm, 1 % sodium diethyldithiocarbamate (2.0 cm3 ) was used as a complexing agent. Reproducibility of the analysis was checked by 3 repeated ex‐ periments. Standard solutions were made from 99.999 % grade copper.

The electrochemical impedance spectroscopy (EIS) was utilized to characterize the alumina barrier-layer properties upon the anodizing bath voltage decrease. The EIS spectra were re‐ corded using a VoltaLab 80 (Radiometer Analytical, France) electrochemical system for fre‐ quencies between 1 and 105 Hz with ten measurements and are presented in the complex capacitance form, that is, the plots of Im*Y*/*ω* versus Re *Y*/*ω*, where *Y* is the admittance and *ω* is the angular frequency [*ω* = 2*πf,* and *f* is the frequency in hertz (Hz)]. These plots allow simple models of the interface to be readily inferred when the electrochemical response ex‐ hibits capacitive behaviour. The amplitude of the applied *AC* signal was set to 10 mV. The spectral data were analyzed/fitted with Z View software (Scribner Associates, South Pines, NC, USA). High-frequency stray capacitance cross-talk was eliminated from the EIS spectra as described earlier (Vanderah et al. 2003).

#### **3.1. Peculiarities of the alumina barrier-layer thinning**

It is well-documented that the thickness of the barrier-layer of alumina templates is propor‐ tional to the voltage applied in the anodizing process (Diggle et al. 1969). Going through the voltage drop, at the end of anodizing, one could expect the decrease in the barrier-layer thickness due to the field-assisted ejection of Al+3 from the oxide lattice to the solution (Harkness & Young 1966). However, this process can proceed only in some pores if a large voltage decrement is applied in a single step (Furneaux et al. 1989) causing the increased heterogeneity in the physical properties of the films. Therefore, one might expect the con‐ comitant changes in the EIS spectra reflecting these phenomena.

with metal nanowires. It is still far from clear which barrier-layer thickness and *AC* voltage value will be optimal for the uniform filling of one or another alumina template for given material. Reported in Paulus et al. study (2001) *d* values for deposition of Fe, Ni, and Co nanowires range from 10 to 20 nm. According to recent study (Sausa et al. 2011), the optimal barrier layer thickness of alumina templates for homogenous and complete filling of all pores with Ni by *AC* treatment approximates 10 nm. It is likely, however, that highly uni‐ form deposition of copper into oxalic acid grown pores by *AC* treatment is significantly more challenging than deposition into sulfuric acid grown template pores because of the dif‐ ferent chemistry and structure of their barrier-layers (Gerein & Haber 2005). Therefore, the influence of *d*is still an open issue since the composition and properties of various alumina barrier-layers are complex and ill defined, especially after thinning through voltage decrease

To shed light on this problem, in this study we focused on the use of the dependency the plots of the quantity of copper deposited within the template pores on the alumina growth and subsequent *AC* deposition conditions expressed as *AC* electrolysis and the template

used for nanowires dissolution, sample rinsing and further double soaking in triply distilled water for 3 min were mixed together, diluted to constant volume and analyzed quantitative‐ ly using a Perkin Elmer Lambda 35 UV/Vis spectrometer. To increase absorbance detection sensitivity of copper analyte at 450 nm, 1 % sodium diethyldithiocarbamate (2.0 cm3

used as a complexing agent. Reproducibility of the analysis was checked by 3 repeated ex‐

The electrochemical impedance spectroscopy (EIS) was utilized to characterize the alumina barrier-layer properties upon the anodizing bath voltage decrease. The EIS spectra were re‐ corded using a VoltaLab 80 (Radiometer Analytical, France) electrochemical system for fre‐

capacitance form, that is, the plots of Im*Y*/*ω* versus Re *Y*/*ω*, where *Y* is the admittance and *ω* is the angular frequency [*ω* = 2*πf,* and *f* is the frequency in hertz (Hz)]. These plots allow simple models of the interface to be readily inferred when the electrochemical response ex‐ hibits capacitive behaviour. The amplitude of the applied *AC* signal was set to 10 mV. The spectral data were analyzed/fitted with Z View software (Scribner Associates, South Pines, NC, USA). High-frequency stray capacitance cross-talk was eliminated from the EIS spectra

It is well-documented that the thickness of the barrier-layer of alumina templates is propor‐ tional to the voltage applied in the anodizing process (Diggle et al. 1969). Going through the voltage drop, at the end of anodizing, one could expect the decrease in the barrier-layer thickness due to the field-assisted ejection of Al+3 from the oxide lattice to the solution (Harkness & Young 1966). However, this process can proceed only in some pores if a large voltage decrement is applied in a single step (Furneaux et al. 1989) causing the increased

Hz with ten measurements and are presented in the complex

nanowires from

) was

) for about 3 min. All solutions

The amount of copper was determined after the complete dissolution of Cu0

periments. Standard solutions were made from 99.999 % grade copper.

a 4.5 cm2 template surface in HNO3: H2O (1:2) solution (2 cm3

and chemical etching.

398 Nanowires - Recent Advances

forming voltages ratio.

quencies between 1 and 105

as described earlier (Vanderah et al. 2003).

**3.1. Peculiarities of the alumina barrier-layer thinning**

Figure 2 displays a set of the EIS spectra obtained for the alumina films of different anodiza‐ tion end-voltage values. As seen, all the spectra possess a typical semicircular shape, which suggest simple capacitive behavior (Vanderah et al. 2003). This indicates the dielectric film that can be modeled as an equivalent circuit of a series of *R*s*C* elements, where *R*s is the solu‐ tion resistance and *C* is the capacitance of the alumina barrier-layer. Though being similar in shape the spectra in Fig.2 show several trends and differences. First, as the final anodizing voltage decreases the diameter of the semi-circles increases, signaling about the increase in capacitance of the barrier layer. Second, the plots of Im*Y/ω* versus Re *Y/ω* of the films ob‐ tained using high end-voltage values (10 - 40 V) tend to approach the limit:limRe*Y* / *ω* →0 *ω*→0 at

the low frequency edge of the EIS spectra indicating nearly perfect capacitive behavior. While, the films obtained using low final anodizing voltages do not. This observation sug‐ gests that the low end-voltage films exhibit significant micro-heterogeneity of the physical properties, while the opposite is true for the high end-voltage films. Taking into account these observations, for fitting the experimental EIS spectra data to the *R*s*C* model, we re‐ placed the capacitance with the constant phase element (*CPE*) as Macdonald (1987) suggest‐ ed. The *CPE* reflects the deviation of alumina impedance from the ideal behavior. In the presence of a *CPE*, the film impedance exhibits a law frequency dependence: *Z*film = (1/*C')<sup>α</sup> (iω)-α*, where *C'* is the constant *CPE*'s coefficient, *α* is the *CPE*'s exponent and *i* is the imagi‐ nary unit. The results of the fitting to this model are summarized in Table 1.As seen from Table 1, the numerical values of *C'* and *α* are consistent with the qualitative features of the Re *Y/ω* vs. Im*Y/ω* spectra, i.e., the gradual lowering of the anodizing end-voltage yields to the increase in the *CPE*'s coefficient and concomitant decrease in the CPE's exponent value.

**Figure 2.** A) *Cversus* 1/*d* plots of alumina in the range of the final anodizing voltages from 40 to 5 V calculated accord‐ ing to the impedance spectra of alumina layers grown in 0.2 mol/L H2C2O4 at 40 V for 70 min with the subsequent lowering of the anodizing voltage in a step-by-step manner to *U*a,fin (V): (1) 40 (no lowering), (2) 20, (3) 10, (4) 5, (5) 1, (6) 0.2 V, shown as Im*Y*/ω vs. Re *Y*/ω plots. Inset shows the elements of the equivalent model circuit.


**Table 1.** EIS fitting data for alumina films grown in 0.2 mol/L oxalic acid bath at 18 °C and different end-voltage values. The electrode surface geometric area is 0.33 cm2. The standard error is indicated next to the values in columns.

It is well documented, that aluminum anodizing voltage and the thickness of alumina barri‐ er-layer are related through the equation *δ*b = κ*U*a,fin where κ is the "anodizing ratio" coeffi‐ cient close to 1 nm V-1(Diggle et al. 1969). The estimated thickness values are tabulated in Table 1, column 1(numbers in the brackets). On the other hand, the capacitance and thick‐ ness of the barrier-layer are related through the following equation:

$$\mathbf{C} = \varepsilon\_0 \varepsilon \frac{A}{d} \boldsymbol{\beta}^\* \boldsymbol{\beta}^\* = \varepsilon\_0 \varepsilon \frac{A}{d} \boldsymbol{\beta} \tag{1}$$

where: *ε*0 is the vacuum permitivitty, 8.85 10-14 F/cm, *ε* is the dielectric constant of alumina, hereinafter, the value 9.8 is accepted (Harkness & Young 1966), *A* is the surface area of the electrode, 0.33 cm2 , *d* – is the thickness of the barrier-layer and *β'* is the factor that accounts for the fraction of the surface occupied by the alumina pores (0 <*β'* < 1), and *β"* is the rough‐ ness factor of the aluminum surface (*β"* > 1) (Saif et al. 2002); *β =β'β"*. Using Eq.(1) and as‐ suming the approximate equality *C*≈*C*' it is possible to verify whether or not the experimental capacitance (constant phase element) values follow the expected barrier-layer thickness sequence. Figure 2A shows the experimental dependence of the *C* vs*.d*-1, which is linear in the range 40 to about 5 nm. The slope of the line is 2.81 10-13 F cm, which yields a quite realistic value of *β*= 0.98. However, below 5 nm there is a clear deviation from linearity (not shown). These results imply that during re-anodizing the barrier-layer thickness linear‐ ly decreases with (*U*a,fin)-1 only down to about 5.0 nm. The further decrease in *U*a,fin results in slower reduction of *d* because the thickness of Al native oxide film exceeds 5 nm at room temperature (Saif et al. 2002). In other words, at *U*a,fin< 5.0 V, the chemical interaction be‐ tween aluminum and the ambient changes the relationship between the *U*a,fin and *d*. In addi‐ tion, it is likely that the constant phase element decrease from 0.98 to about 0.94 reflects the enhanced fluctuations of *d* from pore to pore, yielding more heterogeneous distribution of the physical and chemical properties of the barrier-layer.

#### **3.2. The mCu versus Uv/Ua,fin plots for copper nanowires growth**

Our experimental data suggest that a broad range of *AC* voltages can be used for deposition of copper within the alumina template pores if a suitable composition of the solution is chosen. **Final anodizing voltage, V**

400 Nanowires - Recent Advances

40 (40) 20 (20) 10 (10) 5 (5) 1 (1) 0.2 (0.2)

electrode, 0.33 cm2

**(approximate thickness of the barrier in nm) C', μF <sup>α</sup>**

**Table 1.** EIS fitting data for alumina films grown in 0.2 mol/L oxalic acid bath at 18 °C and different end-voltage values. The electrode surface geometric area is 0.33 cm2. The standard error is indicated next to the values in columns.

> 0 0 ' " *A A <sup>C</sup> d d* = =

ee bb ee b

ness of the barrier-layer are related through the following equation:

the physical and chemical properties of the barrier-layer.

**3.2. The mCu versus Uv/Ua,fin plots for copper nanowires growth**

It is well documented, that aluminum anodizing voltage and the thickness of alumina barri‐ er-layer are related through the equation *δ*b = κ*U*a,fin where κ is the "anodizing ratio" coeffi‐ cient close to 1 nm V-1(Diggle et al. 1969). The estimated thickness values are tabulated in Table 1, column 1(numbers in the brackets). On the other hand, the capacitance and thick‐

where: *ε*0 is the vacuum permitivitty, 8.85 10-14 F/cm, *ε* is the dielectric constant of alumina, hereinafter, the value 9.8 is accepted (Harkness & Young 1966), *A* is the surface area of the

for the fraction of the surface occupied by the alumina pores (0 <*β'* < 1), and *β"* is the rough‐ ness factor of the aluminum surface (*β"* > 1) (Saif et al. 2002); *β =β'β"*. Using Eq.(1) and as‐ suming the approximate equality *C*≈*C*' it is possible to verify whether or not the experimental capacitance (constant phase element) values follow the expected barrier-layer thickness sequence. Figure 2A shows the experimental dependence of the *C* vs*.d*-1, which is linear in the range 40 to about 5 nm. The slope of the line is 2.81 10-13 F cm, which yields a quite realistic value of *β*= 0.98. However, below 5 nm there is a clear deviation from linearity (not shown). These results imply that during re-anodizing the barrier-layer thickness linear‐ ly decreases with (*U*a,fin)-1 only down to about 5.0 nm. The further decrease in *U*a,fin results in slower reduction of *d* because the thickness of Al native oxide film exceeds 5 nm at room temperature (Saif et al. 2002). In other words, at *U*a,fin< 5.0 V, the chemical interaction be‐ tween aluminum and the ambient changes the relationship between the *U*a,fin and *d*. In addi‐ tion, it is likely that the constant phase element decrease from 0.98 to about 0.94 reflects the enhanced fluctuations of *d* from pore to pore, yielding more heterogeneous distribution of

Our experimental data suggest that a broad range of *AC* voltages can be used for deposition of copper within the alumina template pores if a suitable composition of the solution is chosen.

, *d* – is the thickness of the barrier-layer and *β'* is the factor that accounts

0.065 ± 0.001 0.139 ± 0.002 0.288 ± 0.004 0.560 ± 0.006 0.762 ± 0.010 1.440 ± 0.032

± 0.002 ± 0.002 ± 0.003 ± 0.004 ± 0.005 0.943 ± 0.005

(1)

Unfortunately, this is applicable only for alumina grown in sulfuric acid bath where *AC* voltag‐ es *U*v 6.0 to 18.0 V can be successfully applied (Fig. 3, curve 2). All attempts to fill more complete‐ ly alumina templates grown in oxalic and phosphoric acid solutions varying *AC* voltage and the deposition time have failed since only limited voltages can be used in these cases. At higher *AC* voltages or somewhat longer electrolysis durations the barrier-layer breakdown of these tem‐ plates was found to be inevitable (curves 1 in Figs. 4 and 5). As seen from curve 1 in Fig. 3, it is al‐ so a common practice for alumina grown in sulfuric acid bath if *d* is increased at the end of anodizing process up to about 20 nm by an increase of *U*a,fin to 20 V. On the other hand, it was es‐ tablished that the deposition of copper proceeds easier in the same Cu(II) solution if the barrierlayer thickness of alumina is decreased by lowering *U*a. These results are in line with former findings by other investigators while depositing Au (Forrer et al. 2000), Ag (Xu et al. 2002; Sau‐ er et al. 2002) and Fe, Ni, Co (Paulus et al. 2001). The *m*Cu against *U*v/*U*a,fin plots for alumina tem‐ plates grown in oxalic and phosphoric acid solutions are shown in Figs. 4 and 5, respectively. From these plots one can see that the behavior of alumina templates in acidic Cu(II)solution manifests itself through parabolic *m*Cuvs*.U*v/*U*a,fin dependencies. Moreover, we have found that a range of *AC* voltages at which deposition of copper nanowires proceeds within the alumina pores strongly depends on the *U*a,fin, decreasing with the alumina barrier-layer thickness. It should be also noted, that at *AC* voltages higher than *m*Cuvs.*U*v/*U*a,fin curve peak, (*U*v/*U*a,fin)peak, the alumina spalling and peeling off from the substrate becomes critical, especially when *U*a,fin> 20 V.Therefore, attention was focused on the range of *AC* voltages suitable for Cu0 nanowires growth without the barrier-layer breakdown. As clearly seen from the experimental results de‐ picted for various templates in Figs. 3 to 5, the range of voltages suitable for copper deposition shifts to the higher ratio of *U*v/*U*a,fin, the lower *U*a,fin. On the other hand, despite the fact that a much wider range of *AC* voltages can be used for copper deposition within the pores of alumi‐ na with a quite thin barrier-layer, it seems impossible to completely fill such a matrix even at the *U*v/*U*a,fin ratio 3.0 when *d*< 5 nm (*U*a,fin< 5 V). This is due to a sharp decay in the current strength even during first 10-20 s of *AC* electrolysis up to a few mA cm-2; characteristic feature for films with *d* lower than 5.0 nm. In these cases only imperceptible quantity of copper can be deposited within the alumina pores at the electrode edges since the gas bubbles evolved at the Al│oxide interface push away the alumina film from the substrate. The appearance of the gas bubbles at the Al│oxide interface was clearly observed by the naked eye. It is also common for alumina templates having thicker barrier-layers if prolonged *AC* electrolysis and higher *AC* voltages are used. Consequently, it seems difficult to estimate one optimal *U*v/*U*a,fin for Cu0 nanowires depo‐ sition within the pores of different templates. Evidently, the optimal *AC* voltage is lower (*U*v/ *U*a,fin)peak and the final choice is up to uniformity and completeness of the filling of the alumina pores. On the other hand, it has been found that the range of *AC* voltages suitable for copper nanowires fabrication depends also on the composition of Cu(II) solution. The *m*Cuvs.*U*v/*U*a,fin plots presented in Fig. 6 show that an increase in the solution pH widens the range of *AC* voltag‐ es suitable for copper deposition. Although the reasons of such behavior are unclear yet, we be‐ lieved that this is most probably accomplished by a sharp decrease in the gas quantities evolved simultaneously with copper deposition from neutral and alkaline Cu(II) solutions at the Al│oxide boundary (Jagminas et al. 2002).

**Figure 3.** *m*Cuvs*.U*v/*U*a,fin plots for copper nanowires growth for 3 min within the pores of alumina with various barrierlayer thickness attained by increasing (1) or lowering (3) anodizing voltage in 1.53 mol/L sulfuric acid bath at the end of the template growth to *U*a,fin (V): (1) 20, (2) 15, (3) 7.0. An acidic solution of 0.1 CuSO4 + 0.05 mol/L MgSO4 + H2SO4 down to pH 1.50 was used for Cu0 nanowires growth. The alumina thickness,δ = 10 μm. The dashed lines correspond to the obvious damage of alumina template.

**Figure 4.** The same as in Fig. 3 for alumina grown in 0.2 mol/L oxalic acid bath for 70 min (δ=5.0 μm) with reduced *d* by lowering *U*a 40 Vto *U*a,fin (V): (1) 30, (2) 20, (3) 10, and (4) 7.0.

How and Why Alumina Matrix Architecture Influence the Shape and Composition of Nanowires Grown... http://dx.doi.org/10.5772/52589 403

**Figure 5.** The same as in Fig. 3 for alumina grown in 0.4 mol/L phosphoric acid bath at 80 V for 1.5 h (δ=5.0 μm) with reduced *d* by lowering *U*ato *U*a,fin (V): (1) 80, (2) 40, (3) 20, (4) 13, (5) 10, (6) 7.0.

**Figure 3.** *m*Cuvs*.U*v/*U*a,fin plots for copper nanowires growth for 3 min within the pores of alumina with various barrierlayer thickness attained by increasing (1) or lowering (3) anodizing voltage in 1.53 mol/L sulfuric acid bath at the end of the template growth to *U*a,fin (V): (1) 20, (2) 15, (3) 7.0. An acidic solution of 0.1 CuSO4 + 0.05 mol/L MgSO4 + H2SO4 down to pH 1.50 was used for Cu0 nanowires growth. The alumina thickness,δ = 10 μm. The dashed lines correspond

**Figure 4.** The same as in Fig. 3 for alumina grown in 0.2 mol/L oxalic acid bath for 70 min (δ=5.0 μm) with reduced *d*

to the obvious damage of alumina template.

402 Nanowires - Recent Advances

by lowering *U*a 40 Vto *U*a,fin (V): (1) 30, (2) 20, (3) 10, and (4) 7.0.

**Figure 6.** *m*Cuvs*.U*v/*U*a,fin plots for copper nanowires growth within the pores of alumina for 3 min *AC* electrolysis in the solution containing: 0.1 CuSO4 + 0.05 mol/L MgSO4 + H2SO4 down to pH 1.0 (1), 1.50 (2), and 6.0 (3) adjust by adding 0.2 M TEA. The alumina was grown in 0.4 mol/L phosphoric acid bath at 80 V for 1.5 h. *U*a,fin 20 V.

#### **3.3. Concluding remarks**

The above results show that in order to grow copper nanowires within the pores of alumina template obtained by Al anodizing at higher voltages the alumina barrier-layer thickness should be lowered. Using the acidic anodizing baths, a linear dependence of *d* on the final anodizing voltage, *U*a,fin, is observed down to 5 V. The linearity was verified by impedance spectroscopy data, so, this technique might be used to monitor the parameters of the alumi‐ na template formation. However, further *U*a,fin decrease below 5 V results in significant barri‐ er-layer thickness fluctuations, which are possibly due to uneven native oxide formation at the bottom of the pores. Spectrometric analysis of deposited copper content has shown that the range of AC voltage suitable for copper nanowires growth within the alumina pores sharply increases with decrease in *U*a,fin and hence *d*. The most favourable *U*a,fin range for uniform copper nanowires growth is 15 to 7.0 V.

## **4. Alumina template-dependant growth of cobalt nanowire arrays by** *AC* **deposition**

#### **4.1. Depositions into as-grown templates**

In this study, different electrochemical regimes and porous alumina fabricated by aluminum anodizing in either sulphuric or oxalic acid solutions were applied for template synthesis of cobalt nanowire arrays, revealing several peculiar cases. By this way, we found that the growth of cobalt nanowires depends much stronger on the conditions for fabricating the alumina template than other metals like copper, silver or tin. For example, only sulfuric acid alumina templates can be successfully filled in the optimized deposition solutions by Co *nws* using *AC*, while the use of the same solution for uniform growth of Co *nws* in oxalic or phosphoric acid alumina pores is problematic. Therefore, in this study we focus on the in‐ vestigation the peculiarities of the Co *nws* electrochemical growth using oxalic and sulfuric acid alumina templates at different regimes.

In this study, the composition of solution for Co *nws* electrosynthesis within the alumi‐ na pores was organized using CoSO4, as a precursor for cobalt ions, and H3BO3, as a buf‐ fering ingredient, to prevent any pH variation within the alumina pores and to inhibit the formation of cobalt hydroxide species, as a result of hydrogen evolution (Zech & Landolt 2000). Furthermore, MgSO4 was added as Mg2+ source to prevent the break‐ down of alumina barrier-layer (Jagminas et al. 2003) during *AC* depositions. To adjust the pH of solutions, H2SO4 and triethanolamine(TEA) 1:1 solutions in water were used. From a review of patent's literature (Herrman 1972), arises that these components have been frequently used for *AC* coloring of anodized aluminum in the Co2+-containing baths. To optimize the concentration of components and the solution pH, 15 μm thick sulfuric acid alumina templates were applied. All depositions in this setup were performed us‐ ing 50 cm2 specimens, a 50 Hz frequency *AC* and constant peak-to-peak voltage of 32 V, for 15 min. Observation of the alumina color uniformity and the amount of deposited co‐ balt were the main criteria adopted in the search of the optimal composition. Conse‐ quently, the solution composed of 0.2-0.25 CoSO4, 0.65-0.7 H3BO3, 0.005-0.05 mol/L MgSO4 and TEA, to adjust a pH at 5.5-5.75, was determined as optimum for *AC* filling of alumina templates with Co *nws*, having average *Ø*pore of 15 nm. The XRD patterns (not shown herein) of Co *nws* deposited from this solution demonstrated the polycrystalline nature of cobalt in hexagonal closely packed lattice. For any of the used *AC* deposition conditions no phase modification in the XRD patterns was observed. Figure 7 demon‐ strates typical variation of the amount of Co assembled inside the alumina pores (*m*Co) with the deposition time and *AC* current frequency (*f*) used for depositions. As seen, the amount of deposited Co increases linearly during the first 10-20 min of *AC* treatment at constant voltage. However, with further processing the rate of deposition inside the alu‐ mina pores progressively decreases. Moreover, it was observed that this solution allows the growth of cobalt *nws* within a wide range of *AC* frequencies, ca. from 10 to 200 Hz, coloring the template uniformly from bronze to deep black. A further increase in fre‐ quency, up to 1000 Hz, results in a smaller amount of deposited cobalt and therefore in a lighter template color intensity. As seen from the inset of Fig. 7, the maximum amount of cobalt can be deposited using 100 Hz frequency. The bath temperature within 10 to 40°C range was found to have negligible effect on the amount of deposited cobalt as well as on the uniformity of depositions.

**3.3. Concluding remarks**

404 Nanowires - Recent Advances

**deposition**

uniform copper nanowires growth is 15 to 7.0 V.

**4.1. Depositions into as-grown templates**

acid alumina templates at different regimes.

The above results show that in order to grow copper nanowires within the pores of alumina template obtained by Al anodizing at higher voltages the alumina barrier-layer thickness should be lowered. Using the acidic anodizing baths, a linear dependence of *d* on the final anodizing voltage, *U*a,fin, is observed down to 5 V. The linearity was verified by impedance spectroscopy data, so, this technique might be used to monitor the parameters of the alumi‐ na template formation. However, further *U*a,fin decrease below 5 V results in significant barri‐ er-layer thickness fluctuations, which are possibly due to uneven native oxide formation at the bottom of the pores. Spectrometric analysis of deposited copper content has shown that the range of AC voltage suitable for copper nanowires growth within the alumina pores sharply increases with decrease in *U*a,fin and hence *d*. The most favourable *U*a,fin range for

**4. Alumina template-dependant growth of cobalt nanowire arrays by** *AC*

In this study, different electrochemical regimes and porous alumina fabricated by aluminum anodizing in either sulphuric or oxalic acid solutions were applied for template synthesis of cobalt nanowire arrays, revealing several peculiar cases. By this way, we found that the growth of cobalt nanowires depends much stronger on the conditions for fabricating the alumina template than other metals like copper, silver or tin. For example, only sulfuric acid alumina templates can be successfully filled in the optimized deposition solutions by Co *nws* using *AC*, while the use of the same solution for uniform growth of Co *nws* in oxalic or phosphoric acid alumina pores is problematic. Therefore, in this study we focus on the in‐ vestigation the peculiarities of the Co *nws* electrochemical growth using oxalic and sulfuric

In this study, the composition of solution for Co *nws* electrosynthesis within the alumi‐ na pores was organized using CoSO4, as a precursor for cobalt ions, and H3BO3, as a buf‐ fering ingredient, to prevent any pH variation within the alumina pores and to inhibit the formation of cobalt hydroxide species, as a result of hydrogen evolution (Zech & Landolt 2000). Furthermore, MgSO4 was added as Mg2+ source to prevent the break‐ down of alumina barrier-layer (Jagminas et al. 2003) during *AC* depositions. To adjust the pH of solutions, H2SO4 and triethanolamine(TEA) 1:1 solutions in water were used. From a review of patent's literature (Herrman 1972), arises that these components have been frequently used for *AC* coloring of anodized aluminum in the Co2+-containing baths. To optimize the concentration of components and the solution pH, 15 μm thick sulfuric acid alumina templates were applied. All depositions in this setup were performed us‐ ing 50 cm2 specimens, a 50 Hz frequency *AC* and constant peak-to-peak voltage of 32 V, for 15 min. Observation of the alumina color uniformity and the amount of deposited co‐

**Figure 7.** Variation of the amount of Co (*m*Co) inside the alumina pores for the deposition solution made of 0.2 CoSO4, 0.7 H3BO3 and 0.01 mol/L MgSO4 and TEA up to pH 5.7, at a constant *AC* voltage *U*p-p of 32 V, 50 Hz and 20 oC, as a function of the deposition time. *Ø*pore 15 nm; template thickness (δAAO) 15 μm. In the inset: variation of *m*Co versus *AC* frequency (*f*) for the same deposition solution and *U*p-p value for 15 min of deposition.

**Figure 8.** Cross-sectional FESEM images of sulfuric acid alumina templates grown at 25 (A, B) and 15 V (C) following the template etching in 2 mol/L H2SO4 for 15 min and deposition of cobalt *nw* arrays by *AC* (50 Hz) treatment in a same as in Fig. 7 solution at a constant current density of 0.3 A dm-2 and room temperature for: 5 (A), 15 (B) and 60 min (C). δAAO =18 μm, *Ø*pore~28 (A,B) and ~20 nm (C).

In order to visualize the uniformity of Co *nws* growth by *AC* deposition, cross-sections of the alumina templates were investigated using field emission scanning electron microscopy (FESEM). Figures 8A-C show the arrangement of Co *nws* deposited inside the pores of sulfu‐ ric acid alumina templates at the same *AC* frequency (50 Hz) and peak-to-peak voltage (*U*p-p) during 5, 15 and 60 min, respectively. The obtained data show quite uniform growth of Co *nws* from the bottom of almost all pores only at the onset of the process. The pores filling rate depends on *AC* voltage and of the pores diameter, *Ø*pore. In case of sulfuric acid alumina templates formed at 25 V (*Ø*pore 25-30 nm after pore widening), the uniform growth corre‐ sponds to a filling rate *v*Co~ 6 μm/h for the first 10 min of deposition at a constant *U*p-pof 32 V. Processing further, *v*Co decreases due to an increase of the template resistance, causing a reduced *AC* current. A smaller *Ø*pore results in a faster growth of cobalt *nws* under the same *AC* treatment conditions. For example, *v*Co~ 9.3 μm/h was detected at *U*p-pof32 V for alumina templates with average *Ø*pore of 15 nm. It can be observed from Fig. 8B that some cobalt *nws* grew faster than *nws* front. For prolonged *AC* treatment, this progressively leads to the for‐ mation of more and more uneven lengths of Co *nws* emerging onto the template surface in cobalt caps ( Fig. 8C). The *nws* height uniformity was found to be independent on the *f.* Typ‐ ical morphology of cobalt *nw* arrays assembled inside the sulfuric acid alumina pores by short-term AC deposition after template etching is presented in Figure 9. As seen, in case of 1-2 μm length of Co nws they are densely packed and quite uniform.

#### **4.2. Depositions through a reconstructed barrier-layer**

**Figure 8.** Cross-sectional FESEM images of sulfuric acid alumina templates grown at 25 (A, B) and 15 V (C) following the template etching in 2 mol/L H2SO4 for 15 min and deposition of cobalt *nw* arrays by *AC* (50 Hz) treatment in a same as in Fig. 7 solution at a constant current density of 0.3 A dm-2 and room temperature for: 5 (A), 15 (B) and 60

In order to visualize the uniformity of Co *nws* growth by *AC* deposition, cross-sections of the alumina templates were investigated using field emission scanning electron microscopy (FESEM). Figures 8A-C show the arrangement of Co *nws* deposited inside the pores of sulfu‐ ric acid alumina templates at the same *AC* frequency (50 Hz) and peak-to-peak voltage (*U*p-p) during 5, 15 and 60 min, respectively. The obtained data show quite uniform growth of Co *nws* from the bottom of almost all pores only at the onset of the process. The pores filling rate depends on *AC* voltage and of the pores diameter, *Ø*pore. In case of sulfuric acid alumina templates formed at 25 V (*Ø*pore 25-30 nm after pore widening), the uniform growth corre‐ sponds to a filling rate *v*Co~ 6 μm/h for the first 10 min of deposition at a constant *U*p-pof 32 V. Processing further, *v*Co decreases due to an increase of the template resistance, causing a reduced *AC* current. A smaller *Ø*pore results in a faster growth of cobalt *nws* under the same *AC* treatment conditions. For example, *v*Co~ 9.3 μm/h was detected at *U*p-pof32 V for alumina templates with average *Ø*pore of 15 nm. It can be observed from Fig. 8B that some cobalt *nws* grew faster than *nws* front. For prolonged *AC* treatment, this progressively leads to the for‐ mation of more and more uneven lengths of Co *nws* emerging onto the template surface in cobalt caps ( Fig. 8C). The *nws* height uniformity was found to be independent on the *f.* Typ‐ ical morphology of cobalt *nw* arrays assembled inside the sulfuric acid alumina pores by short-term AC deposition after template etching is presented in Figure 9. As seen, in case of

1-2 μm length of Co nws they are densely packed and quite uniform.

min (C). δAAO =18 μm, *Ø*pore~28 (A,B) and ~20 nm (C).

406 Nanowires - Recent Advances

The key feature of *AC* deposition process of the densely packed Co *nws* is that only sulfuric acid alumina templates can be successfully applied. Fabrication of Co *nws* in the nanochan‐ nels of alumina template formed in the oxalic or phosphoric acid anodizing baths, however, is problematic by*AC* deposition way. One possible explanation for this effect is the adsorp‐ tion and incorporation of acid anions at some depth of the alumina barrier-layer, changing the state of alumina/solution interface at the bottom of pores (surface charge, free energy, etc.) and preventing the discharge of Co2+ ions. Besides, highly ordered oxalic and phos‐ phoric acid alumina templates are usually formed at higher voltages (Masuda & Fukuda 1995; Masuda et al. 1997; Li et al. 2000) and, therefore, present much thicker barrier-layers at the metal│oxide interface. To use these templates for *AC* deposition of various materials, the step-wise voltage decreasing at the end of anodizing process has been proposed (Fur‐ neaux et al. 1989) and successfully used in several works. We found herein that this is help‐ ful also for the Co case, however, only for short-time processing. The Co *nw* array produced by long*AC* treatment, i.e. longer than 15 min, viewed nonuniform from pore to pore with some mushroomed Co fragments (Fig. 10) outgrowing from the breakdown sites of the alu‐ mina barrier-layer. The modification of alumina barrier-layers through (i) the decreasing of anodizing voltage (*U*a ) at the end of oxalic acid alumina growth down within 13 to 5 V, (ii) the cathodic treatment in the same anodizing bath for 3 min at various potentials and (iii) the chemical etching in the solution of sulfuric acids inhibit the uniformity of the deposi‐ tions (see Fig. 11). The most uniform alumina color was obtained after decreasing *U*a and chemical etching in the solution of sulfuric acid. Nevertheless, in this case the SEM crosssectional observations of templates revealed the formation of Co *nws* tufts in random areas of template (Fig. 12A). It is worth to note that these tufts were found to arise from cobalt balls (Figs 11B and 12 C) formed at the metal/template interface.

**Figure 9.** Top view FESEM images for fragments of Co *nw* arrays fabricated under the optimized *AC* deposition condi‐ tions in the sulfuric acid alumina template pores after the template etching in 0.5 mol/L H3PO4 for 70 (A) and 100 min (B).

**Figure 10.** Typical top view SEM images of Co 'mushroom' grown by *AC* treatment of oxalic acid alumina template in the solution of this study. Before deposition the thickness of alumina barrier- layer was reduced by lowering the anod‐ izing voltage down to final value (*U*a,fin) equal to 15 V.

**Figure 11.** Top view FESEM image for fragments of Co *nw* arrays fabricated in the oxalic acid alumina pores by *AC* (50 Hz) deposition from the optimized solution at *j*ac ~ 0.5 A dm-2 (*U*p-p from 18 to 32 V) and room temperature for 10 (A) and 20 min (B). Before depositions, the barrier-layer of the as-grown templates was thinned by decreasing anodizing voltage down to 20 V and etching in 0.5 mol/L H3PO4 at 30 oC for 32 min. Following depositions the template was etched in the same 0.5 mol/L H3PO4 solution.

Variations in the conditions of cathodic treatment and chemical etching of the oxalic acid template as well as variations of the deposition potential were found to be ineffective for rod-like Co formation by *AC* deposition through the remained barrier-layer. Moreover, in case of *DC* deposition, the detachment of alumina template from the substrate even after several minutes of treatment took place.

#### **4.3. Galvanostatic DC deposition**

An alternative approach for Co *nws* deposition was further examined for oxalic acid alumi‐ na templates by a *DC* constant current density deposition, after removing or perforating the barrier-layer on the pores bottoms. In this setup, we used either an electrochemical/chemical method for the barrier layer perforation, or we detached the alumina from the substrate, re‐ moved the barrier layer and made a subsequent evaporation of Cr/Au layer, acting later as a conducting pad. The resulting Co *nws* released from the as-filled templates are shown in im‐ ages A and B of Fig. 13. We found here that by applying a low current density during the entire deposition process, ca. ≤ 0.2 mA cm-2, well-ordered, densely packed, continuous and highly aligned Co *nw* arrays, faithfully reproducing the shape of the pores and with height up to several tens of micrometers, can be synthesized by this way. Furthermore, after alumi‐ na dissolution these *nws* seems not to collapse and stuck together, as in the case of *AC* syn‐ thesis, implying an easier their application in future nanoelectronics and novel efficient sensors.

**Figure 10.** Typical top view SEM images of Co 'mushroom' grown by *AC* treatment of oxalic acid alumina template in the solution of this study. Before deposition the thickness of alumina barrier- layer was reduced by lowering the anod‐

**Figure 11.** Top view FESEM image for fragments of Co *nw* arrays fabricated in the oxalic acid alumina pores by *AC* (50 Hz) deposition from the optimized solution at *j*ac ~ 0.5 A dm-2 (*U*p-p from 18 to 32 V) and room temperature for 10 (A) and 20 min (B). Before depositions, the barrier-layer of the as-grown templates was thinned by decreasing anodizing voltage down to 20 V and etching in 0.5 mol/L H3PO4 at 30 oC for 32 min. Following depositions the template was

Variations in the conditions of cathodic treatment and chemical etching of the oxalic acid template as well as variations of the deposition potential were found to be ineffective for rod-like Co formation by *AC* deposition through the remained barrier-layer. Moreover, in case of *DC* deposition, the detachment of alumina template from the substrate even after

An alternative approach for Co *nws* deposition was further examined for oxalic acid alumi‐ na templates by a *DC* constant current density deposition, after removing or perforating the barrier-layer on the pores bottoms. In this setup, we used either an electrochemical/chemical method for the barrier layer perforation, or we detached the alumina from the substrate, re‐ moved the barrier layer and made a subsequent evaporation of Cr/Au layer, acting later as a conducting pad. The resulting Co *nws* released from the as-filled templates are shown in im‐ ages A and B of Fig. 13. We found here that by applying a low current density during the

izing voltage down to final value (*U*a,fin) equal to 15 V.

408 Nanowires - Recent Advances

etched in the same 0.5 mol/L H3PO4 solution.

several minutes of treatment took place.

**4.3. Galvanostatic DC deposition**

**Figure 12.** SEM images of reconstructed oxalic acid alumina templates after deposition of Co *nws* by *AC* treatment in the solution as in Fig. 7 at *U*p-p 32 V and room temperature for 20 min. δAAO 13 μm, *Ø*pore~ 45 nm, τw 30 min in 0.5 mol/L H3PO4. (A) top-side view after dissolution some part of alumina; (B) *in plane* view; (C) cross-sectional view.

A typical XRD profile of a template filled with Co *nws* via *DC* deposition at a constant cur‐ rent density of 0.12 mA cm-2 for 5 hours is shown in Figure 14. Only a single peak is observa‐ ble at 2*Θ* = 41.59. According to the XRD library patterns for bulk Co (PDF 89-4308), this peak corresponds to the (100) reflection of the hexagonal closely packed Co lattice. Further, some additional weak signal situated at 2*Θ* = 75.89, ascribed to hexagonal Co phase in (110) direc‐ tion, can be observed. This weak feature probably inferred that Co *nws* are not single crys‐ tals but consist of oriented polycrystals with a preferred (100) growth direction perpendicular to the substrate. We note that the preferential growth of hexagonal Co phase in (100) direction is not a trivial case and differs from the growth of Co *nws* inside the alumi‐ na pores via *AC* and potentiostatic depositions reported by Kartopu et al. (2008) where the formation of Co *nw* arrays with a preferred (110) orientation has been demonstrated.

**Figure 13.** Top-view FESEM images of Co *nw* arrays after dissolution of oxalic acid alumina template fully (A )and in part (B). Template fabrication: anodizing 3 hours; *U*a,fin 5.0 V; τw 30 min in 0.5 mol/L H3PO4. Following the detachment of aluminafrom the substrate and a back-side evaporation of a Cr/Au layer, the deposition of Co was conducted in the solution: 0.45 CoSO4, 0.7 mol/L H3BO3 and TEA up to pH 5.7 at a constant *DC*current density of 0.12 mA cm-2 for 5 hours.

**Figure 14.** Typical X-ray diffraction spectrum for a Co *nws* array fabricated inside the oxalic acid alumina pores by DC deposition under the same galvanostatic conditions as in Fig. 13.

## **5. Compositional, structural and optical properties of bismuth selenide nws synthesized by template approach**

Bulk bismuth selenide, Bi2Se3, is a V/VI semiconducting material with a band gap of ~ 0.35 eV and belongs to semiconductors group with a set of optical and physical properties which permits their use in photosensitive, photoelectrochemical and optoelectronic devices. In onedimensional structures, especially when the nanometric dimensions turn down to Bohr radi‐ us (Bi2Se3*r*Bh~ 8 nm: Hillhouse & Tuominen 2001), these properties can be drastically enhanced due to quantum confinement effects (Li & Wang 2005). In the past decade, these effects have stimulated extremely active investigations on creating methods for fabrication of densely packed arrays of semiconductor species varied in composition, nm-scaled dimen‐ sions, shape and spacing.

Chemical deposition (Pejova & Grozdanov 2002; Gracia et al. 1997), electrodeposition (Tor‐ ene et al. 1998), vapor deposition (Giani et al. 2002), reactive and compound evaporation (Augustine et al. 2005) and others (Sankapal et al. 2002) have been employed to date for the formation of Bi2Se3 thin films demonstrating their different optical properties and band gap values varied up to 2.3 eV.

During the last decade, several methods have also been proposed for the synthesis of nmscaled bismuth selenide products. Yang et al. (2005) reported a solvothermal procedure for nanotubes and nanorods fabrication. Bi2Se3nanosheets and nanotubes have also been syn‐ thesized by hydrothermal co-reduction (Cui et al. 2004a) and nanorods(Cui et al. 2004b) and nanobelts(Xu et al. 2005) by a photochemical and sonochemical routes, respectively. The preparation of Bi2Se3nanocrystals has been also reported by Wang's group (2003). Recently, we have proposed a simple way for electrochemical formation of crystalline Bi2Se3 nanowire arrays inside the alumina template pores with an average pore diameter, *Ø*pore= 40-50 nm (Jagminas et al. 2008). By this way, densely packed *nw* arrays from quite pure Bi2Se3 *nws* at a high aspect ratio can be successfully fabricated.

**Figure 13.** Top-view FESEM images of Co *nw* arrays after dissolution of oxalic acid alumina template fully (A )and in part (B). Template fabrication: anodizing 3 hours; *U*a,fin 5.0 V; τw 30 min in 0.5 mol/L H3PO4. Following the detachment of aluminafrom the substrate and a back-side evaporation of a Cr/Au layer, the deposition of Co was conducted in the solution: 0.45 CoSO4, 0.7 mol/L H3BO3 and TEA up to pH 5.7 at a constant *DC*current density of 0.12 mA cm-2 for 5

**Figure 14.** Typical X-ray diffraction spectrum for a Co *nws* array fabricated inside the oxalic acid alumina pores by DC

**5. Compositional, structural and optical properties of bismuth selenide**

Bulk bismuth selenide, Bi2Se3, is a V/VI semiconducting material with a band gap of ~ 0.35 eV and belongs to semiconductors group with a set of optical and physical properties which permits their use in photosensitive, photoelectrochemical and optoelectronic devices. In onedimensional structures, especially when the nanometric dimensions turn down to Bohr radi‐ us (Bi2Se3*r*Bh~ 8 nm: Hillhouse & Tuominen 2001), these properties can be drastically enhanced due to quantum confinement effects (Li & Wang 2005). In the past decade, these effects have stimulated extremely active investigations on creating methods for fabrication

deposition under the same galvanostatic conditions as in Fig. 13.

**nws synthesized by template approach**

hours.

410 Nanowires - Recent Advances

In this article, we report the results of investigation on peculiarities of bismuth selenide elec‐ trode position by*AC* treatment in different alumina templates varied in pore diameter, *Ø*pore, within 10 to 100 nm range demonstrating, for the first time, a strong dependency of formed *nws* composition, morphology and their optical properties on the *Ø*pore.

Porous alumina templates, 5.0 to 10.0 μm thick, were grown *via* two-step direct current (*dc*) anodizing of specimens for 0.5 to 20 hours in thermostated and vigorously stirred solutions under conditions indicated in Table 2. First anodization lasted two hours following the alu‐ mina film stripping in 0.2 M CrO3 + 0.7 M H3PO4 at 60 ºC for 5 hours, while the second one – as indicated in Table 2. The barrier layer of alumina films after the second anodizing was set to about 15 nm by reduction of the anodizing voltage step-wise (2-1 V per 30-60 s) as in the study of Furneaux et al. (1989). All depositions of bismuth selenide *nws* within the alumina template pores were performed at room temperature in a glass cell where two graphite rods were used as the auxiliary and Al/alumina as a working electrode. The solution containing 0.02Bi2(SO4)3 and 0.024 mol/L of H2SeO3 as the sources for BixSey *nws* growth and two chelat‐ ing agents for Bi3+, e.g. 0.25 [TEA] and 0.07 mol/L [EDTA], kept at pH between 5.0 and 5.3 by addition of H2SO4 (1:1) was used in this study. The ratio of selenious acid to bismuth salt concentration was approximated to 1.2. This value has been recently determined by us as optimal for stoichiometric Bi2Se3*nws* synthesis inside the pores of oxalic acid alumina tem‐ plates with *Ø*pore = 40-50 nm (Jagminas et al. 2008). For depositions, an alternating current, 50 Hz in frequency, centered at 0 V, under constant *AC* current density (*j*ac) control of 0.3 ± 0.02 A dm-2 was applied. Reagents for the preparation of deposition solutions were: bismuth(III) sulphate, Bi2(SO4)3, triethanolamine, N(C2H4OH)3 – TEA, ethylendiaminetetracetic acid, 2(HOOCH2C)NCH2CH2N(CH2COOH)2 - EDTA, and selenious acid, H2SeO3, of analytical re‐ agent quality, purchased from Aldrich, and water from Milli-Q water system.


**Table 2.** Summary of anodizing and post treatment variables employed for fabrication and etching of alumina templates.

To increase crystallinity of the final-products, samples were annealed for 3 hours in vacuum. The optimal annealing temperatures (*T*ann) for various alumina templates were found in this study experimentally and approximated to: 250 o C for 100 and 50 nm, 200 o C for 28 and 13 nm and 170 o C for 10 nm *Ø*pore templates. The different values of *T*ann were chosen due to the well-known dependency of melting and crystallization temperatures of nanomaterials on their size (Noh et al. 2007).

The alumina templates intended for XRD and UV-vis-IR investigations were separated from the electrode surface by one-side sequential etching of the electrode window in a solution of 1.5 mol/LNaOHand then in 10 wt% HCl and 0.1 mol/L CuCl2 followed by thorough rinsing and drying in a nitrogen stream. X-ray diffraction studies were performed with a diffrac‐ tometer D8 (Bruker AXS, Germany) equipped with a Göbel mirror (primary beam mono‐ chromator) for CuKα radiation. A step-scan mode was used in the 2*Θ* range from 18 to 55o with a step of 0.04o and a counting time of 15 s per step.

A FESEM (model FESEM LEO 1530) and a FEI Helios NanoLab Workstation were used for cross-sectional observations of mechanically fractured samples and their surfaces chemically etched in 0.5 mol/LNaOH as well as *nw* arrays freed up from the templates.

To prepare TEM samples, deposited species were liberated by dissolving alumina template in 0.1 mol/L sodium hydroxide at 40 o C. The released products were then rinsed many times and finally dispersed in ethanol. At each stage, solvent exchange was carried out by centri‐ fuging, extracting the supernatant and adding fresh solvent. Finally, free-standing nanospe‐ cies were re-dispersed in ethyl alcohol. For TEM observations, specimens were prepared by placing a drop of suspension on a Lacey carbon grid and left overnight at room temperature to evaporate the solvent. Nanostructured products were examined with a TEM microscope (model MORGAGNI 268) operating at 80 kV.

Optical properties of bismuth selenide arrays fabricated inside the alumina template pores were studied by recording the transmittance spectra within the 190 to 3150 nm wavelength range with respect to pure alumina template using a Shimadzu UV-3101PC spectrophotom‐ eter. The transmission data were manipulated for calculating the absorption coefficient de‐ pendency on the photon energy.

How and Why Alumina Matrix Architecture Influence the Shape and Composition of Nanowires Grown... http://dx.doi.org/10.5772/52589 413

**Figure 15.** A typical cross-sectional FESEM view of alumina templates with average *Ø*pore: 10 (A), 13 nm (B), 25 (C) and 50 nm (D) after *AC* treatment in the solution containing 0.02 Bi2(SO4)3, 0.15 TEA, 0.07 EDTA and 0.024 mol/L H2SeO3 and H2SO4 to adjust pH to 5.0at *j*ac=0.3 A dm-2 for 3 min.

#### **5.1. Results**

**No.** Average *Ø* pore, (nm)

412 Nanowires - Recent Advances

10 13 25-28 45-50 100

templates.

nm and 170 o

their size (Noh et al. 2007).

with a step of 0.04o

in 0.1 mol/L sodium hydroxide at 40 o

(model MORGAGNI 268) operating at 80 kV.

pendency on the photon energy.

Bath composition

1.2 H2SO4 1.2 H2SO4 0.5 H2SO4 0.3 H2C2O4 0.04 H2C2O4

study experimentally and approximated to: 250 o

(M) **Anodizing conditions Pore widening conditions**

C for 100 and 50 nm, 200 o

C. The released products were then rinsed many times


C for 28 and 13

5 V *dc*; 5.00 ± 0.01 °C; 20 h 10 V *dc*; 10.00 ± 0.01 °C; 3 h 25 V *dc*; 3.00 ± 0.01 °C; 2 h 40 V *dc*; 17.00 ± 0.02 °C; 1 h 90 V *dc*; 1.00 ± 0.02 °C; 1.2 h

**Table 2.** Summary of anodizing and post treatment variables employed for fabrication and etching of alumina

To increase crystallinity of the final-products, samples were annealed for 3 hours in vacuum. The optimal annealing temperatures (*T*ann) for various alumina templates were found in this

well-known dependency of melting and crystallization temperatures of nanomaterials on

The alumina templates intended for XRD and UV-vis-IR investigations were separated from the electrode surface by one-side sequential etching of the electrode window in a solution of 1.5 mol/LNaOHand then in 10 wt% HCl and 0.1 mol/L CuCl2 followed by thorough rinsing and drying in a nitrogen stream. X-ray diffraction studies were performed with a diffrac‐ tometer D8 (Bruker AXS, Germany) equipped with a Göbel mirror (primary beam mono‐ chromator) for CuKα radiation. A step-scan mode was used in the 2*Θ* range from 18 to 55o

A FESEM (model FESEM LEO 1530) and a FEI Helios NanoLab Workstation were used for cross-sectional observations of mechanically fractured samples and their surfaces chemically

To prepare TEM samples, deposited species were liberated by dissolving alumina template

and finally dispersed in ethanol. At each stage, solvent exchange was carried out by centri‐ fuging, extracting the supernatant and adding fresh solvent. Finally, free-standing nanospe‐ cies were re-dispersed in ethyl alcohol. For TEM observations, specimens were prepared by placing a drop of suspension on a Lacey carbon grid and left overnight at room temperature to evaporate the solvent. Nanostructured products were examined with a TEM microscope

Optical properties of bismuth selenide arrays fabricated inside the alumina template pores were studied by recording the transmittance spectra within the 190 to 3150 nm wavelength range with respect to pure alumina template using a Shimadzu UV-3101PC spectrophotom‐ eter. The transmission data were manipulated for calculating the absorption coefficient de‐

and a counting time of 15 s per step.

etched in 0.5 mol/LNaOH as well as *nw* arrays freed up from the templates.

C for 10 nm *Ø*pore templates. The different values of *T*ann were chosen due to the

Uniform filling of alumina pores by densely packed Bi2Se3*nws* with an average pore diame‐ ter *Ø*pore= 40-50 nm in the solution and conditions adopted in this study were demonstrated by us in (Jagminas et al. 2008). Figure15 presents typical FESEM images of the final products encapsulated within the alumina template pores with average diameter 10, 13, 25 and 50 nm showing that the diameters of *nws* grown inside the alumina pores by *ac* deposition are in agreement with the nominal pore diameter of templates while the height of deposited prod‐ ucts depends on the current density, *j*ac, deposition time, *τ*dep, and *Ø*pore. Under the same dep‐ osition conditions, an increase in the *Ø*pore resulted in filling of pores of a lower height. Furthermore, variations in *Ø*pore do not noticeably altered the completeness of the pore fill‐ ing; for all cases most of the pores, especially at the metal/oxide interface with heights up to 1 μm, seem nicely filled. On the other hand, the dispersity of *nws* lengths increases with *τ*dep, especially at higher *j*ac, and *AC* voltages. Surprisingly, the pitting of alumina films, frequent‐ ly observed in other solutions as a result of the alumina barrier-layer breakdown and crys‐ tallization of salts (Jagminas 2002), was not observed for the solution of this study under a wide range of deposition conditions: *j*ac up to 0.5 A dm-2, *τ*dep up to 45 min, and *Ø*pore up to 100 nm. The influence of the solution temperature on the composition of products was also investigated here. In all cases increase in the solution temperature higher than 40 °C led to non-uniform depositions within the alumina pores of products in the lower quantity most likely due to alumina pore sealing, as could be expected.

Figure 16 shows XRD patterns of porous alumina templates with average *Ø*pore 100, 50, 25, 13 and 10 nm filled with BixSey*Nw* arrays by *AC* electrolysis in the same optimized solution at the same constant *j*ac, ca. 0.3 A/dm2 , for 25 min. From the patterns, it has been found that under the same electrolysis conditions Bi2Se3, Bi3Se2 or both phases of selenides can be de‐ posited. As seen from Fig. 16a, a quite pure, Se-rich phase, Bi2Se3, grows when templates with *Ø*pore≥ 50 nm are used. However, Bi-rich phase, Bi3Se2, appears to form more readily when fine structure templates with *Ø*pore≤ 13 nm are employed (Fig. 16b). Moreover, the *Nws* array grown under the same conditions within extremely tiny pores, with *Ø*pore= 10 nm, was found to be composed of Bi3Se2 and some Bi0 inclusions while in the case of *Ø*pore= 28 nm the *nws* are composed of Bi2Se3 with some amount of Bi3Se2. In the case of tiny pores (*Ø*pore= 13 and 10 nm), increase in the molar ratio of the selenium and bismuth precursors (αSe/Bi) from 1.0 to 2.0 results in the formation of BiSe (αSe/Bi = 1.5) and finely grained Bi3Se2 (αSe/Bi = 2.0) without Bi0 inclusions. Also, with αSe/Bi increase a somewhat slower bath voltage growth has been determined during the deposition process at a constant *AC* current density.

The influence of the solution temperature on the composition of products was also investi‐ gated here. In all cases increase in the solution temperature higher than 40 °C led to nonuniform depositions within the alumina pores of products in the lower quantity most likely due to alumina pore sealing, as could be expected. For the same *Ø*pore, however, no changes in the phase composition of deposited products have been observed within 5 to 35 °C. The compositional variations of deposited *nws* with the size of alumina template pores can be explained as follows:

It is known that Se-rich thin films of bismuth selenide, namely Bi2Se3, may be electrodeposit‐ ed only when enough Se is present in the reaction zone. However, as it has been shown ear‐ lier by us (Jagminas et al. 2005), the discharge of SeO3 -2 ions from aqueous solutions of selenious acid at the bottom of tinny pores under *ac* bias is hampered. As a result, the size of *a*-Se species that can be deposited drastically decrease with *Ø*pore shrinking. Surprisingly, the increase in the *ac* current density and electrolysis time influenced the content of deposited *a*-Se in these pores only negligible. In contrast, the content of selenium deposited in wider pores, ca. 40-50 nm, under the same conditions increased many folds. Thus, the formation of Se-rich bismuth selenide nanowires in the oxalic acid alumina pores (*Ø*pore≥ 40 nm) can sim‐ ply be released. Again, only Bi-rich phases can be deposited in the sulfuric acid alumina templates with *Ø*pore≤ 13 nm.

How and Why Alumina Matrix Architecture Influence the Shape and Composition of Nanowires Grown... http://dx.doi.org/10.5772/52589 415

ing; for all cases most of the pores, especially at the metal/oxide interface with heights up to 1 μm, seem nicely filled. On the other hand, the dispersity of *nws* lengths increases with *τ*dep, especially at higher *j*ac, and *AC* voltages. Surprisingly, the pitting of alumina films, frequent‐ ly observed in other solutions as a result of the alumina barrier-layer breakdown and crys‐ tallization of salts (Jagminas 2002), was not observed for the solution of this study under a wide range of deposition conditions: *j*ac up to 0.5 A dm-2, *τ*dep up to 45 min, and *Ø*pore up to 100 nm. The influence of the solution temperature on the composition of products was also investigated here. In all cases increase in the solution temperature higher than 40 °C led to non-uniform depositions within the alumina pores of products in the lower quantity most

Figure 16 shows XRD patterns of porous alumina templates with average *Ø*pore 100, 50, 25, 13 and 10 nm filled with BixSey*Nw* arrays by *AC* electrolysis in the same optimized solution at

under the same electrolysis conditions Bi2Se3, Bi3Se2 or both phases of selenides can be de‐ posited. As seen from Fig. 16a, a quite pure, Se-rich phase, Bi2Se3, grows when templates with *Ø*pore≥ 50 nm are used. However, Bi-rich phase, Bi3Se2, appears to form more readily when fine structure templates with *Ø*pore≤ 13 nm are employed (Fig. 16b). Moreover, the *Nws* array grown under the same conditions within extremely tiny pores, with *Ø*pore= 10 nm, was found to be composed of Bi3Se2 and some Bi0 inclusions while in the case of *Ø*pore= 28 nm the *nws* are composed of Bi2Se3 with some amount of Bi3Se2. In the case of tiny pores (*Ø*pore= 13 and 10 nm), increase in the molar ratio of the selenium and bismuth precursors (αSe/Bi) from 1.0 to 2.0 results in the formation of BiSe (αSe/Bi = 1.5) and finely grained Bi3Se2 (αSe/Bi = 2.0) without Bi0 inclusions. Also, with αSe/Bi increase a somewhat slower bath voltage growth has

been determined during the deposition process at a constant *AC* current density.

The influence of the solution temperature on the composition of products was also investi‐ gated here. In all cases increase in the solution temperature higher than 40 °C led to nonuniform depositions within the alumina pores of products in the lower quantity most likely due to alumina pore sealing, as could be expected. For the same *Ø*pore, however, no changes in the phase composition of deposited products have been observed within 5 to 35 °C. The compositional variations of deposited *nws* with the size of alumina template pores can be

It is known that Se-rich thin films of bismuth selenide, namely Bi2Se3, may be electrodeposit‐ ed only when enough Se is present in the reaction zone. However, as it has been shown ear‐

selenious acid at the bottom of tinny pores under *ac* bias is hampered. As a result, the size of *a*-Se species that can be deposited drastically decrease with *Ø*pore shrinking. Surprisingly, the increase in the *ac* current density and electrolysis time influenced the content of deposited *a*-Se in these pores only negligible. In contrast, the content of selenium deposited in wider pores, ca. 40-50 nm, under the same conditions increased many folds. Thus, the formation of Se-rich bismuth selenide nanowires in the oxalic acid alumina pores (*Ø*pore≥ 40 nm) can sim‐ ply be released. Again, only Bi-rich phases can be deposited in the sulfuric acid alumina

, for 25 min. From the patterns, it has been found that


likely due to alumina pore sealing, as could be expected.

lier by us (Jagminas et al. 2005), the discharge of SeO3

the same constant *j*ac, ca. 0.3 A/dm2

414 Nanowires - Recent Advances

explained as follows:

templates with *Ø*pore≤ 13 nm.

**Figure 16.** a) XRD patterns for various alumina templates filled by *AC* treatment at *j*ac 0.3 A dm-2 for 30 min in the solution containing 0.02 Bi2(SO4)3, 0.15 TEA, 0.07 EDTA and 0.024 mol/L H2SeO3 and H2SO4 to adjust pH to 5.0 after annealing in vacuum for 3 h. (b) The same as in part (a) within the 26↔32 2Θ range.

In this study, the optical properties of bismuth selenide *Nw* arrays fabricated within the pores of various alumina templates were investigated by recording the transmittance UVvis-NIR spectra using the same pure alumina templates as reference. In this set-up, the thickness of alumina templates varied between 5.0 and 7.5 ± 0.5 μm and both as-grown and annealed in vacuum at 170, 200 and 250 o C templates with various *Ø*pore were studied. To achieve more precise results, the working and reference samples were anodized, post treat‐ ed and annealed together. Again, the optical band-gap, *E*g, for as-grown and annealed ar‐ rays was calculated using a well-known Tauc's relation:

$$\alpha \left( h\nu \right) = A \left( h\nu - E\_{\text{g}} \right)^{\nu/2},\tag{2}$$

where *α* is the absorption coefficient, A is a constant, *hν* is the photon energy, *E*g is the band gap, and *n* depends on the nature of transition, being equal to 1 or 3 for direct-allowed or direct-for‐ bidden and 4 or 6 for indirect-allowed or indirect-forbidden transitions, respectively. The ab‐ sorption coefficient was calculated from the transmittance spectra using a simple relation:

$$\alpha = \quad -\ln T \;/\, h\_{\text{BUS}e} \tag{3}$$

The average height of bismuth selenide *nws*, *h*BiSe, was determined from the FESEM observa‐ tions of cross-sectioned templates after the optical measurements. By this way, the predomi‐ nant direct band-to-band transition across the gap of different wavelengths was verified for all bismuth selenide *nw* arrays fabricated in this study. Typical plots of *a*<sup>2</sup> *versus* the corre‐ sponding values of photon energy, *hν*, for BixSey *nws* deposited inside the alumina pores with different *Ø*pore are given in Fig. 17. As seen, variation of *α* vs. *hν* demonstrates a wide light absorption region from NIR to UV. In case of alumina templates with average *Ø*pore of 100 nm (curve 4) extrapolating the straight line part of the curve *α<sup>2</sup>* vs. *hν* to the energy-axis, the value of *E*g,dir equal to 0.4 eV was obtained for as-formed Bi2Se3*nw* arrays that is close to *E*g= 0.35 eV of bulk bismuth selenide. Note that with decrease in the diameter of Bi2Se3*nws*, the absorption of higher energy light increases. In case of *Ø*pore = 25 nm, the shape of *α*<sup>2</sup> vs. *hν* plot implied two absorption edges, perhaps due to the deposition within such alumina pores of species composed of a Bi2Se3 and Bi3Se2 mixture, as it has been shown above by XRD investigations. For Bi3Se2nanoscaled products encased within the alumina template pores with *Ø*pore 10 and 13 nm, the *α*<sup>2</sup> vs. *hν* plots demonstrate similar shapes (see curves 1 and 2). However, we found that the straight parts both in *α* vs. *hν* and *α<sup>2</sup>* vs. *hν* plots are not clear and thus numerous tangents can be extrapolated to the energy-axis from these plots indicating, for example, that *E*g value for 13 nm *nws* could be between 0.9 and 1.7 eV, while for 10 nm *nws E*<sup>g</sup> approximated to from ~2.2 to 2.7 eV. Consequently, the effective band gaps of these arrays cannot be precisely distinguished from the absorption spectra. Notice that these results are in line with the results presented in the recent publication (Sun et al. 2008) where the same problem raised analyzing the absorption spectra of CdTe quantum wires. Nevertheless, an obvious blue shift of *α*<sup>2</sup> vs. *hν* plot is observed for 10 nm nanowired prod‐ ucts, e.g. when the diameter of alumina template pores approach to the Bohr radius of bis‐ muth selenides (see *Inset* in Fig. 17).

**Figure 17.** Variations of absorption coefficient, α*<sup>2</sup>*, with photon energy, hν, for bismuth selenide *nw* arrays fabricated by *AC* deposition from the solution as in Fig. 15 at *j*ac0.3 A dm-2 for 3 min (*h*BiSe 0.75 – 1.0 μm) within the alumina tem‐ plate pores differing in pore diameter: (1) 10; (2) 13; (3) 25; (4)100 nm.

How and Why Alumina Matrix Architecture Influence the Shape and Composition of Nanowires Grown... http://dx.doi.org/10.5772/52589 417

The average height of bismuth selenide *nws*, *h*BiSe, was determined from the FESEM observa‐ tions of cross-sectioned templates after the optical measurements. By this way, the predomi‐ nant direct band-to-band transition across the gap of different wavelengths was verified for

sponding values of photon energy, *hν*, for BixSey *nws* deposited inside the alumina pores with different *Ø*pore are given in Fig. 17. As seen, variation of *α* vs. *hν* demonstrates a wide light absorption region from NIR to UV. In case of alumina templates with average *Ø*pore of

the value of *E*g,dir equal to 0.4 eV was obtained for as-formed Bi2Se3*nw* arrays that is close to *E*g= 0.35 eV of bulk bismuth selenide. Note that with decrease in the diameter of Bi2Se3*nws*, the absorption of higher energy light increases. In case of *Ø*pore = 25 nm, the shape of *α*<sup>2</sup>

plot implied two absorption edges, perhaps due to the deposition within such alumina pores of species composed of a Bi2Se3 and Bi3Se2 mixture, as it has been shown above by XRD investigations. For Bi3Se2nanoscaled products encased within the alumina template

clear and thus numerous tangents can be extrapolated to the energy-axis from these plots indicating, for example, that *E*g value for 13 nm *nws* could be between 0.9 and 1.7 eV, while for 10 nm *nws E*<sup>g</sup> approximated to from ~2.2 to 2.7 eV. Consequently, the effective band gaps of these arrays cannot be precisely distinguished from the absorption spectra. Notice that these results are in line with the results presented in the recent publication (Sun et al. 2008) where the same problem raised analyzing the absorption spectra of CdTe quantum wires.

ucts, e.g. when the diameter of alumina template pores approach to the Bohr radius of bis‐

**Figure 17.** Variations of absorption coefficient, α*<sup>2</sup>*, with photon energy, hν, for bismuth selenide *nw* arrays fabricated by *AC* deposition from the solution as in Fig. 15 at *j*ac0.3 A dm-2 for 3 min (*h*BiSe 0.75 – 1.0 μm) within the alumina tem‐

*versus* the corre‐

vs. *hν*

vs. *hν* to the energy-axis,

vs. *hν* plots are not

vs. *hν* plots demonstrate similar shapes (see curves 1

vs. *hν* plot is observed for 10 nm nanowired prod‐

all bismuth selenide *nw* arrays fabricated in this study. Typical plots of *a*<sup>2</sup>

100 nm (curve 4) extrapolating the straight line part of the curve *α<sup>2</sup>*

and 2). However, we found that the straight parts both in *α* vs. *hν* and *α<sup>2</sup>*

pores with *Ø*pore 10 and 13 nm, the *α*<sup>2</sup>

416 Nanowires - Recent Advances

Nevertheless, an obvious blue shift of *α*<sup>2</sup>

plate pores differing in pore diameter: (1) 10; (2) 13; (3) 25; (4)100 nm.

muth selenides (see *Inset* in Fig. 17).

**Figure 18.** High resolution FESEM (A,B) and TEM (C) views of the Bi2Se3 product encased inside the alumina template pores by *AC* deposition as in Fig. 15 at *j*ac 0.3 A dm-2 for 3.0 (A) and 30 min (B,C) before (A) and after template dissolu‐ tion (B,C). *Ø*pore = 50nm. In the *Inset* high magnification (250,000×) Bi2Se3*Nw* FESEM fragments acquired at 1 kV using a concentric BackScatter detector are shown.

To understand absorption variables of bismuth selenide arrays fabricated herein, we further studied the morphology of products, deposited within the alumina templates with *Ø*pore 50 nm and 13 nm using modern high-resolution FESEM and TEM techniques. Shown in Figure 18 are the high resolution panoramic (A) and top-side (B) FESEM images of alumina films encased with Bi2Se3 species before (A) and following the template etching with a drop of 0.5 mol/LNaOH, while (C) depicts the TEM image of the same product disengaged from the template through the template dissolution and collection of the remained species by centri‐ fugation and several washings. As seen from images B and C, Bi2Se3 deposited within the 50 nm pores has a granular shape. This granular structure can be also visualized from the high magnification FESEM observation of nanowired product (see *Inset*). Through the TEM ob‐ servation the size of granules varied within the 17 to 40 nm range implying that 50 nm *nws* of Bi2Se3 are composed of weakly connected nanocrystals. A similar morphology was also observed for disengaged species of bismuth selenides deposited within 25-28 nm pores. In the case of tiny pores, however, the structure of the deposited material after dissolution of alumina matrix was found to differ significantly from the 50 nm products in that it consists of short and tinny *nw* fragments in length of from 0.2 to 1.0 μm (see Fig. 19) even after the template etching and liberation from the matrix procedures. It is some confusing knowing that in the case of *Ø*pore= 13 nm the deposited material is not a phase pure material but ac‐ cording to XRD results consists of a Bi3Se2 and Bi2Se3 mixture.

#### **5.2. Discussion**

Bi2Se3 is reported as a direct band gap semiconductor material. Of special note is that the band gap (*E*g) values of the Bi2Se3 crystals determined from the optical measurements by dif‐ ferent scientists disagree strongly both for bulk materials and nanomaterials. For example, for bulk Bi2Se3 Novoselova(1978) reported *E*g = 0.35 eV, while Lide(1991) only *E*g = 0.16 eV. In the case of Bi2Se3 films fabricated by electroless deposition, the presence of two edges corre‐ sponding to *E*g = 0.354 eV and *E*g = 1.03 eV has been reported by Bhattacharya and Prama‐ nik(1980). Moreover, for thin Bi2Se3 films the band gaps as high as 2.3 eV has been reported by Pejova and Grozdanov (2002) linking such high values with a nanocrystalline film na‐ ture. The great *E*g variations have been also reported for nm-scaled Bi2Se3 including 1.59 eV(Jiang et al. 2006) and 2.25 eV(Ota et al. 2006). Noteworthy that the reported variations in the band gap values of bismuth selenide thickfilms and nanomaterials frequently are ascri‐ bed more to the morphology and purity of this semiconductor than to the size quantization effects. Consequently our findings of high *E*g of bismuth selenide species deposited within alumina pores under conditions of this study cannot be considered as surprising.

**Figure 19.** A top-side FESEM view of alumina surface chemically etched by a drop of 0.5 mol/LNaOH for a case when alumina template with average Øpore= 13 nm is encased with bismuth selenide nws by AC deposition as in Fig. 15. In the Inset, 13 nm in diameter a product fragment acquired at 1 kV using a concentric BackScatter detector is shown.

#### **5.3. Concluding remarks**

the case of tiny pores, however, the structure of the deposited material after dissolution of alumina matrix was found to differ significantly from the 50 nm products in that it consists of short and tinny *nw* fragments in length of from 0.2 to 1.0 μm (see Fig. 19) even after the template etching and liberation from the matrix procedures. It is some confusing knowing that in the case of *Ø*pore= 13 nm the deposited material is not a phase pure material but ac‐

Bi2Se3 is reported as a direct band gap semiconductor material. Of special note is that the band gap (*E*g) values of the Bi2Se3 crystals determined from the optical measurements by dif‐ ferent scientists disagree strongly both for bulk materials and nanomaterials. For example, for bulk Bi2Se3 Novoselova(1978) reported *E*g = 0.35 eV, while Lide(1991) only *E*g = 0.16 eV. In the case of Bi2Se3 films fabricated by electroless deposition, the presence of two edges corre‐ sponding to *E*g = 0.354 eV and *E*g = 1.03 eV has been reported by Bhattacharya and Prama‐ nik(1980). Moreover, for thin Bi2Se3 films the band gaps as high as 2.3 eV has been reported by Pejova and Grozdanov (2002) linking such high values with a nanocrystalline film na‐ ture. The great *E*g variations have been also reported for nm-scaled Bi2Se3 including 1.59 eV(Jiang et al. 2006) and 2.25 eV(Ota et al. 2006). Noteworthy that the reported variations in the band gap values of bismuth selenide thickfilms and nanomaterials frequently are ascri‐ bed more to the morphology and purity of this semiconductor than to the size quantization effects. Consequently our findings of high *E*g of bismuth selenide species deposited within

alumina pores under conditions of this study cannot be considered as surprising.

**Figure 19.** A top-side FESEM view of alumina surface chemically etched by a drop of 0.5 mol/LNaOH for a case when alumina template with average Øpore= 13 nm is encased with bismuth selenide nws by AC deposition as in Fig. 15. In the Inset, 13 nm in diameter a product fragment acquired at 1 kV using a concentric BackScatter detector is shown.

cording to XRD results consists of a Bi3Se2 and Bi2Se3 mixture.

**5.2. Discussion**

418 Nanowires - Recent Advances

We initiated this study to show for the first time that the composition and morphology of bismuth selenide deposited within the alumina template pores by means of alternating cur‐ rent deposition depend on the diameter of pores. Under the same electrolysis conditions *nw* arrays from Se-rich (Bi2Se3), Bi-rich (Bi3Se2) or both phases can be successfully fabricated if the templates with average *Øpore*≥ 50 nm, *Øpore*= 28 nm, and *Øpore*≤ 13 nm, respectively, are used. Also, the optical properties of bismuth selenide *nws* differing in size and phase com‐ position, differ significantly. The band-edge absorption at 0.4 eV, characteristic to bulk *Eg* of Bi2Se3,was clearly evidenced only for bismuth selenides deposited within the alumina pores approximated to 100 nm. With decrease in *Øpore* and diameter of bismuth selenide *nws* the blue shift of absorption edge is obvious although in the case of very tiny pores, when *Øpore* approaches the Bohr radius, the determination of effective band gaps for deposited BixSey *nw* arrays was found to be somewhat problematic using only the experimental transmission spectra perhaps due to at least dual composition of *nws*.

## **Author details**

#### Arūnas Jagminas

State Research Institute Center for Physical Sciences and Technology, Lithuania

## **References**


[21] Jagminas, A.; Gailiūtė, I.; Niaura, G. & Giraitis, R. (2005). Template-assisted fabrica‐ tion of pure Se nanocrystals in controllable dimensions*. Chemija*, Vol. 16(3/4), pp. 15

[7] Cui, H.; Liu, H.; Li, X.; Wang, J.; Han, F. & Zhang, X. (2004b). Synthesis of Bi2Se3 ther‐ moelectric nanosheets and nanotubes through hydrothermal co-reduction method. *J.*

[8] Diggle, J.W.; Downie, T.C. & Goulding, C.W. (1969). Anodic oxide films on alumini‐

[9] Doughty, A.S.; Thompson, G.E.; Richardson, J.A. & Wood, G.C. (1975). Investigation of the electrolytic colouring of porous anodic films on aluminium using electron mi‐

[10] Forrer, P.; Schlottig, F.; Siegenthaler, H. & Textor, M. (2000). Electrochemical prepara‐ tion and surface properties of gold nanowire arrays formed by the template techni‐

[11] Furneaux, R.C.; Rigby, W.R. Davidson, A.P. (1989). The formation of controlled-po‐ rosity membranes from anodically oxidized aluminium. *Nature*, Vol. 337, pp. 147-149.

[12] Gerein, N.J. & Haber, J.A. (2005). Effect of ac electrodeposition conditions on the growth of high aspect ratio copper nanowires in porous aluminum oxide templates.

[13] Giani, A.; Bayaz, A.Al.; Foucaran, A.; Pascal-Delannoy, F. & Boyer, A. (2002). Elabo‐ ration of Bi2Se3 by metalorganic chemical vapor deposition. *J. Cryst. Growth*, Vol. 236,

[14] Goad, D.G.W. Moskovits. (1978). Colloidal metal in aluminium oxide. *J. Appl. Phys*.,

[15] Gracia, V.M.; Nair, M.T.S.; Nair, P.K. & Zingaro, R.A. (1997). Chemical deposition of bismuth selenide thin films using N,N-dimethylselenourea. *Semicond. Sci. Technol*.,

[16] Harkness, A.C. & Young, L. (1966). High resistance anodic oxide films on aluminium.

[17] Hermann, E. (1972). Elektrolytisches Färben von Anodisiertem. Aluminium. Galva‐

[18] Hillhouse, H.W. & Tuominen, M.T. (2001). Modelling the thermoelectric transport properties of nanowires embeded in oriented microporous and mesoporous films.

[19] Jagminas, A. (2002). Influence of magnesium and aluminium ions on the copper ac deposition into aluminium anodic oxide film nanotubes. *J. Appl. Electrochem*., Vol. 32,

[20] Jagminas, A.; Lichušina, S.; Kurtinaitienė, M. & Selskis, A. (2003). Concentration ef‐ fect of the solutions for alumina template ac filling by metal arrays. *Appl. Surf. Sci.,*

*Solid. State Chem*., Vol. 177, pp. 4001-4006.

um. *Chem. Rev*., Vol. 69, pp. 365-405.

croscopy. *Trans. IMF*, Vol. 53, pp. 33-39.

que. *J. Appl. Electrochem*., Vol. 30, pp. 533-541.

*J. Phys. Chem. B*, Vol. 109, pp. 17372-17385.

pp. 217-220.

420 Nanowires - Recent Advances

Vol. 49, pp. 2929-2934.

Vol. 12, pp. 645-649.

pp. 1201-1209.

Vol. 211, pp. 194-202.

*Can. J. Chem*., Vol. 44, pp 2409-2413.

notechnik *Galvanotechnik* Vol. 63, pp. 110-121.

*Miroporous Mesoporous Mater*., Vol. 47, pp. 39-50.


[49] Römer, W. Steinem, C. (2004). Impedance Analysis and Single-Channel Recordings on Nano-Black Lipid Membranes Based on Porous Alumina. *Biophysical Journal,* Vol. 86, pp. 955-965.

[35] Mardilovich, P.P; Govyadinov, A.N.; Mukhurov, N.I.; Rzhevskii, A.M. Paterson, R. (1995). New and modified anodic alumina membranes. Part 1. Thermotreatment of

[36] Masuda H. Fukuda, K. (1995). Ordered metal nanohole arrays made by a two-step replication of honeycomb structures of anodic alumina. *Science*, Vol. 268, pp.

[37] Metzger, R. M.; Konovalov, V. V.; Xu, M. S. T.; Zangari, G.; Xu, B.; Benakli M. & Doyle, W. D. (2000). Magnetic nanowires in hexagonally ordered pores of alumina.

[38] Nielsch, K.; Choi, J.; Schwirn, K.; Wehrspohn, R.B. Gösele, U. (2002). Self-ordering re‐ gimes of porous alumina: the 10 porosity rule. *Nano Lett.*, Vol. 2(7), pp. 677-680.

[39] Noh, K.W.; Woo, E. & Shin, K. (2007). Alteration of crystal structure of bismuth con‐

[40] Novoselova, A.B. (Ed.), Physical and Chemical Properties of Semiconductors, Hand‐

[41] O'Sullivan, J.P. Wood, G.C. (1970). Morphology and mechanism of formation of po‐ rous anodic films on aluminium. *Proc. Royal Soc. London, A*., Vol. 317, pp. 511-543.

[42] Ota, J. R.; Roy, P.;. Srivastava, S. K. & Popovitz-Biro, R. A simple hydrothermal meth‐ od for the growth of Bi2Se3 nanorods. *Nanotechnology*, Vol. 17, pp. 1700-1705.

[43] Pan, H.; Sun, H.; Poh, C.; Feng, Y. Lin, J. (2005). Single-crystal growth of metallic nanowires with preferred orientation. *Nanotechnology*, Vol. 16, pp. 1559-1564.

[44] Paulus, P.M.; Luis, F.; Kröll, M.; Schmid, G. & de Jongh, L.J. (2001). Low-temperature study of the magnetization reversal and magnetic anisotropy of Fe, Ni and Co nano‐

[45] Pejova, B. & Grozdanov, I. (2002). Chemical deposition and characterization of glassy

[46] Picht, O.; Müller, S.; Albert, I.; Rouber, M.; Lensch-Falk, J.; Medlin, D. L.; Neumann, R. & Toimil-Molares, M.E. (2012). Tuning the geometrical and crystallographic char‐ acteristics of Bi2Te3 nanowiresby electrodeposition in ion-track membranes. *J. Phys.*

[47] Qin, J.; Nogues, J.; Mikhaylova, M.; Roig, A.; Munoz, J. S. Muhammed, M. (2005). Dif‐ ferences in the magnetic properties of Co, Fe, and Ni 250-300 nm wide nanowires electrodeposited in amorphous anodized alumina templates. *Chem. Mater*., Vol. 17(7),

[48] Routkevich, D.; Bigioni, T.; Moskovits, M. & Xu, J..M. (1996). Electrochemical fabrica‐ tion of CdS nanowire arrays in porous anodic aluminum oxide templates. *J. Phys.*

bismuth(III) selenide thin films. *Thin Solid Films*, Vol. 408, pp. 6-10.

wires. *J. Magn. Magn. Mater*., Vol. 224, pp. 180-196.

*Chem. C*, Vol. 116, pp. 5367-75.

*Chem*., Vol. 100, pp. 14037-14047.

pp. 1829-1834.

fined in cylindrical nanopores. *Chem. Phys. Lett*., Vol. 444, pp. 130-134.

anodic alumina membranes. *J. Membr. Sci*. 98, 131-142.

*IEEE Trans. Magn*., Vol. 36(1), pp. 30-35.

1466-1468.

422 Nanowires - Recent Advances

book, Moscow, 1978.


[63] Xu, T.; Zangari, G. Metzger, R.M. (2002). Periodic holes with 10 nm diameter pro‐ duced by grazing Ar+ milling of the barrier layer in hexagonally ordered nanoporous

[64] Yan, X.B.; Puodel, Y.; Ma, W.S.; Liu, G.; Joshi, H.; Wang, Y.; Lan, G.; Wang, D.; Chen G. & Ren, Z.F. (2010). Experimental studies on anisotropic thermoelectric properties

[65] Yang, X.; Wang, X. & Zhang, Z. (2005). Synthesis and optical properties of singlecrystalline bismuth selenide nanorods via a convenient route. *J. Cryst. Growth*, Vol.

[66] Yoo, W.-Ch. Lee, J.-K. (2004). Field-dependent growth patterns of metals electroplat‐ ed in nanoporous alumina membranes. *Adv. Mater*., Vol. 16, pp. 1097-1101.

[67] Zheng, M.J.; Zhang, L.D.; Li, G. H. & Shen, W.Z. (2002). Fabrication and optical prop‐ erties of large-scale uniform zinc oxide nanowire arrays by one-step electrochemical

and structures of n-type Bi2Te2.7Se0.5. *Nano Lett.*, Vol. 10, pp. 3373-3378.

deposition technique. *Chem. Phys. Lett*., Vol. 363, pp. 123-125.

alumina. *Nano Lett*., Vol. 2, pp. 37-41.

276, pp. 566-570.

424 Nanowires - Recent Advances

## *Edited by Xihong Peng*

One dimensional nanoscale structures such as nanowires have drawn extensive research interests in recent years. The size miniature brings unique properties to nanowires due to quantum confinement. The large surface-to-volume ratio renders nanowires with high sensitivity to surface effects. The unique geometrical advantages and properties facilitate the utilization of nanowires in nano-electronics. InTech scientific publisher has initialized a series of books focusing on fundamental research in nanowires, which largely boosted the widespread of knowledge among the research society. This book is intended to provide an updated review on the applications of various nanowires and the associated advancements in synthesis and properties characterization. The topics include recent progress in metal oxide nanowires, silicon nanowires, carbon based nanotubes and nanowires.

Photo by Rost-9D / iStock

Nanowires - Recent Advances

Nanowires

Recent Advances

*Edited by Xihong Peng*