**Growth and Characteristics of High-quality InN by Plasma-Assisted Molecular Beam Epitaxy Plasma-Assisted Molecular Beam Epitaxy**

**Growth and Characteristics of High-Quality InN by** 

Chen-Chi Yang, Ikai Lo, Cheng-Hung Shih, Chia-Hsuan Hu, Ying‑Chieh Wang, Yu‑Chiao Lin, Cheng-Da Tsai, Hui‑Chun Huang, Mitch M. C. Chou, Cheng‑Chang Yu and Der-Jun Jang Shih, Chia-Hsuan Hu, Ying-Chieh Wang, Yu-Chiao Lin, Cheng-Da Tsai, Hui-Chun Huang, Mitch M. C. Chou, Cheng-Chang Yu and Der-Jun Jang

Additional information is available at the end of the chapter Additional information is available at the end of the chapter

Chen-Chi Yang, Ikai Lo, Cheng-Hung

http://dx.doi.org/10.5772/65812

#### **Abstract**

[72] Lormes W., Hundhausen M., and Ley L. Time resolved photoluminescence of amorphous

[73] Pelant I., and Valenta J. Luminescence Spectroscopy of Semiconductors. Oxford: Univer-

[74] Giorgis F., Mandracci P., Dal Negro L., Mazzoleni M., and Pavesi P. Optical absorption and luminescence properties of wide-band gap amorphous silicon based alloys. Journal

[76] Tessler L. R., and Solomon I. Photoluminescence of tetrahedrally coordinated a-Si1<sup>−</sup>xCx:H.

[77] Estes M. J., and Moddel G. Luminescence from amorphous silicon nanostructures. Phys-

[78] Robertson J. Recombination and photoluminescence mechanism in hydrogenated amor-

[79] Chollon G. Oxidation behaviour of ceramic fibres from the Si–C–N–O system and related

[80] Tayagaki T., Fukatsu S., and Kanemitsu Y. Photoluminescence dynamics and reduced Auger recombination in Si1−xGex/Si superlattices under high-density photoexcitation.

[81] Grieshaber W., Schubert E. F., Goepfert I. D., Karlicek R. F., Schurman M. J., and Tran C. Competition between band gap and yellow luminescence in GaN and its relevance for

sub-systems. Journal of European Ceramics Society 2000;20:1959.

optoelectronic devices. Journal of Applied Physics 1996;80:4615.

[75] Monroe D. Hopping in exponential band tails. Physical Review Letters 1985;54:146.

hydrogenated carbon. Journal of Non-Crystalline Solids 1998;227:570.

sity Press; 2012.

of Non-Crystalline Solids 2000;266:588.

302 Modern Technologies for Creating the Thin-film Systems and Coatings

phous carbon. Physical Review B 1996;53:16302.

Physical Review B 1995;52:10962.

Physical Review B 2009;79:041301.

ical Review B 1996;54:14633.

The high-quality InN epifilms and InN microdisks have been grown with InGaN buffer layers at low temperatures by plasma-assisted molecular beam epitaxy. The samples were analyzed using X-ray diffraction, scanning electron microscopy, high-resolution transmission electron microscopy, and photoluminescence. The characteristics of the InN epifilms and InN microdisks were studied, and the role of InGaN buffer was evaluated.

**Keywords:** InN microdisk, InGaN buffer, Molecular beam epitaxy

### **1. Introduction**

III-Nitride semiconductor compounds have been extensively studied for applications in optoelectronic devices, such as solar cells and light emitting diodes (LEDs) [1–5]. The wide direct band-gap gallium nitride (GaN) and aluminum nitride (AlN) compounds, with energy gaps covering the ultraviolet spectrum, are the dominant materials for solid-state lighting devices and have been well studied to date. The molecular beam epitaxy (MBE) technique can be used to grow a thin epifilm in an ultrahigh vacuum (~10−10 torr) and low temperature condition [6]. Under such conditions, materials in the effusion cells of the MBE system are heated and they move toward the substrate to form epitaxial high purity films. The low-temperature condition is crucial to grow the compounds with a low volatilized temperature (such as In atom, 650°C). Because of the improvement of InN films grown by MBE, the direct band-gap of the indium

© 2016 The Author(s). Licensee InTech. This chapter is 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. © 2017 The Author(s). Licensee InTech. This chapter is 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.

nitride (InN) compound was demonstrated with the value of 0.64 eV rather than 1.9 eV [7, 8]. This important finding indicated that one can tune the band-gap energy to achieve the fullcolor spectrum (red, green, and blue) devices by changing the III-group alloy ratio without any phosphor. Besides, InN is a high potential material in optoelectronic applications due to its outstanding material properties, such as the smallest effective mass, the highest peak and saturation electron drift velocity, and the largest mobility among the nitride semiconductors [9, 10].

On the other hand, the development of the full-color spectrum micron LED is very important for the high-resolution display. The general method to fabricate micron LED is etching process to reach micro scale. However, it is not easy to downsize to 1–10 μm by etching process. In order to fabricate micron LED, a suitable micron growth base is the top priority. In recent years, the growth and characteristics of InN nanowire on Si (1 0 0) by the vapor-liquid-solid mechanism and on Si (1 1 1) by plasma-assisted molecular beam epitaxy (PA-MBE) were reported [11–13]. The wire diameter was less than 100 nm. In our previous work, we have grown the high-quality self-assembled *c*-plane GaN (0 0 0 1¯) hexagonal microdisks with a diameter of 4 μm on γ-LiAlO<sup>2</sup> (LAO) substrates. The diameter of microdisk can be adjusted to optimize the quantum effect for nanodevice applications [5]. Besides, we developed a back process to fabricate an electrical contact for the GaN hexagonal microdisk on a transparent p-type GaN template [14]. Consequently, the InN microdisk provides an opportunity to fabricate the InGaN/ GaN microdisk quantum well for the application of full-color micron LED without the sapphire substrate, which is mostly used for the bulk GaN-based quantum wells in commercial LEDs but has a large lattice mismatch with InN [1, 2]. In this chapter, we will show the growth of InN (0 0 0 1¯) hexagonal thin wurtzite microdisks on the γ-LAO substrate by PA-MBE.

### **2. High-quality InN epifilms**

### **2.1. InGaN buffer layer**

When engineering the band structure of III-nitrides, it is difficult to grow high-quality InN thin film due to the low decomposition temperature of InN (<600°C) and the large lattice mismatch between InN and common substrates (e.g., sapphire or silicon) [15]. Therefore, determining an appropriate substrate for the growth of high-quality InN film is one of the main issues in the fabrication of full-color optoelectronic devices. The lattice mismatch between InN (*a* = 0.3537 and *c* = 0.5704 nm) and sapphire (*a* = 0.4785 and *c* = 1.2991 nm) on the *c*-plane is about 26.1%.

The initial methods prior to InN growth, including substrate nitridation and buffer layer deposition, have very important effects on the growth of high-quality InN films with a flat surface on a sapphire substrate. Xiao et al. grew InN films with 20 min nitridation and a lowtemperature InN (LT-InN) buffer layer. By X-ray diffraction (XRD) and room temperature photoluminescence (PL) analyses, it was found that these InN films grown with LT-InN buffer layer have better quality than those without LT-InN buffer layer [16].

Meanwhile, Saito et al. reported the growth of InN films on sapphire with 1 hour nitridation and low-temperature intermediate InN buffer layers, and they found that the growth of thicker InN with a uniform surface was very difficult without intermediate layers and the electron mobility was improved by improvement of surface flatness [17]. Besides, Lu et al. studied the effect of an AlN buffer layer on the epitaxial growth of InN on the sapphire substrate by MBE and found that by using the AlN buffer layer, the structural and electrical properties of InN could be greatly improved. It was also found that a thicker AlN buffer layer was preferred when growing the InN epilayer, which could lead to better electrical properties and surface morphology [18]. From these studies, we can find that it is helpful to improve the quality of InN thin films by introducing an appropriate buffer layer. In general, a thick GaN film (>4 μm) can be grown on sapphire substrate (0 0 0 1) to form a GaN template. In this chapter, we will show firstly the high-quality epitaxial growth of InN epifilms on GaN template with an appropriate InGaN buffer layer by PA-MBE system. We designed a series of samples to study the effect of InGaN buffer layer with growthtemperature dependence.

### **2.2. Growth of InN epifilms**

nitride (InN) compound was demonstrated with the value of 0.64 eV rather than 1.9 eV [7, 8]. This important finding indicated that one can tune the band-gap energy to achieve the fullcolor spectrum (red, green, and blue) devices by changing the III-group alloy ratio without any phosphor. Besides, InN is a high potential material in optoelectronic applications due to its outstanding material properties, such as the smallest effective mass, the highest peak and saturation electron drift velocity, and the largest mobility among the nitride semiconductors [9, 10]. On the other hand, the development of the full-color spectrum micron LED is very important for the high-resolution display. The general method to fabricate micron LED is etching process to reach micro scale. However, it is not easy to downsize to 1–10 μm by etching process. In order to fabricate micron LED, a suitable micron growth base is the top priority. In recent years, the growth and characteristics of InN nanowire on Si (1 0 0) by the vapor-liquid-solid mechanism and on Si (1 1 1) by plasma-assisted molecular beam epitaxy (PA-MBE) were reported [11–13]. The wire diameter was less than 100 nm. In our previous work, we have grown the high-quality self-assembled *c*-plane GaN (0 0 0 1¯) hexagonal microdisks with a diameter of 4

304 Modern Technologies for Creating the Thin-film Systems and Coatings

(LAO) substrates. The diameter of microdisk can be adjusted to optimize the

quantum effect for nanodevice applications [5]. Besides, we developed a back process to fabricate an electrical contact for the GaN hexagonal microdisk on a transparent p-type GaN template [14]. Consequently, the InN microdisk provides an opportunity to fabricate the InGaN/ GaN microdisk quantum well for the application of full-color micron LED without the sapphire substrate, which is mostly used for the bulk GaN-based quantum wells in commercial LEDs but has a large lattice mismatch with InN [1, 2]. In this chapter, we will show the growth of InN (0 0 0 1¯) hexagonal thin wurtzite microdisks on the γ-LAO substrate by PA-MBE.

When engineering the band structure of III-nitrides, it is difficult to grow high-quality InN thin film due to the low decomposition temperature of InN (<600°C) and the large lattice mismatch between InN and common substrates (e.g., sapphire or silicon) [15]. Therefore, determining an appropriate substrate for the growth of high-quality InN film is one of the main issues in the fabrication of full-color optoelectronic devices. The lattice mismatch between InN (*a* = 0.3537 and *c* = 0.5704 nm) and sapphire (*a* = 0.4785 and *c* = 1.2991 nm) on the *c*-plane

The initial methods prior to InN growth, including substrate nitridation and buffer layer deposition, have very important effects on the growth of high-quality InN films with a flat surface on a sapphire substrate. Xiao et al. grew InN films with 20 min nitridation and a lowtemperature InN (LT-InN) buffer layer. By X-ray diffraction (XRD) and room temperature photoluminescence (PL) analyses, it was found that these InN films grown with LT-InN buffer

Meanwhile, Saito et al. reported the growth of InN films on sapphire with 1 hour nitridation and low-temperature intermediate InN buffer layers, and they found that the growth

layer have better quality than those without LT-InN buffer layer [16].

μm on γ-LiAlO<sup>2</sup>

**2. High-quality InN epifilms**

**2.1. InGaN buffer layer**

is about 26.1%.

Four samples were grown on 2 inch *c*-plane (0 0 0 1) Si-doped GaN/sapphire template substrates that consisted of 3.5 μm intrinsic GaN, 65 nm Si-doped AlGaN and 2 μm Si-doped GaN were grown by metal-organic chemical vapor deposition (MOCVD). The InN thin film was grown on the InGaN buffer layer by the PA-MBE system (Veeco Applied-GEN 930) with standard effusion cells for In- and Ga-evaporation and an rf-plasma cell with 450 W for the N2 -plasma source. Before mounting on a holder, the template substrates were cleaned with acetone (5 min), isopropanol (5 min), and de-ionized water (5 min) in an ultrasonic bath, and then dried with nitrogen gas immediately. After the chemical cleaning, the substrates were out-gassed at 750°C for 10 min in the MBE chamber before epitaxial growth. The temperature was defined by a thermal couple equipped at the backside of the substrates. Thereafter, the substrate temperature was decreased down to growth temperatures. The epitaxial growth of GaN was performed on the GaN template at 700°C with a flux ratio N/Ga = 42.9 represented by beam equivalent pressure (BEP) of evaporative III-group sources from standard effusion cell against that of N<sup>2</sup> source from rf-plasma cell [19] and the duration time of the epitaxial growth for all samples was 10 min. Thereafter, the substrate temperature was ramped to growth temperatures with a flux ratio In/Ga = 2.0, and the duration time of the InGaN buffer layer for all samples was 10 min. Four samples were grown under varied temperatures of InGaN buffer layers: 500, 540, 570, and 600°C. Finally, the substrate temperature was ramped down to growth temperature at 410°C with a flux ratio N/In = 40.0 and the duration time of the InN for all samples was 10 min to grow the InN epifilms.

### **2.3. Analysis of InN epifilms**

The *in situ* reflection high-energy electron diffraction (RHEED) was used to monitor the growth of InN epifilms with 15 kV and 14 mA. The structural properties and crystalline preferred orientations were characterized by an X-ray diffractometer (Bede D1) and a field emission transmission electron microscope (FE-TEM; Phillips Tecnai F-20) with an electron voltage of 200 kV. The cross-sectional TEM specimens were prepared by a focus ion beam (FIB; Seiko SII-3050). The FIB was performed with accelerated voltage of 30 kV to cut the samples roughly and then refined the sample further by accelerated voltage of 5 kV. The surface morphology was evaluated by the field emission scanning electron microscope (FE-SEM; Seiko SII-3050) and the atomic force microscope (AFM; Dimension 3100). AFM images were taken with tapping mode by silicon probe and the scanning data were characterized by software NanoScope (R) III (Digital Instruments, version 5.12r2). The photoluminescence (PL) measurement was carried out by Ti:sapphire laser (Traix-320) with a light source from 808-nm laser and 208 mW power from room temperature (300 K) to 14 K to investigate the optical emission properties of the InN epifilms.

The crystal structure of all samples was characterized by XRD measurements. **Figure 1** shows the XRD results of all samples and indicates that *c*-plane InN epifilms were epitaxially grown on GaN templates. From the peaks of X-ray diffraction pattern at 2*θ* = 32.84°, 33.13°, 33.15°, and 33.76°, we estimated the content of indium of Inx Ga1−xN on the basis of Vegard's law to be about 52, 43, 42, and 23%, respectively [20]. The peaks at 2*θ* = 31.22°, 32.97°, 34.57°, and 34.82° were corresponding to the X-ray diffraction patterns from **c**-plane InN (0 0 0 2), In (1 0 1), GaN (0 0 0 2), and Si-doped AlGaN (0 0 0 2), respectively. These peak positions for the X-ray diffraction patterns were obtained by software Quick Graph (version 2.0) with the Asymmetric Double Sigmoidal linear curve fitting. We can observe that the peak of Inx Ga1−xN shifts from left to right with the increasing growth-temperature of the InGaN buffer layer. This shows that the content of indium decreases with the increasing growth temperature, and the diffraction of Indium was observed corresponds to In droplet on the surface of sample 3. In order to eliminate the influence of In drops in further measurements, the acid treatment (H3 PO4 :HNO3 :CH3 COOH:H<sup>2</sup> O = 50:2:10:9) was employed for sample 3 to remove the In drops on the surface. As compared to other samples, the interference fringes of InN grown on sample 1 exhibit prominent oscillations. Qualitatively, it shows that sample 1 is a very high-quality and layer-by-layer epitaxial growth InN epifilm. **Figure 2(a** and **b)** shows the rocking curve and full-width at half maximum (FWHM) values of the plane of InN (0 0 0 2) and Inx Ga1−xN (0 0 0 2), respectively. The FWHM values of the plane of InN (0 0 0 2) grown on samples 1, 2, 3, and 4 are 435.7, 651.5, 682.6, and 777.1 arc-sec,

**Figure 1.** The X-ray 2 Theta-Omega scans of growing samples.

and then refined the sample further by accelerated voltage of 5 kV. The surface morphology was evaluated by the field emission scanning electron microscope (FE-SEM; Seiko SII-3050) and the atomic force microscope (AFM; Dimension 3100). AFM images were taken with tapping mode by silicon probe and the scanning data were characterized by software NanoScope (R) III (Digital Instruments, version 5.12r2). The photoluminescence (PL) measurement was carried out by Ti:sapphire laser (Traix-320) with a light source from 808-nm laser and 208 mW power from room temperature (300 K) to 14 K to investigate the optical emission properties

The crystal structure of all samples was characterized by XRD measurements. **Figure 1** shows the XRD results of all samples and indicates that *c*-plane InN epifilms were epitaxially grown on GaN templates. From the peaks of X-ray diffraction pattern at 2*θ* = 32.84°,

Vegard's law to be about 52, 43, 42, and 23%, respectively [20]. The peaks at 2*θ* = 31.22°, 32.97°, 34.57°, and 34.82° were corresponding to the X-ray diffraction patterns from **c**-plane InN (0 0 0 2), In (1 0 1), GaN (0 0 0 2), and Si-doped AlGaN (0 0 0 2), respectively. These peak positions for the X-ray diffraction patterns were obtained by software Quick Graph (version 2.0) with the Asymmetric Double Sigmoidal linear curve fitting. We can observe that the

InGaN buffer layer. This shows that the content of indium decreases with the increasing growth temperature, and the diffraction of Indium was observed corresponds to In droplet on the surface of sample 3. In order to eliminate the influence of In drops in further mea-

:CH3

sample 3 to remove the In drops on the surface. As compared to other samples, the interference fringes of InN grown on sample 1 exhibit prominent oscillations. Qualitatively, it shows that sample 1 is a very high-quality and layer-by-layer epitaxial growth InN epifilm. **Figure 2(a** and **b)** shows the rocking curve and full-width at half maximum (FWHM) values

plane of InN (0 0 0 2) grown on samples 1, 2, 3, and 4 are 435.7, 651.5, 682.6, and 777.1 arc-sec,

Ga1−xN shifts from left to right with the increasing growth-temperature of the

COOH:H<sup>2</sup>

Ga1−xN (0 0 0 2), respectively. The FWHM values of the

Ga1−xN on the basis of

O = 50:2:10:9) was employed for

33.13°, 33.15°, and 33.76°, we estimated the content of indium of Inx

306 Modern Technologies for Creating the Thin-film Systems and Coatings

PO4

:HNO3

of the InN epifilms.

peak of Inx

surements, the acid treatment (H3

of the plane of InN (0 0 0 2) and Inx

**Figure 1.** The X-ray 2 Theta-Omega scans of growing samples.

**Figure 2.** (a) Rocking curve of the plane of InN (0 0 0 2). (b) Rocking curve of the plane of Inx Ga1−xN (0 0 0 2). The inset is the FWHM value vs. the growth temperature.

respectively. The FWHM values of the plane of Inx Ga1-xN (0 0 0 2) grown on sample 1, 2, 3, and 4 are 359.6, 541.6, 513.7, and 523.8 arc-sec, respectively. It reveals that the intensity value of the plane of InN (0 0 0 2) grown on sample increases and the FWHM value of the plane of InN (0 0 0 2) grown on sample decreases with the decreasing growth temperature of InGaN grown on sample. Therefore, the maximum intensity and minimum FWHM value of the plane of InN (0 0 0 2) grown on sample 1 from **Figure 2(a)** is obtained, which shows that it is helpful to grow high-crystal quality InN epifilms by decreasing the growth temperature of the InGaN buffer layer and the growth parameter of sample 1 is suitable to grow a highquality InN epifilm.

**Figure 3(a–d)** shows the surface morphology of *c*-plane (0 0 0 2) InN grown on samples 1, 2, 3, and 4, respectively, obtained by SEM (SII-3050). By comparing the RHEED patterns of InN grown on samples 1, 2, 3, and 4 along with [1 1¯ 0 0] InN, we found that the RHEED pattern of sample 1 is streaky but others are spotty patterns, indicating that the growth mode of InN on sample 1 was established by the two-dimensional (2D) Frank-van der Merwe epitaxial growth mode. As compared to the XRD results of InN grown on sample 1 with prominent oscillations, sample 1 is a high-quality and layer-by-layer epitaxial 2D-growth sample. From SEM analysis, we observed the flatness of *c*-plane InN epifilm was getting smoother from sample 4 to sample 1 except for sample 3 because of the In drops left on the surface. The surface morphology of samples 1, 2, 3, and 4 were also analyzed by AFM with the root-mean-square (RMS) roughness, as shown in **Figure 4**. The RMS values of samples 1, 2, 3, and 4 are 0.636, 1.537, 9.821, and 1.910 nm, respectively. It shows that the surface of sample 1 is the flattest surface morphology in the samples, and supports the conclusion of XRD, RHEED, and SEM analyses. It also indicates that it is helpful to grow flat InN epifilms by decreasing the growth

**Figure 3.** SEM images of *c*-plane InN thin films grown on (a) sample 1, (b) sample 2, (c) sample 3 and (d) sample 4. The scale bar is 1 μm.

temperature of InGaN buffer layer and the growth parameter of sample 1 offers a better condition to grow flat InN epifilms. Sample 1 was then studied in more detail.

grown on samples 1, 2, 3, and 4 along with [1 1¯ 0 0] InN, we found that the RHEED pattern of sample 1 is streaky but others are spotty patterns, indicating that the growth mode of InN on sample 1 was established by the two-dimensional (2D) Frank-van der Merwe epitaxial growth mode. As compared to the XRD results of InN grown on sample 1 with prominent oscillations, sample 1 is a high-quality and layer-by-layer epitaxial 2D-growth sample. From SEM analysis, we observed the flatness of *c*-plane InN epifilm was getting smoother from sample 4 to sample 1 except for sample 3 because of the In drops left on the surface. The surface morphology of samples 1, 2, 3, and 4 were also analyzed by AFM with the root-mean-square (RMS) roughness, as shown in **Figure 4**. The RMS values of samples 1, 2, 3, and 4 are 0.636, 1.537, 9.821, and 1.910 nm, respectively. It shows that the surface of sample 1 is the flattest surface morphology in the samples, and supports the conclusion of XRD, RHEED, and SEM analyses. It also indicates that it is helpful to grow flat InN epifilms by decreasing the growth

308 Modern Technologies for Creating the Thin-film Systems and Coatings

**Figure 3.** SEM images of *c*-plane InN thin films grown on (a) sample 1, (b) sample 2, (c) sample 3 and (d) sample 4. The

scale bar is 1 μm.

**Figure 5** shows that PL spectra of sample 1 for different temperatures. The PL measurements were carried out by Ti:sapphire laser (Traix-320) with a light source from 808-nm laser and 208 mW power from 300 to 14 K. When the temperature was changed from 300 to 14 K, the position of major peak shifted from 0.698 to 0.703 eV, in good agreement with the recent data (~0.7 eV) [7, 8]. The intensity of major peak also increased. The major peaks measured at different temperatures were confirmed by a multipeak Gaussian-function curve fitting with the software Origin (Pro. 8.0). The result of the multipeak Gaussian-function curve fitting showed that the major peak was composed by three peaks and all the peak centers shifted to higher energy when the temperature was changed from 300 to 14 K. Among three fitting peaks, only one peak can be described by Varshni's equation [21]:

$$E\_{\rm g}(T) = E\_{\rm g}(0) - \frac{\alpha T^2}{T + \beta}$$

**Figure 4.** AFM images of surface of growing samples by 5 × 5 μm<sup>2</sup> scan: (a) sample 1, (b) sample 2, (c) sample 3, and (d) sample 4. The scale bar is 100 nm

**Figure 5.** The PL spectra taken at different temperatures. The inset is PL peak energy as a function of temperature. The uncertainty of PL peak position is the result of the multipeak Gaussian-function curve fitting. The red curve is the theoretical fitting to Varshni's equation.

In the inset of **Figure 5**, the theoretical fitting to Varshni's equation is obtained with *E*<sup>g</sup> (0) = 0.681 eV, α = 0.18 meV/K and β = 416 K. As compared to *E*<sup>g</sup> (0) = 0.69 eV, α = 0.41 meV/K and β = 454 K from Wu et al. [8], we find that the values of *E*<sup>g</sup> (0) and β are consistent with the results of Wu et al., within the variation of β. The value of β, Debye temperature at 0 K, is in the range from 370 to 650 K for hexagonal InN, estimated by Davydov et al. [22]. However, the different values of α, the Varshni thermal coefficient, might be due to the different InN thicknesses of the sample used in this study. The other two peaks were nearly independent of temperature, and attributed to the defect levels.

The cross-sectional TEM specimen of sample 1 was prepared by a dual-beam focus ion beam (Seiko SII-3050), with the cleavage plane along the [1 1¯ 0 0] direction on the *c*-plane InN (0 0 0 1) and carbon was used as a preservation layer to avoid the damage from the Ga-ion beam during the preparation. The microstructure of sample 1 was analyzed by the field emission transmission electron microscope (FE-TEM; Phillips Tecnai F-20) with an electron voltage of 200 kV. From the TEM bright field image with [1 1 2¯ 0] zone axis in **Figure 6(a)**, we deduced the selective area diffraction (SAD) pattern for sample 1, as shown in **Figure 6(b)**. It shows three distinguishing rectangular diffraction spots in the SAD pattern, indicating that sample 1 was formed by the high-quality GaN, InGaN buffer layer and InN crystals. All of the spots are very clear and with no distortion. It shows that there are few stacking-faults in sample 1. The *d***-spacing** of {0 0 0 1} and {[1 1¯ 0 0]} planes of GaN were measured to be (*d*0 0 0 1 = 0.5109 nm and d1 1¯ 0 0=0.2753 nm). The *d***-spacing** of {0 0 0 1} and {[1 1¯ 0 0]} planes of InGaN were measured to be (*d*0 0 0 1 = 0.5371 nm and d1 1¯ 0 0=0.2821 nm). The *d***-spacing** of {0 0 0 1} and {[1 1¯ 0 0]} planes of InN were measured to be (*d*0 0 0 1 = 0.5634 nm and d1 1¯ 0 0=0.3043 nm). This indicates that the In content of Inx Ga1−xN in Sample 1 is, about 50% determined from these *d*0 0 0 1 data, which is consistent with the XRD analysis. The scanning transmission electron microscope (STEM) measurement will show the high contract images among GaN, InGaN, and InN layers. We show the STEM result for sample 1 in **Figure 6(c)**. It clearly exhibits, with a high-resolution STEM image, that the InN epifilm was well formed on the InGaN buffer layer and the InGaN buffer layer was well established on GaN template. The thicknesses of

In the inset of **Figure 5**, the theoretical fitting to Varshni's equation is obtained with *E*<sup>g</sup>

results of Wu et al., within the variation of β. The value of β, Debye temperature at 0 K, is in the range from 370 to 650 K for hexagonal InN, estimated by Davydov et al. [22]. However, the different values of α, the Varshni thermal coefficient, might be due to the different InN thicknesses of the sample used in this study. The other two peaks were nearly independent of

**Figure 5.** The PL spectra taken at different temperatures. The inset is PL peak energy as a function of temperature. The uncertainty of PL peak position is the result of the multipeak Gaussian-function curve fitting. The red curve is the

The cross-sectional TEM specimen of sample 1 was prepared by a dual-beam focus ion beam (Seiko SII-3050), with the cleavage plane along the [1 1¯ 0 0] direction on the *c*-plane InN (0 0 0 1) and carbon was used as a preservation layer to avoid the damage from the Ga-ion beam during the preparation. The microstructure of sample 1 was analyzed by the field emission transmission electron microscope (FE-TEM; Phillips Tecnai F-20) with an electron voltage of 200 kV. From the TEM bright field image with [1 1 2¯ 0] zone axis in **Figure 6(a)**, we deduced the selective area diffraction (SAD) pattern for sample 1, as shown in **Figure 6(b)**. It shows three distinguishing rectangular diffraction spots in the SAD pattern, indicating that sample 1 was formed by the high-quality GaN, InGaN buffer layer and InN crystals. All of the spots

0.681 eV, α = 0.18 meV/K and β = 416 K. As compared to *E*<sup>g</sup>

310 Modern Technologies for Creating the Thin-film Systems and Coatings

β = 454 K from Wu et al. [8], we find that the values of *E*<sup>g</sup>

temperature, and attributed to the defect levels.

theoretical fitting to Varshni's equation.

(0) =

(0) = 0.69 eV, α = 0.41 meV/K and

(0) and β are consistent with the

**Figure 6.** TEM analysis of sample 1: (a) the bright field image of TEM, the scale bar is 20 nm, (b) the selective area diffraction patterns, the scale bar is 5 (1/nm), (c) the STEM image, the scale bar is 50 nm. (d)–(f) the high-resolution TEM images of InN, InN-InGaN interface, and InGaN-GaN interface, respectively, the scale bar is 5 nm, (g) the enlarged SAD pattern of the square in (b).

InN and InGaN buffer layer were evaluated from the STEM image to be about 50 and 30 nm, respectively. The high-quality crystalline microstructures of InN, InGaN and GaN layers were also confirmed by the high-resolution TEM images. **Figure 6(d–f)** showed that InGaN buffer layer was well-stacked on GaN and high-quality InN epifilm was well-stacked on the InGaN buffer layer with some minor structural defects (e.g., dislocations or stacking faults) occurred in InN and InGaN layers.

### **2.4. Characteristics of InN epifilms**

From the crystal structural analyses by XRD and TEM, we found that the crystal quality was significantly improved by decreasing the growth temperature of InGaN buffer layer. From the SEM images and AFM analyses, we also found that the surface of InN epifilm became smoother by decreasing the growth temperature of the InGaN buffer layer. From the PL measurements, we showed that the energy of 0.681 eV emitted from the InN epifilm of sample 1 (the growth temperature of InGaN buffer layer is 500°C) by the fitting to Varshni's equation. Finally, it is suggestive that one can grow high-quality and flat InN epifilms by decreasing the growth temperature of the InGaN buffer layer. Therefore, the influence of InGaN buffer layer is very effective to grow high-quality InN epifilms and InN microstructures as well. We therefore grow InN hexagonal microdisks on the LAO substrate with the InGaN buffer layer.

### **3. InN hexagonal microdisks**

### **3.1. Growth of InN microdisks**

The two-orientation growth of GaN nanopillars on the LAO substrate has been reported in our previous papers [23, 24]. In this paper, we applied the two-orientation growth to grow the 2D *M*-plane InN epifilm and 3D *c*-plane InN hexagonal microdisks on the LAO substrate with the InGaN buffer layer at low-growth temperature (470°C). The sample was grown on a high-quality 1 × 1 cm<sup>2</sup> LAO (1 0 0) substrate with the InGaN buffer layer by a low-temperature PA-MBE system (Veeco Applied-GEN 930). The LAO substrate was cut from the crystal ingot, which was fabricated by the traditional Czochralski pulling technique. Then, we grew InN hexagonal microdisks with an InGaN buffer layer on the γ-LiAlO<sup>2</sup> substrate by plasmaassisted molecular beam epitaxy. The details of growth parameters can be obtained from the previous paper [25].

### **3.2. Analysis of InN microdisks**

The crystal structure of the microdisk sample is characterized by the high-resolution X-ray diffraction (XRD; Bede D1) measurement and is shown in **Figure 7**. From the peak of X-ray diffraction pattern at 2*θ* = 31.69°, we estimated the content of indium of Inx Ga1−xN on the basis of Vegard's law to be about 20% [20]. The peaks at 2*θ* = 29.07°, 31.31°, 32.29°, and 34.69° represent the X-ray diffraction patterns from **M**-plane InN (1 1¯ 0 0), *c*-plane InN (0 0 0 2¯), *M*-plane GaN (1 1¯ 0 0) and LAO (1 0 0), respectively. By the asymmetric double sigmoidal linear curve fitting with the software Quick Graph (version 2.0), these XRD peak positions were obtained that agreed with those data of the standard wurtzite structure bulk InN (JCPDS file No. 50-1239). The *d***-spacing** between {0 0 0 2¯} planes of InN was evaluated to be *d*0 0 0 2 = 0.28216 nm from the Bragg's law (2*d*sin*θ* = *n*λ) with Cu Kα wavelength λ = 0.1540562 nm. The lattice constant of wurtzite InN microdisk is smaller than that of bulk InN by 1.09%, as compared with the value on JCPDS file, *d*0 0 0 2 = 0.28528 nm.

InN and InGaN buffer layer were evaluated from the STEM image to be about 50 and 30 nm, respectively. The high-quality crystalline microstructures of InN, InGaN and GaN layers were also confirmed by the high-resolution TEM images. **Figure 6(d–f)** showed that InGaN buffer layer was well-stacked on GaN and high-quality InN epifilm was well-stacked on the InGaN buffer layer with some minor structural defects (e.g., dislocations or stacking faults) occurred

From the crystal structural analyses by XRD and TEM, we found that the crystal quality was significantly improved by decreasing the growth temperature of InGaN buffer layer. From the SEM images and AFM analyses, we also found that the surface of InN epifilm became smoother by decreasing the growth temperature of the InGaN buffer layer. From the PL measurements, we showed that the energy of 0.681 eV emitted from the InN epifilm of sample 1 (the growth temperature of InGaN buffer layer is 500°C) by the fitting to Varshni's equation. Finally, it is suggestive that one can grow high-quality and flat InN epifilms by decreasing the growth temperature of the InGaN buffer layer. Therefore, the influence of InGaN buffer layer is very effective to grow high-quality InN epifilms and InN microstructures as well. We therefore grow InN hexagonal microdisks on the LAO substrate with the

The two-orientation growth of GaN nanopillars on the LAO substrate has been reported in our previous papers [23, 24]. In this paper, we applied the two-orientation growth to grow the 2D *M*-plane InN epifilm and 3D *c*-plane InN hexagonal microdisks on the LAO substrate with the InGaN buffer layer at low-growth temperature (470°C). The sample was grown on

ture PA-MBE system (Veeco Applied-GEN 930). The LAO substrate was cut from the crystal ingot, which was fabricated by the traditional Czochralski pulling technique. Then, we grew

assisted molecular beam epitaxy. The details of growth parameters can be obtained from the

The crystal structure of the microdisk sample is characterized by the high-resolution X-ray diffraction (XRD; Bede D1) measurement and is shown in **Figure 7**. From the peak of X-ray

basis of Vegard's law to be about 20% [20]. The peaks at 2*θ* = 29.07°, 31.31°, 32.29°, and 34.69° represent the X-ray diffraction patterns from **M**-plane InN (1 1¯ 0 0), *c*-plane InN (0 0 0 2¯),

InN hexagonal microdisks with an InGaN buffer layer on the γ-LiAlO<sup>2</sup>

diffraction pattern at 2*θ* = 31.69°, we estimated the content of indium of Inx

LAO (1 0 0) substrate with the InGaN buffer layer by a low-tempera-

substrate by plasma-

Ga1−xN on the

in InN and InGaN layers.

InGaN buffer layer.

**2.4. Characteristics of InN epifilms**

312 Modern Technologies for Creating the Thin-film Systems and Coatings

**3. InN hexagonal microdisks**

**3.1. Growth of InN microdisks**

a high-quality 1 × 1 cm<sup>2</sup>

previous paper [25].

**3.2. Analysis of InN microdisks**

The surface morphology of the sample was evaluated by the field emission scanning electron microscope (FE-SEM, SII-3050). **Figure 7(a)** showed the top-view SEM image of the sample. The morphology of the sample exhibited that 3D *c*-plane InN hexagonal microdisks and 2D *M*-plane InN epifilm were grown on the LAO substrate. **Figure 7(b)** showed the tilt-view SEM image of the InN microdisk shown in the center of **Figure 7(a)**, and the diameter of the InN microdisk was 0.60 μm. The micrographic image of the sample showed that the 3D *c*-plane InN hexagonal microdisks and nanopillars were grown atop an anionic hexagonal basal plane of LAO, while the 2D *M*-plane InN epifilm were developed along with the lateral orientation [112¯0]InN//[001]LAO.

The microstructure of the sample was analyzed by a field emission transmission electron microscope (FE-TEM; Phillips Tecnai F-20) at an electron voltage of 200 kV. The cross-sectional TEM specimen was prepared by a dual-beam focus ion beam (FIB; Seiko SII-3050), on the

**Figure 7.** The X-ray 2 Theta-Omega scan of the sample. In the inset of (a) the top-view SEM image of the sample, the scale bar is 1 μm. (b) Enlarged SEM image with a tilted angle of InN hexagonal thin disk, the scale bar is 0.5 μm.

cleavage plane along [11¯00] direction of the *c*-plane InN hexagonal thin disk. **Figure 8(a)** showed the bright field image with [112¯0]InN//[001]LAO zone axis. The thicknesses of *M*plane InN, *M*-plane InGaN and *M*-plane GaN were measured to be about 265, 51, and 137 nm, respectively. The height for the *c*-plane InN hexagonal thin disk from neck to top was about 188 nm. The high-resolution TEM images with [112¯0]InN//[001]LAO zone axis were performed in the areas HR01 and HR02 of the sample, as shown in **Figure 8(a)**. From highresolution TEM analyses, we found the staking faults at the boundary between *M*-plane and *c*-plane GaN, which released the strains between the misfit *M*-plane and *c*-plane wurtzite structures of GaN and InGaN. The *c*-plane wurtzite structure was followed up to the neck area and formed a uniform *c*-plane InGaN pyramid-shaped structure. The wave-shaped InN was produced by the staking faults between the misfit *c*-plane wurtzite structures of InGaN and InN. In **Figure 8(b)**, the wave-shaped InN became uniform in the area HR01 and followed further to form the InN hexagonal thin disk structure. In **Figure 8(c)**, the highquality crystalline structure of the InN thin disk is shown in the area HR02. **Figure 8(d–i)** shows the selective area diffraction (SAD) patterns taken along the growth direction from the bottom to the top (labeled from DP01 to DP06), which covered *c*-plane GaN, *M*-plane GaN, *c*-plane InN, and *M*-plane InN. **Figure 8(d)** simply shows one clear single rectangular diffraction pattern (white) at the location of DP03, indicating that the hexagonal thin disk was uniquely formed by the *c*-plane wurtzite InN crystal. The *d***-spacing** between {0 0 0 1¯} planes and {1 1¯ 0 0} planes of InN hexagonal thin disk were measured to be *dc* = 0.5687 nm and *dM* = 0.3025 nm, respectively. Compared with the values given in JCPDS file No. 50-1239 which are 0.5703 and 0.30647 nm, respectively, the difference between wurtzite InN thin disk and bulk InN for *dc* and *dM* are 0.28 and 1.24%, respectively, revealing that the

**Figure 8.** TEM analyses of the InN hexagonal thin disk: (a) the bright field image with [1 1 2¯ 0]InN//[0 0 1]LAO zone axis. The high-resolution TEM images taken at the points shown in (a) are presented in (b) and (c), the scale bar is 2 nm. The selective area diffraction patterns taken at the points shown in (a) are presented in [(d) – (i)], the scale bar is 2 (1/nm). The ball-stick model for InN epilayer: (j) the chemical bonds of (0 0 0 1¯) surface, (k) the hexagonal thin disk.

lattice constant of InN thin disk is smaller than that of bulk InN. The result is consistent with the XRD analysis. In **Figure 8(e)**, the SAD patterns showed the overlapping diagram of two rectangles and two hexagons at the neck area of the disk (location of DP02), indicating that a *c*-plane InN (white rectangle) was formed in addition to the *c*-plane GaN (red rectangle), *M*-plane InN (blue hexagon), and *M*-plane GaN (yellow hexagon) at the neck area. We checked the *M*-plane InN (blue hexagon) and *M*-plane GaN (yellow hexagon) by the SAD patterns, as shown in **Figure 8(g–i)**. These two hexagons are identical to those shown in **Figure 8(h)**, indicating that the *M*-plane wurtzite InN and *M*-plane wurtzite GaN were grown in the same crystalline direction. From the analyses of SAD patterns, we found that the *c*-plane wurtzite nanocrystal was embedded between *M*-plane wurtzite nanocrystal areas at the beginning of nucleation when GaN was grown on the LAO substrate. We demonstrated a ball-stick model for the self-assembled InN hexagonal thin disk to establish the growth mechanism of the InN hexagonal thin disk. The ball-stick model for the standard wurtzite InN (JCPDS file No. 50-1239) with *a* = *b* = 0.3537 nm, *c* = 0.5703 nm, and u=α¯/c= 3/8 was used to simulate the *c*-plane InN thin disk in **Figure 8(j)**, where blue balls and red balls represented In atoms and N atoms, respectively. In our previous paper, we showed that the GaN (0 0 0 1¯) microdisk with a tilted angle of θ = tan−1(*dM*/*dc*) = 28° was established with the capture of N atoms by the β¯-dangling bonds of the most-outside Ga atoms for each *dc***-spacing** during the GaN lateral overgrowth [5]. In the case of InN thin disk, when the growth temperature was lowered to 470°C, the *c*-plane InN (0 0 0 1¯) hexagonal thin disk was built up with the capture of N atoms by the β¯-dangling bonds of the most-outside In atoms and then the lateral overgrowth occurred; by capture of In atoms by β¯-dangling bonds of N atoms, to form the thin disk. The lateral overgrowth along the (1 1¯ 0 0) direction was extended to six *dM***-spacings** for each *dc***-spacing**, resulting in the angle of 73 off the *c*-axis. Based on the ball-stick model, the edge was then tilted off the *c*-axis [0 0 0 1¯] direction by the angle of *ϕ* = tan−1(6*dM*/*dc*) = 72.76° as shown in **Figure 8(k)**. We also calculated the angle from the measured SAD data at the InN hexagonal thin disk in **Figure 8(d)**, and obtained that the *d*-spacing between {0 0 0 1¯} planes was *dc* = 0.5687 nm and the *d***-spacing** between {1 1¯ 0 0} planes was *dM* = 0.3025 nm, resulting in *ϕ* = tan−1(6*dM*/*dc*) = 72.60°, which was in good agreement with the model predicted.

#### **3.3. Characteristics of InN microdisks**

cleavage plane along [11¯00] direction of the *c*-plane InN hexagonal thin disk. **Figure 8(a)** showed the bright field image with [112¯0]InN//[001]LAO zone axis. The thicknesses of *M*plane InN, *M*-plane InGaN and *M*-plane GaN were measured to be about 265, 51, and 137 nm, respectively. The height for the *c*-plane InN hexagonal thin disk from neck to top was about 188 nm. The high-resolution TEM images with [112¯0]InN//[001]LAO zone axis were performed in the areas HR01 and HR02 of the sample, as shown in **Figure 8(a)**. From highresolution TEM analyses, we found the staking faults at the boundary between *M*-plane and *c*-plane GaN, which released the strains between the misfit *M*-plane and *c*-plane wurtzite structures of GaN and InGaN. The *c*-plane wurtzite structure was followed up to the neck area and formed a uniform *c*-plane InGaN pyramid-shaped structure. The wave-shaped InN was produced by the staking faults between the misfit *c*-plane wurtzite structures of InGaN and InN. In **Figure 8(b)**, the wave-shaped InN became uniform in the area HR01 and followed further to form the InN hexagonal thin disk structure. In **Figure 8(c)**, the highquality crystalline structure of the InN thin disk is shown in the area HR02. **Figure 8(d–i)** shows the selective area diffraction (SAD) patterns taken along the growth direction from the bottom to the top (labeled from DP01 to DP06), which covered *c*-plane GaN, *M*-plane GaN, *c*-plane InN, and *M*-plane InN. **Figure 8(d)** simply shows one clear single rectangular diffraction pattern (white) at the location of DP03, indicating that the hexagonal thin disk was uniquely formed by the *c*-plane wurtzite InN crystal. The *d***-spacing** between {0 0 0 1¯} planes and {1 1¯ 0 0} planes of InN hexagonal thin disk were measured to be *dc* = 0.5687 nm and *dM* = 0.3025 nm, respectively. Compared with the values given in JCPDS file No. 50-1239 which are 0.5703 and 0.30647 nm, respectively, the difference between wurtzite InN thin disk and bulk InN for *dc* and *dM* are 0.28 and 1.24%, respectively, revealing that the

314 Modern Technologies for Creating the Thin-film Systems and Coatings

**Figure 8.** TEM analyses of the InN hexagonal thin disk: (a) the bright field image with [1 1 2¯ 0]InN//[0 0 1]LAO zone axis. The high-resolution TEM images taken at the points shown in (a) are presented in (b) and (c), the scale bar is 2 nm. The selective area diffraction patterns taken at the points shown in (a) are presented in [(d) – (i)], the scale bar is 2 (1/nm). The

ball-stick model for InN epilayer: (j) the chemical bonds of (0 0 0 1¯) surface, (k) the hexagonal thin disk.

We have grown InN hexagonal thin microdisks on the LAO substrate with the InGaN buffer layer by PA-MBE. From the SEM images and TEM analyses, we found that *c*-plane wurtzite was established at the nucleation of GaN on the LAO substrate and *c*-plane InN hexagonal thin disks were built up at low temperature (470°C) after insetting the InGaN buffer layer. The *c*-plane InN (0 0 0 1¯) hexagonal thin disk was produced with the capture of N atoms by the β¯-dangling bonds of the most-outside In atoms, and then laterally over-grown along [1 1¯ 0 0] direction by six *dM***-spacings** for each *dc***-spacing**. The oblique angle of InN hexagonal thin disk was formed by the lateral overgrowth of the wurtzite structure. Based on the standard wurtzite InN, the angle of *ϕ* = tan−1(6*dM*/*dc*) = 72.76° was evaluated. The oblique angle of InN hexagonal thin disk can be examined directly from the SAD pattern and high-resolution TEM analyses to be 72.60° and about 73°, respectively.

### **4. Conclusion**

In this paper, we have reported the growth and characteristics of 2D *c*-plane InN (0 0 0 1) epifilms and 3D *c*-plane InN (0 0 0 1¯) hexagonal thin microdisks with InGaN buffer layers at low temperatures by PA-MBE. By decreasing the growth temperature of the InGaN buffer layer, we can grow high-quality and flat InN epifilms. Besides, InGaN buffer layer can also provide the growth base to form InN hexagonal thin microdisks. By introducing InGaN buffer layers, the high-quality InN epifilms and microstructures can be grown under suitable growth conditions. Consequently, the InN hexagonal thin microdisk provides an opportunity to fabricate the InGaN/GaN microdisk quantum well for the application of full-color micron LED.

### **Acknowledgements**

The project was supported by the Ministry of Science and Technology of Taiwan and the Core Facilities Laboratory for Nanoscience and Nanotechnology in Kaohsiung and Pintung Area.

### **Author details**

Chen-Chi Yang1,<sup>2</sup> , Ikai Lo1,<sup>2</sup> \*, Cheng-Hung Shih1,<sup>2</sup> , Chia-Hsuan Hu1,<sup>2</sup> , Ying-Chieh Wang1,<sup>2</sup> , Yu-Chiao Lin1,<sup>2</sup> , Cheng-Da Tsai1,<sup>2</sup> , Hui-Chun Huang1,<sup>2</sup> , Mitch M. C. Chou1,<sup>2</sup> , Cheng-Chang Yu1,<sup>2</sup> and Der-Jun Jang1,<sup>2</sup>

\*Address all correspondence to: ikailo@mail.phys.nsysu.edu.tw

1 Department of Physics, Center for Nanoscience and Nanotechnology, National Sun Yat-Sen University, Kaohsiung, Taiwan

2 Department of Materials and Optoelectronics Science, National Sun Yat-Sen University, Kaohsiung, Taiwan

### **References**


Alx Ga1−x N/GaN heterostructures. Physics Review B. 2002;**65**:161306. DOI: 10.1103/ PhysRevB.65.161306

[4] I. Lo, W. T. Wang, M. H. Gau, J. K. Tsai, S. F. Tsay, and J. C. Chiang. Gate-controlled spin splitting in GaN/AlN quantum wells. Applied Physics Letters. 2006;**88**:082108. DOI: 10.1063/1.2178505

**4. Conclusion**

micron LED.

Area.

Yu1,<sup>2</sup>

**Acknowledgements**

**Author details**

Chen-Chi Yang1,<sup>2</sup>

Kaohsiung, Taiwan

**References**

and Der-Jun Jang1,<sup>2</sup>

University, Kaohsiung, Taiwan

Yu-Chiao Lin1,<sup>2</sup>

, Ikai Lo1,<sup>2</sup>

, Cheng-Da Tsai1,<sup>2</sup>

316 Modern Technologies for Creating the Thin-film Systems and Coatings

\*Address all correspondence to: ikailo@mail.phys.nsysu.edu.tw

In this paper, we have reported the growth and characteristics of 2D *c*-plane InN (0 0 0 1) epifilms and 3D *c*-plane InN (0 0 0 1¯) hexagonal thin microdisks with InGaN buffer layers at low temperatures by PA-MBE. By decreasing the growth temperature of the InGaN buffer layer, we can grow high-quality and flat InN epifilms. Besides, InGaN buffer layer can also provide the growth base to form InN hexagonal thin microdisks. By introducing InGaN buffer layers, the high-quality InN epifilms and microstructures can be grown under suitable growth conditions. Consequently, the InN hexagonal thin microdisk provides an opportunity to fabricate the InGaN/GaN microdisk quantum well for the application of full-color

The project was supported by the Ministry of Science and Technology of Taiwan and the Core Facilities Laboratory for Nanoscience and Nanotechnology in Kaohsiung and Pintung

, Chia-Hsuan Hu1,<sup>2</sup>

, Mitch M. C. Chou1,<sup>2</sup>

, Ying-Chieh Wang1,<sup>2</sup>

, Cheng-Chang

,

\*, Cheng-Hung Shih1,<sup>2</sup>

, Hui-Chun Huang1,<sup>2</sup>

1 Department of Physics, Center for Nanoscience and Nanotechnology, National Sun Yat-Sen

2 Department of Materials and Optoelectronics Science, National Sun Yat-Sen University,

[1] S. Nakamura, M. Senoh, N. Iwasa, S. I. Nagahama, T. Yamada, and T. Mukai. Super-bright green InGaN single-quantum-well-structure light-emitting diodes. Japanese Journal of Applied Physics. 1995;**34**(Part 2, Number 10B):L1332-L1335. DOI: 10.1143/JJAP.34.L1332

[2] S. Nakamura, S. Pearton, and G. Fasol, editors. The Blue Laser Diode: The Complete Story. Berlin: Springer Science & Business Media; 20 0 0 . 367 p. DOI: 10.1007/978-3-662-04156-7

[3] I. Lo, J. K. Tsai, W. J. Yao, P. C. Ho, L. W. Tu, T. C. Chang, S. Elhamri, W. C. Mitchel, K. Y. Hsieh, J. H. Huang, H. L. Huang, and W. C. Tsai. Spin splitting in modulation-doped


### **Chemical Solution Deposition Technique of Thin-Film Ceramic Electrolytes for Solid Oxide Fuel Cells Chemical Solution Deposition Technique of Thin-Film Ceramic Electrolytes for Solid Oxide Fuel Cells**

Mridula Biswas and Pei-Chen Su Mridula Biswas and Pei-Chen Su

Additional information is available at the end of the chapter Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/66125

#### **Abstract**

[17] Y. Saito, T. Yamaguchi, H. Kanazawa, K. Kano, T. Araki, Y. Nanishi, N. Teraguchi, and A. Suzuki. Growth of high-quality InN using low-temperature intermediate layers by RF-MBE. Journal of Crystal Growth. 2002;237-239:1017-1021. DOI: 10.1016/

[18] H. Lu, W. J. Schaff, J. Hwang, H. Wu, G. Koley, and L. F. Eastman. Effect of an AlN buffer layer on the epitaxial growth of InN by molecular-beam epitaxy. Applied Physics

[19] J-K. Tsai, I. Lo, K-L. Chuang, L-W.Tu, J-H. Huang, C-H. Hsieh, and K-Y. Hsieh. Effect of N to Ga flux ratio on the GaN surface morphologies grown at high temperature by plasma-assisted molecular-beam epitaxy. Journal of Applied Physics. 2004;**95**:460. DOI:

[20] A. R. Denton and N. W. Ashcroft. Vegard's law. Physical Review A. 1991;**43**:3161-3164.

[21] Y.P. Varshni. Temperature dependence of the energy gap in semiconductors. Physica.

[22] V. Yu. Davydov, V. V. Emtsev, I. N. Goncharuk, A. N. Smirnov, V. D. Petrikov, V. V. Mamutin, V. A. Vekshin, S. V. Ivanov, M. B. Smirnov and T. Inushima. Experimental and theoretical studies of phonons in hexagonal InN. Applied Physics Letters. 1999;**75**:3297-

[23] C. H. Hsieh, I. Lo, M. H. Gau, Y. L. Chen, M. C. Chou, W. Y. Pang, Y. I. Chang, Y. C. Hsu, M. W. Sham, J. C. Chiang, and J. K. Tsai. Self-assembled c-plane GaN nanopillars on

[24] I. Lo, C. H. Hsieh, Y. L. Chen, W. Y. Pang, Y. C. Hsu, J. C. Chiang, M. C. Chou, J. K. Tsai,

[25] C. C. Yang, I. Lo, C. H. Hu, H. C. Huang, and M.M. C. Chou. Growth of InN hexagonal

molecular beam epitaxy. Applied Physics Letters. 2008;**92**:202106. DOI: 10.1063/1.2924288

substrate grown by plasma-assisted molecular-beam epitaxy. Japanese Journal

by plasma-assisted

S0022-0248(01)02119-4

10.1063/1.1634388

DOI: 10.1103/PhysRevA.43.3161

3299. DOI: 10.1063/1.125330

γ-LiAlO<sup>2</sup>

Letters. 2001;79:1489-1491. DOI: 10.1063/1.1402649

318 Modern Technologies for Creating the Thin-film Systems and Coatings

1967;34:149-154. DOI: 10.1016/0031-8914(67)90062-6

of Applied Physics. 2008;**47**:891-895. DOI: 10.1143/JJAP.47.891

and D. M. Schaadt. Line defects of M-plane GaN grown on g-LiAlO<sup>2</sup>

microdisks. AIP Advances. 2016;**6**:085015. DOI: 10.1063/1.4961699

Chemical solution deposition (CSD) technique is recently gaining momentum for the fabrication of electrolyte materials for solid oxide fuel cells (SOFCs) due to its costeffectiveness, high yield, and simplicity of the process requirements. The advanced vacuum deposition techniques such as sputtering, atomic layer deposition (ALD), pulsed laser deposition (PLD), metallo-organic chemical vapor deposition (MOCVD) are lacking in scalability and cost-effectiveness. CSD technique includes a variety of approaches such as sol-gel process, chelate process, and metallo-organic decomposition. The present chapter discusses briefly about the evolution of CSD method and its subsequent entry to the field of SOFCs, various solution methods associated with different chemical compositions, film deposition techniques, chemical reactions, heat treatment strategies, nucleation and growth kinetics, associated defects, etc. Examples are cited to bring out the history dating back to the discovery of amorphous zirconia film through the successful fabrication of the crystalline fluorite-type films such as yttria-stabilized zirconia (YSZ), scandia-doped ceria (SDC), and crystalline perovskitetype films such as yttria-doped barium zirconate (BZY) and yttria-doped barium cerate (BCY), to name a few.

**Keywords:** chemical solution deposition, solid oxide fuel cell, ceramic electrolyte, thin films

### **1. Introduction**

The high-temperature solid oxide fuel cells (HTSOFCs, ≥750°C) are yet to find widespread commercialization due to its high cost and short lifetime associated with its high-temperature operation. Thus, the demand for low-cost solid oxide fuel cells (SOFCs) has stimulated

© 2017 The Author(s). Licensee InTech. This chapter is 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. © 2017 The Author(s). Licensee InTech. This chapter is 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.

research to develop low-temperature SOFCs (LTSOFCs, ≤500°C) [1]. Since the high performance of SOFCs requires high operating temperature to activate electrochemical reactions and charge transport processes, reduction in operating temperature will sacrifice performance of SOFCs. Therefore, attempts have been made to find new materials and fabrication technologies for LTSOFCs so that performance remains the same or gets enhanced. One of the main challenges in decreasing the operating temperature is the lower electrolyte conductivity, which resulted in high ohmic resistance, and deteriorates the fuel cell performance. Efforts have been made to enhance performance via reducing the thickness of ceramic electrolyte [2–8]. As the resistance of ionic charge transportation across the electrolyte obeys Ohm's law, thinner film offers less resistance to the ionic conduction and provides lower area-specific resistance. Various thin-film fabrication methods have been developed to date including vacuum-based [9–15] and non-vacuum-based methods [16–18]. Among them, chemical solution deposition (CSD) technique has been a promising technique for its high yield, versatility, and low investment cost. Moreover, the characteristic of CSD method allows easy deposition of the film over the large area up to several square meters [19]. The following paragraphs will be discussing the progress of the concept of thin film and discovery and progress of CSD method.

#### **1.1. The concept of thin-film electrolyte and its progress**

The concept of the thin-film electrolyte was first introduced with the fabrication of 400 μm thick stabilized zirconia (SZ) electrolyte in the 1960s [20]. The trend continued with the development of 30 μm thick yttria-stabilized zirconia (YSZ) electrolyte which, for the first time, successfully demonstrated the remarkable reduction of ohmic resistance from 1 to 3 × 10−3 Ω owing to minimization of the thickness of electrolyte from 1 mm to 30 μm [16]. In 1977, one electrochemical experiment, for the first time, achieved 0.91 V of open-circuit voltage (OCV) at 400°C with 0.05–1.7 μm thick calcia-stabilized zirconia (CSZ) film [10], which is a major breakthrough for LTSOFCs. With the progress in the R&D sector, Westinghouse Electric Corporation first launched cathode-supported tubular cell with 50 μm thick electrolyte, which was the first appearance of SOFC with a film electrolyte in the commercial sector [14]. All these developments were carried out with the vacuum-based methods (i.e., electrochemical vapor deposition, physical vapor deposition, chemical vapor deposition, sputtering, etc.) which lack in scalability and cost-effectiveness. As the alternative to those methods, slurry and solutionbased methods were adopted. They are atmospheric or vacuum plasma spraying [21], spray pyrolysis [22], slurry coating [23, 24], and CSD [17, 18]. Although these methods are successful for micrometer thick film, the thickness of the electrolyte could not be brought down to the sub-micrometer level. With the miniaturization of SOFCs, demand for sub-micrometer thin electrolyte has been generated. CSD-based method has recently demonstrated its ability to produce films with sub-micrometer thickness.

#### **1.2. Discovery and progress of CSD**

The CSD method was introduced with the discovery of silica (SiO2) gel from silicon alkoxide in humidified atmosphere in the middle of the nineteenth century [25]. The potential of this technique was realized with the application of single and multilayered coating of titania (TiO2), zirconia (ZrO2), alumina (Al2O3), etc. on SiO2 glass in the 1950s, and henceforth commercialization followed [26–28]. The development of YSZ film fabrication method started with the invention of amorphous coating of ZrO2 film in the Central Glass and Ceramic Research Institute, India, in 1984 [17]; the Lawrence Berkeley National Laboratory took a pioneering role to develop crack-free, smooth crystalline 0.1–2 μm thin YSZ electrolyte film with sufficiently high conductivity at a temperature of 450°C [29]. Gastightness of YSZ film was demonstrated for the first time with the achievement of promising open-circuit voltage (OCV) of 0.85 V at 600°C [30]. This optimistic result casts light on the rapid progress of CSD method in the field of SOFCs. In 1912, Korea Institute of Science and Technology, South Korea, reported the fabrication of a dense and gastight bilayer electrolyte of YSZ/gadolinia-doped ceria (GDC) with thickness of 100/400 nm. The success of this process was established with the achievement of 1.0 V OCV at 650°C [31]. Numerous significant progresses across the globe are discussed in various articles [3, 32–46].

The fabrication strategies were extended from binary to ternary oxide electrolytes. The fabrication of multiple oxide electrolytes faced difficulties with single-phase formation because of the presence of several oxides and their different crystallization kinetics. The possibility of developing a single-phase ternary oxide thin film was advanced with the successful fabrication of lead zirconate titanate (PZT) film via metallo-organic decomposition (MOD) and sol-gel route in the 1980s [18, 47, 48]. For the first time, phase-pure ternary oxide electrolyte film of Yb-doped strontium zirconate (YDSZ) was successfully obtained via sol-gel method [49]. Our group from Nanyang Technological University, Singapore, has recently reported the fabrication of a dense and crack-free yttria-doped barium zirconate (BZY) thin film by modified CSD technique along with various sintering strategies [38, 40, 50–52] at remarkably low sintering temperature of 800–1000°C. The following sections will give the detail about the various solution preparation strategies and sintering methods.

### **2. CSD methods**

research to develop low-temperature SOFCs (LTSOFCs, ≤500°C) [1]. Since the high performance of SOFCs requires high operating temperature to activate electrochemical reactions and charge transport processes, reduction in operating temperature will sacrifice performance of SOFCs. Therefore, attempts have been made to find new materials and fabrication technologies for LTSOFCs so that performance remains the same or gets enhanced. One of the main challenges in decreasing the operating temperature is the lower electrolyte conductivity, which resulted in high ohmic resistance, and deteriorates the fuel cell performance. Efforts have been made to enhance performance via reducing the thickness of ceramic electrolyte [2–8]. As the resistance of ionic charge transportation across the electrolyte obeys Ohm's law, thinner film offers less resistance to the ionic conduction and provides lower area-specific resistance. Various thin-film fabrication methods have been developed to date including vacuum-based [9–15] and non-vacuum-based methods [16–18]. Among them, chemical solution deposition (CSD) technique has been a promising technique for its high yield, versatility, and low investment cost. Moreover, the characteristic of CSD method allows easy deposition of the film over the large area up to several square meters [19]. The following paragraphs will be discussing the progress of the concept of thin film and discovery and progress of CSD method.

The concept of the thin-film electrolyte was first introduced with the fabrication of 400 μm thick stabilized zirconia (SZ) electrolyte in the 1960s [20]. The trend continued with the development of 30 μm thick yttria-stabilized zirconia (YSZ) electrolyte which, for the first time, successfully demonstrated the remarkable reduction of ohmic resistance from 1 to 3 × 10−3 Ω owing to minimization of the thickness of electrolyte from 1 mm to 30 μm [16]. In 1977, one electrochemical experiment, for the first time, achieved 0.91 V of open-circuit voltage (OCV) at 400°C with 0.05–1.7 μm thick calcia-stabilized zirconia (CSZ) film [10], which is a major breakthrough for LTSOFCs. With the progress in the R&D sector, Westinghouse Electric Corporation first launched cathode-supported tubular cell with 50 μm thick electrolyte, which was the first appearance of SOFC with a film electrolyte in the commercial sector [14]. All these developments were carried out with the vacuum-based methods (i.e., electrochemical vapor deposition, physical vapor deposition, chemical vapor deposition, sputtering, etc.) which lack in scalability and cost-effectiveness. As the alternative to those methods, slurry and solutionbased methods were adopted. They are atmospheric or vacuum plasma spraying [21], spray pyrolysis [22], slurry coating [23, 24], and CSD [17, 18]. Although these methods are successful for micrometer thick film, the thickness of the electrolyte could not be brought down to the sub-micrometer level. With the miniaturization of SOFCs, demand for sub-micrometer thin electrolyte has been generated. CSD-based method has recently demonstrated its ability to

The CSD method was introduced with the discovery of silica (SiO2) gel from silicon alkoxide in humidified atmosphere in the middle of the nineteenth century [25]. The potential of this technique was realized with the application of single and multilayered coating of titania

**1.1. The concept of thin-film electrolyte and its progress**

320 Modern Technologies for Creating the Thin-film Systems and Coatings

produce films with sub-micrometer thickness.

**1.2. Discovery and progress of CSD**

CSD techniques are specially characterized by its mass transport process (**Figure 1**) which maintains liquid phase as the mass transport media for the transportation of precursors from the source to the substrate. The major advantages of CSD method are homogeneity of the product and lower processing temperature compared to the temperature for solid-state sintering [29]. Morphological control over the deposited film can be gained through varying composition, viscosity, pH, concentration of the solution, etc.

**Figure 1.** Chemical solution deposition technique.

Based on various requirements of different film morphology, CSD process can be categorized into three major groups: sol-gel process, chelate process, and MOD technique [53]. In this section, the CSD methodologies will be reviewed, with an emphasis on the underlying chemical aspects of the solution. The characteristics of the main three methods have been summarized in **Table 1**.


**Table 1.** The comparison among three major CSD processes.

#### **2.1. Sol-gel process**

A classical sol-gel process typically involves metal alkoxides and alcohols (M(OR)x and ROH). Common alcohols are methanol and ethanol; 2-methoxyethanol and 1,3-propanediol are also widely used [54, 55]. The selection of cationic components and solvent is crucial for controlling subsequent hydrolysis and condensation reaction, which is the basis for the development of short polymeric species and metal-oxygen-metal (M–O–M) bond upon heat treatment. The reactions are described below [56]:

Hydrolysis:

$$M\text{(OR)}\_{x} + H\_{2}O \rightarrow M\text{(OR)}\_{x-1}\text{(OH)} + ROH \tag{1}$$

Condensation (alcohol elimination):

$$2M\text{(OR)}\_{x\cdot 1}\text{(OH)} + M\_2\text{O}\text{(OR)}\_{2x\cdot 3}\text{(OH)} + \text{ROH}\tag{2}$$

Condensation (water elimination):

Based on various requirements of different film morphology, CSD process can be categorized into three major groups: sol-gel process, chelate process, and MOD technique [53]. In this section, the CSD methodologies will be reviewed, with an emphasis on the underlying chemical aspects of the solution. The characteristics of the main three methods have been

**chemistry**

High Low

Moderate Moderate

Low High

**Simplicity**

**Method Precursors and solvents Control of**

**•** Acetylacetonate and acetic acid as chelating agent and

**•** Long-chain metal carboxylates such as 2-ethylhexanoate, dimethoxy dineodecanoate, and neodecanoate as

A classical sol-gel process typically involves metal alkoxides and alcohols (M(OR)x and ROH). Common alcohols are methanol and ethanol; 2-methoxyethanol and 1,3-propanediol are also widely used [54, 55]. The selection of cationic components and solvent is crucial for controlling subsequent hydrolysis and condensation reaction, which is the basis for the development of short polymeric species and metal-oxygen-metal (M–O–M) bond upon heat treatment. The

( ) ( ) ( ) *M OR H O M OR OH ROH x x* <sup>2</sup> -<sup>1</sup> +® + (1)

( ) ( ) *<sup>2</sup>* ( ) ( ) *x-1 2x - 3 2M OR OH + M O OR OH + ROH* (2)

summarized in **Table 1**.

Metallo-organic decomposition (MOD)

**2.1. Sol-gel process**

Hydrolysis:

Sol-gel **•** Metal alkoxides as precursors

322 Modern Technologies for Creating the Thin-film Systems and Coatings

**•** Alcohols as solvents

Chelate **•** Metal carboxylate, alkoxide, and β-diketonate as precursors

solvents

precursors

**Table 1.** The comparison among three major CSD processes.

reactions are described below [56]:

Condensation (alcohol elimination):

**•** Xylene as solvent (inert)

**•** Acid or base as catalyst **•** Water for polymerization

$$2M\text{(OR}\,\text{(OR}\,\text{)}\_{\text{x-1}}\text{(OH}\,\text{)} + M\_2\text{O}\,\text{(OR}\,\text{)}\_{\text{2x-2}} + H\_2\text{O}\,\text{(}\tag{3}$$

The primarily evolution of inorganic networks occurs through the generation of small particle resulting into colloidal suspension (sol), followed by the formation of continuous network in liquid matrix (gel) [57, 58]. The formation of sol and gel is aided by addition of water, base, and acid [59]. Base and acid act as catalysts. Upon drying and heating, the gel gets converted into amorphous film along with densification. With further heating, amorphous film ceramizes and rate of densification slows down.

In case of multicomponent system, all the alkoxide precursors may not have equal tendency to get dissolved in the same solvent because of their different polarity and ionicity/covalency. Those alkoxides are sometimes pre-hydrolyzed so that the final solution becomes compositionally homogeneous.

Hydrolysis of a particular M–O–R bond depends on its polarity. As the bond polarity increases, the tendency of being hydrolysis also increases. This character plays an important role in determining the processing window which gives an amount of the ratio of reagent to water and precursor concentration. Addition of water to the solution should be controlled to hinder precipitation. Otherwise, powder formation will take place. There are two major strategies to address the problem of hydrolysis associated with polar compounds: alcohol exchange reaction and chelation. In both processes, susceptibility of the reactants toward water is reduced. The alcohol exchange reaction is described below [59]:

$$\text{M(OR)}\_{\text{x}} + \text{xR'}OH \rightarrow \text{M(OR')}\_{\text{x}} + \text{xROH} \tag{4}$$

or

$$\text{M(OR)}\_{\text{x}} + \text{xR'OH} \rightarrow \text{M(OR)}\_{\text{x-1}}\text{(OR')} + \text{ROH} + \text{(x-1)R'OH} \tag{5}$$

The alcohol, 2-methoxyethnol, is widely used for alcohol exchange reaction due to its bidentate nature. Hence, the newly generated alkoxide is less prone to hydrolysis, thereby allowing easy formation of gel instead of precipitate [35]. As for example, 2-methoxyethanol was used to dissolve Ba metal and partially substitute propoxide of zirconium propoxide for the fabrication of epitaxial BZY film. Acetic acid is used for chelation. In some typical film fabrication method, both the acetic acid and 2-methoxyethanol are used together [60]. As 2-methoxyethnol does not change the solution pH, it is often used for the dilution of solution.

#### **2.2. Chelate process**

This is a specialized process of sol-gel technique which employs chelation reaction as the key process in the preparation of the precursor solution. This process with a wide range of application and variety of chemical compositions has opened a separate branch called chelate process. The process aims at the reduction in over-reactive tendency of alkoxides via addition of acetylacetonate or diethanolamine. In most compositions, metal carboxylate, β-diketonate, and alkoxides are used as precursors [59]. Mostly, transition metal alkoxides show polarity with high tendency toward rapid hydrolysis and condensation, requiring complexing or chelating ligands to limit uncontrolled reactions. Acetic acid and acetylacetone are added to the precursor solution to alleviate the issue. Acetic acid can act as both chelating and bridging agent, while acetylacetone only acts as chelating agent. This chelating agent blocks hydrolysis site with the replacement of reactive alkoxide ligand. A typical reaction with acetic acid is described below:

$$\text{M(OR)}\_{\text{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\text{H}}}}}}}}} + \text{xCH}\_{\text{\tiny{\text{\tiny{\text{H}}}}}} \text{COOH} \rightarrow \text{M(OR)}\_{\text{\tiny{\text{H}}-\text{x}}} \text{(CH}\_{\text{\tiny{\text{\tiny{\text{H}}}}})} + \text{xROH} \tag{6}$$

This typical reaction states that the species contains both the acetate and alkoxide ligands. Being bidentate in nature [61] and sterically larger than alkoxy group, the acetate ligands are not much susceptible to hydrolysis. In addition to this, acetylacetone plays the role of stabilizer of colloidal solution as well since it prevents aggregation of colloidal particles by creating steric hindrance [62]. In a typical chelate process, YDSZ was prepared via chelate route using acetylacetone [49]. This chelate process has been widely used to fabricate YSZ and BZY films for SOFC electrolytes [36, 40, 50–52, 63].

#### **2.3. Metallo-organic decomposition (MOD)**

This method involves high molecular weight precursors such as water-insensitive carboxylates and 2-ethylhexanoates. This process is less common than the other methods [59, 64]. This method is straightforward without necessitating precise control of the chemistry. Long-chain carboxylate compounds such as lead 2-ethylhexanoate, titanium dimethoxy dineodecanoate, zirconium neodecanoate, etc. are used as precursors, whereas the common solvent is xylene. The method involves simply dissolution of the metallo-organic compounds in a common solvent. The organic moieties of long-chain length compounds enhance dissolution tendency and concomitantly hinder hydrolysis tendency. They are normally dissolved in common solvent such as xylene [59, 64–66]. As these precursors are water insensitive and nonreactive to one another, they do not undergo any structural or chemical change [56]. This process is the simplest one among the three methods since no skill for controlling the hydrolysis and condensation is necessary. Still, this process suffers from several limitations. First, the large organic chains may cause crack during its decomposition upon heat treatment. Second, modification of the solution properties is limited; hence, the microstructure of the thin film cannot be tailored. This method has been applied specially for the ferroelectric materials [56]. The application of this method is not so popular in the field of SOFCs.

### **2.4. Other processing routes**

**2.2. Chelate process**

324 Modern Technologies for Creating the Thin-film Systems and Coatings

described below:

for SOFC electrolytes [36, 40, 50–52, 63].

**2.3. Metallo-organic decomposition (MOD)**

This is a specialized process of sol-gel technique which employs chelation reaction as the key process in the preparation of the precursor solution. This process with a wide range of application and variety of chemical compositions has opened a separate branch called chelate process. The process aims at the reduction in over-reactive tendency of alkoxides via addition of acetylacetonate or diethanolamine. In most compositions, metal carboxylate, β-diketonate, and alkoxides are used as precursors [59]. Mostly, transition metal alkoxides show polarity with high tendency toward rapid hydrolysis and condensation, requiring complexing or chelating ligands to limit uncontrolled reactions. Acetic acid and acetylacetone are added to the precursor solution to alleviate the issue. Acetic acid can act as both chelating and bridging agent, while acetylacetone only acts as chelating agent. This chelating agent blocks hydrolysis site with the replacement of reactive alkoxide ligand. A typical reaction with acetic acid is

*M OR xCH COOH M OR CH COO xROH <sup>n</sup>* 3 3 *n x <sup>x</sup>* ( ) ( )( ) - +® + (6)

This typical reaction states that the species contains both the acetate and alkoxide ligands. Being bidentate in nature [61] and sterically larger than alkoxy group, the acetate ligands are not much susceptible to hydrolysis. In addition to this, acetylacetone plays the role of stabilizer of colloidal solution as well since it prevents aggregation of colloidal particles by creating steric hindrance [62]. In a typical chelate process, YDSZ was prepared via chelate route using acetylacetone [49]. This chelate process has been widely used to fabricate YSZ and BZY films

This method involves high molecular weight precursors such as water-insensitive carboxylates and 2-ethylhexanoates. This process is less common than the other methods [59, 64]. This method is straightforward without necessitating precise control of the chemistry. Long-chain carboxylate compounds such as lead 2-ethylhexanoate, titanium dimethoxy dineodecanoate, zirconium neodecanoate, etc. are used as precursors, whereas the common solvent is xylene. The method involves simply dissolution of the metallo-organic compounds in a common solvent. The organic moieties of long-chain length compounds enhance dissolution tendency and concomitantly hinder hydrolysis tendency. They are normally dissolved in common solvent such as xylene [59, 64–66]. As these precursors are water insensitive and nonreactive to one another, they do not undergo any structural or chemical change [56]. This process is the simplest one among the three methods since no skill for controlling the hydrolysis and condensation is necessary. Still, this process suffers from several limitations. First, the large organic chains may cause crack during its decomposition upon heat treatment. Second, modification of the solution properties is limited; hence, the microstructure of the thin film cannot be tailored. This method has been applied specially for the ferroelectric materials [56].

The application of this method is not so popular in the field of SOFCs.

Although the abovementioned three processes have found extensive application, there are several other routes such as Pechini method (old method), aqueous solution, citrate, nitrate routes, etc. [56, 59]. Pechini method is an aqueous chemical route which involves dissolution of metal cations and hydroxycarboxylic acid (such as citric acid, etc.) and ethylene glycol in deionized water. Till date, simple Pechini method has not shown any remarkable progress in the field of SOFC electrolytes. Pechini method modified with ethylenediaminetetraacetic acid (EDTA) has been proved to be an efficient technique for the fabrication of nonporous thin-film gadolinia-doped barium cerate (BCG) for SOFC [67]. Citrate process is also similar to the Pechini method [56] without involving ethylene glycol. This method modified with EDTA as chelating agent successfully produced electrolyte film without through-film crack [38]. Aqueous method includes dissolution of metal nitrates or chlorides and other polymerizing agents such as polyvinylpyrrolidone (PVP) in deionized water [34]. A SOFC cell with 0.5 μm thick crack-free YSZ electrolyte film was fabricated via this method. YSZ thin films obtained by both the aqueous and the nonaqueous processes are identical as per the surface morphology, which gives the direction toward the development of aqueous method in future.

#### **2.5. Combined colloidal CSD method**

The limitation of sol-gel, chelate, and MOD processes lies in obtaining film thicker than 0.5 μm with the application of single-layer coating [68]. Therefore, the multi-coating approach was required to obtain thicker film. Still, there is limitation for obtaining crack-free film thicker than 10 μm due to the constraining effect of the substrate. Hence, combined colloidal CSD method was introduced [32, 63, 68, 69]. This method involves addition of pre-synthesized nano-powder to CSD solution. This newly formed system consists of precursors, nano-powder (either synthesized via sol-gel route or commercially purchased), and at least one solvent [69]. Nanoparticles are introduced in the chemical solution to encounter the external constraint due to the presence of substrate and reduce the extent of differential densification in the planar dimension. As the metal-oxide network shrinks faster than the substrate, shrinkage mismatch occurs between the film and the substrate. Since the nanoparticles are supposed to sinter at a significantly lower rate than the metal-oxide framework [31], nanoparticles were added to the solution. To date, there are several significant research works on the YSZ thin film and doped ceria with the thickness ranging from nanometer to several micrometer via CSD method or combined CSD method [32–34, 39, 60, 63, 68, 70–72]. Several groups have already made micro-SOFCs based on CSD techniques [39, 70, 72].

### **3. Deposition techniques**

The precursor solution is deposited over the substrate via a number of coating techniques. The most widely used coating techniques are spin coating, dip coating, and spray coating, as illustrated in **Figure 2**. In this section, the film deposition techniques and their application to SOFC electrolytes will be discussed.

**Figure 2.** Schematic diagrams of (a) spin, (b) dip, and (c) spray coating.

#### **3.1. Spin coating**

Spin coating is a simple process to deposit uniform thin films on relatively flat substrates (**Table 2**). Typically, the substrate to be coated is held in place by vacuum chuck, and the coating solution is dispensed onto the substrate. The substrate is then accelerated to the desired rotation speed for certain duration to obtain desirable film thickness. The excess liquid is spread out from the substrate due to the action of centrifugal force, leaving a thin uniform coating on the surface of the substrate. The major advantages of spin coating are reproducibility, uniformity, simplicity, ability to use different substrate materials, and low cost. The main disadvantage of the method is the requirement of the smooth and flat substrate.


**Table 2.** Three coating techniques used for CSD processes.

The thickness of the coating film is influenced by the spinning speed and time as well as the solution viscosity. According to the empirical equation, the film thickness (t) is inversely proportional to the square root of the spin speed (ω: angular velocity) [73]:

$$t \propto \frac{1}{\sqrt{\omega}}\tag{7}$$

Spinning speed of 2000–3000 rpm is usually used for depositing SOFC electrolyte thin films. The film thickness of 30–100 nm is obtained after one layer deposition depending on the viscosity of the solution and duration of spinning. Typical film thicknesses of SOFC electrolytes are kept usually below 1 μm [31, 34, 38, 44, 74–76]. Several works on electrolyte fabrication with spin coating technique have been reported in the field of SOFCs [31, 38, 76].

### **3.2. Dip coating**

**Figure 2.** Schematic diagrams of (a) spin, (b) dip, and (c) spray coating.

326 Modern Technologies for Creating the Thin-film Systems and Coatings

Evenly coats the substrates with CSD solution due to the rotation of the substrate placed on vacuum

Immersion of a substrate into CSD solution and its subsequent removal

Deposition of the aerosol of CSD solution, via a nebulizer or a nozzle, on the substrate

**Table 2.** Three coating techniques used for CSD processes.

chuck

Spin coating is a simple process to deposit uniform thin films on relatively flat substrates (**Table 2**). Typically, the substrate to be coated is held in place by vacuum chuck, and the coating solution is dispensed onto the substrate. The substrate is then accelerated to the desired rotation speed for certain duration to obtain desirable film thickness. The excess liquid is spread out from the substrate due to the action of centrifugal force, leaving a thin uniform coating on the surface of the substrate. The major advantages of spin coating are reproducibility, uniformity, simplicity, ability to use different substrate materials, and low cost. The main

> **•** Relatively low throughput **•** Requirement of smooth and flat substrate

**•** Variation in film thickness **•** Coating on the both sides of the substrate simultaneously **•** Semiconductors **•** Photoresist **•** Insulators

**•** SOFC

**•** Expensive **•** Coating on dielectrics

**•** Coating for corrosion protection, etc.

**•** Organic semiconductors

**•** Flat or cylindrical substrates **•** Optical coating on glass, etc.

electrolyte and cathode, etc.

disadvantage of the method is the requirement of the smooth and flat substrate.

**Technique Description Advantages Disadvantages Application**

**•** Simplicity **•** Thin and uniform coating **•** Low cost

**•** Simplicity **•** Controllable film thickness **•** Coating on irregular and complex shaped substrates

**•** Uniform coating even on highly structured surfaces

**3.1. Spin coating**

Spin coating

Dip coating

Spray coating Dip coating is a simple, flexible, and cost-effective solution deposition technique that allows coating on both large area and complex shaped substrates. The process involves immersion of substrate in the solution and subsequent removal. A coherent liquid film is entrained on the withdrawal of the substrate from the coating fluid, and the film is subsequently consolidated by drying and accompanying chemical reactions. Typical film thickness obtained is in the micrometer range.

The process of film formation follows fluid mechanical equilibrium between the entrained film and the receding liquid. The equilibrium is governed by several forces. Viscous drag and gravitational forces play the most significant role. Other forces like surface tension, inertial force, or disjoining pressure also play an important role [77]. A competition between these forces in the film deposition region governs the thickness of the film. The film thickness is given by the Landau-Levich equation [78, 79]:

$$t = 0.94 \frac{\eta \mathcal{U}^{2/3}}{\mathcal{Y}^{1/6} \rho \mathcal{Y}^{1/2}} \tag{8}$$

where *η* is the liquid viscosity, *U* is the withdrawal speed, *γ* is the surface tension, and *ρ* is the liquid density. This technique is not applied for electrolyte fabrication of SOFCs because this technique aids in the formation of film on both sides of the substrate.

### **3.3. Spray coating**

The spray coating technique is based on the transformation of a liquid precursor solution into a fine aerosol by atomizer or nebulizer [80]. These fine droplets are then deposited on a substrate surface either with carrier gas or with an electrostatic field or by gravity. The substrate may be at room temperature or above. The different spray techniques are mainly distinguished by the method of atomization [81] and, hence, produce different morphologies of the film. For example, YSZ thin films deposited using electrostatic spray deposition (ESD) and pressurized spray deposition (PSD) on the substrate [82] showed different morphologies from dense to porous. Other parameters such as flow rate, substrate temperature, and deposition time also influence the morphology of the deposited film. The flow rate should not exceed a certain value for a particular composition of the solution and associated parameters; otherwise, cracks may form. Spray pyrolysis has also been applied for the deposition of SOFC electrolyte thin films [71, 83, 84]. To mention a significant work, a gastight bilayer electrolyte fabricated via spray coating technique demonstrated a power density of 750 mW/cm2 with the achievement of 1.01 V OCV at 770°C [84].

### **4. Solution chemistry**

The chemistry of the solution determines the morphology of the film. A wide knowledge of chemistry is required to formulate workable solution. Attention must be paid to several issues such as reactivity among the precursors and solvent, homogeneity of the solution, solvent vapor pressure, wettability of the solvent to the substrate surface, reaction products, pH and viscosity of the solution, etc.

The hydrolysis and condensation reaction needs to be controlled carefully to tailor the morphology of the film. Precursors with more than two hydrolysis sites are highly sensitive toward hydrolysis and form three-dimensional networks during gelation, which provides rigidity to the M–O–M network and inhibits densification of the final sintered film. Water, chelating agent, and modifying ligand are added to block the hydrolysis sites of the precursors. Precursors with two unblocked sites form linear structure, which facilitates almost stress-free densification in all directions.

Transition metal alkoxide precursors require special attention because of their high reactivity and inclination toward coordination expansion. Due to their coordination expansion, these metals become coordinatively unsaturated. In order to satisfy their coordination, they sometimes get integrated with water and subsequently undergo precipitation. These highly sensitive alkoxides need to be handled in glove box initially. Diethanolamine (DEA) and triethanolamine (TEA) are generally used to stabilize alkoxides of transition metals [85]. Stabilization also occurs via chelation and alcohol exchange method. After modification, they do not react with moisture in air.

Compositional homogeneity is desirable for the successful fabrication of film without any defect. Striation is a very common problem associated with heterogeneous solution. Striation is a series of ridges resulting into variation in thickness throughout the film [86]. Heterogeneity is associated with the separation of polymer-rich and polymer-deficient portion of the solution due to the presence of both polar and nonpolar precursors in a multicomponent system. Therefore, a single solvent with both the characters is desirable to maintain homogeneity. For example, 2-methoxyethanol having both the characters is a widely used solvent. Heterogeneous solution also results from the mixture of the solvents of different characters such as specific gravity. Phase separation is another issue associated with the heterogeneous solution, which occurs due to the different rate of the hydrolysis of different components. Refluxing treatment is carried out to address the issue. This treatment helps in random combination of cations [87].

Solvent vapor pressure is one important parameter because solvent determines the film thickness and its rigidity. The short-chain alcohols are generally used for thinner film, while the long-chain alcohols are for thicker films [88, 89]. Short-chain alcohols have higher tendency to leave film faster because of its higher vapor pressure. Higher vapor pressure generates higher capillary force, which drives precursors in greater proximity, thereby causing higher cross-linking among metal-oxide precursors. This cross-link offers rigidity to the film, producing crack in the film [29]. On the other hand, the solvents with low vapor pressure hinder cross-linking reaction, resulting into the crack-free film.

The solution pH and the product generated during condensation reaction have immense influence over the rate of condensation reaction [90, 91]. The reaction products are alcohol and water. Alcohol is eliminated during deposition, which forces the condensation reaction to shift toward the forward direction. Thus, more M–O–M cross-linked network forms. The presence of water in the solution or in the ambient atmosphere slows down the condensation process. The early work showed that the density of the film was enhanced with the presence of more water [91].

Viscosity and concentration of the solution are other variables to control the thickness and the initiation of the crack throughout the film. Early work demonstrated that the higher concentration of the solution produced the thicker film with crack. As per the literature report, the critical thickness limit for the film is governed by the following equation [92]:

$$h = \frac{K\_{\rm loc}}{\sigma \Omega\_{\rm c} \left(\Sigma\right)}\tag{9}$$

where *h* is the critical thickness, *KIC* is the critical stress intensity factor, *σ* is the tensile stress in the film, and *Ωc*(*Σ*) is the ratio of Young's modulus of the film to that of the substrate.


**Table 3.** Solution properties.

example, YSZ thin films deposited using electrostatic spray deposition (ESD) and pressurized spray deposition (PSD) on the substrate [82] showed different morphologies from dense to porous. Other parameters such as flow rate, substrate temperature, and deposition time also influence the morphology of the deposited film. The flow rate should not exceed a certain value for a particular composition of the solution and associated parameters; otherwise, cracks may form. Spray pyrolysis has also been applied for the deposition of SOFC electrolyte thin films [71, 83, 84]. To mention a significant work, a gastight bilayer electrolyte fabricated via spray coating technique demonstrated a power density of 750 mW/cm2 with the achievement of

The chemistry of the solution determines the morphology of the film. A wide knowledge of chemistry is required to formulate workable solution. Attention must be paid to several issues such as reactivity among the precursors and solvent, homogeneity of the solution, solvent vapor pressure, wettability of the solvent to the substrate surface, reaction products, pH and

The hydrolysis and condensation reaction needs to be controlled carefully to tailor the morphology of the film. Precursors with more than two hydrolysis sites are highly sensitive toward hydrolysis and form three-dimensional networks during gelation, which provides rigidity to the M–O–M network and inhibits densification of the final sintered film. Water, chelating agent, and modifying ligand are added to block the hydrolysis sites of the precursors. Precursors with two unblocked sites form linear structure, which facilitates almost stress-free

Transition metal alkoxide precursors require special attention because of their high reactivity and inclination toward coordination expansion. Due to their coordination expansion, these metals become coordinatively unsaturated. In order to satisfy their coordination, they sometimes get integrated with water and subsequently undergo precipitation. These highly sensitive alkoxides need to be handled in glove box initially. Diethanolamine (DEA) and triethanolamine (TEA) are generally used to stabilize alkoxides of transition metals [85]. Stabilization also occurs via chelation and alcohol exchange method. After modification, they

Compositional homogeneity is desirable for the successful fabrication of film without any defect. Striation is a very common problem associated with heterogeneous solution. Striation is a series of ridges resulting into variation in thickness throughout the film [86]. Heterogeneity is associated with the separation of polymer-rich and polymer-deficient portion of the solution due to the presence of both polar and nonpolar precursors in a multicomponent system. Therefore, a single solvent with both the characters is desirable to maintain homogeneity. For example, 2-methoxyethanol having both the characters is a widely used solvent. Heterogeneous solution also results from the mixture of the solvents of different characters such as specific gravity. Phase separation is another issue associated with the heterogeneous solution, which

1.01 V OCV at 770°C [84].

328 Modern Technologies for Creating the Thin-film Systems and Coatings

**4. Solution chemistry**

viscosity of the solution, etc.

densification in all directions.

do not react with moisture in air.

The equation describes the dependence of the critical thickness on the tensile stress exerted on it. Critical thickness decreases with the increase in tensile stress. The critical thickness of film can be increased with the use of longer-chain solvents [89]. Another approach to deal with this issue is to increase the adhesion between the substrate and film [29]. During shrinkage, formation of crack occurs due to the large mismatch of thermal expansion coefficients of the film and the substrate. The strain energy is relieved via the formation of crack, but this strain energy can be balanced by the strength of adhesion of the film to the substrate. Excellent substrate-film adhesion may provide relaxation for the fabrication of thicker film. Therefore, modification of substrate surface before deposition has significance. The properties of solution and their effects are summarized in **Table 3**.

### **5. Heat treatment**

#### **5.1. Physical changes occurring during heat treatment**

Several phenomena occur after deposition of film. They are hydrolysis, drying, condensation, gelation, and densification [93]. Generally, gelation and drying phenomena occur simultaneously during deposition and continue afterward. The deposited film acts as a viscoelastic solid, which is an inorganic framework with organic moieties entrapping solvent [94]. These organics are removed with heat treatment via either pyrolysis or thermolysis process. Pyrolysis occurs in the presence of oxygen, whereas thermolysis occurs in the absence of oxygen. Followed by organic removal, crystallization occurs. Partial densification takes place in the amorphous stage, while the final densification occurs after crystallization.

Hydrolysis-condensation reaction generally occurs in the temperature range of 80 to 400°C. This reaction generates water and alcohols, which are removed via drying process. During drying or organic removal process, a gas-liquid interface is generated within the pores because of the evaporation of liquid. In addition to this, gas-solid and solid-liquid interfaces also exist. These interfaces generate pressure, which gives birth to capillary contraction. The magnitude of capillary contraction depends on the specific energies across these interfaces, nature of the liquid, and pore size. This capillary contraction is responsible for producing the driving force to the collapse of the amorphous network. This driving force is proportional to the pore diameter. At this stage, the film might be prone to crack because of the pressure differential occurring due to the presence of pores with different diameters. The total stress due to drying, capillary contraction, and network consolidation is normally around 100 MPa [95].

Gelation occurs due to the continuous removal of water and organics. During gelation, M–O–M network starts forming. With the heat treatment, a good number of M–O–M linkages form and densification proceeds. Skeletal densification occurs with the structural rearrangement of M–O–M bond. The structure approaches to the state of metastable liquid in the temperature range of 400–600°C. With the increase in temperature, viscous flow occurs [51], followed by crystallization above 600°C. However, crystallization starts after complete removal of organics. Crystallization kinetics depends on its own nature of crystallization (i.e., glass formers have slow rate of crystallization, while other oxides show moderate to high rate of crystallization) and the heating schedule. Structural relaxation occurs during crystallization process. The presence of organic groups delays structural relaxation and crystallization to a higher temperature. Following this principle, crystallization is purposefully delayed so that densification gets over before crystallization [50]. All the processes overlap one another. There is no distinct temperature range. Depending on the solution chemistry, the processes may occur faster or slower.

### **5.2. Phase transformation**

The equation describes the dependence of the critical thickness on the tensile stress exerted on it. Critical thickness decreases with the increase in tensile stress. The critical thickness of film can be increased with the use of longer-chain solvents [89]. Another approach to deal with this issue is to increase the adhesion between the substrate and film [29]. During shrinkage, formation of crack occurs due to the large mismatch of thermal expansion coefficients of the film and the substrate. The strain energy is relieved via the formation of crack, but this strain energy can be balanced by the strength of adhesion of the film to the substrate. Excellent substrate-film adhesion may provide relaxation for the fabrication of thicker film. Therefore, modification of substrate surface before deposition has significance. The properties of solution

Several phenomena occur after deposition of film. They are hydrolysis, drying, condensation, gelation, and densification [93]. Generally, gelation and drying phenomena occur simultaneously during deposition and continue afterward. The deposited film acts as a viscoelastic solid, which is an inorganic framework with organic moieties entrapping solvent [94]. These organics are removed with heat treatment via either pyrolysis or thermolysis process. Pyrolysis occurs in the presence of oxygen, whereas thermolysis occurs in the absence of oxygen. Followed by organic removal, crystallization occurs. Partial densification takes place in the amorphous

Hydrolysis-condensation reaction generally occurs in the temperature range of 80 to 400°C. This reaction generates water and alcohols, which are removed via drying process. During drying or organic removal process, a gas-liquid interface is generated within the pores because of the evaporation of liquid. In addition to this, gas-solid and solid-liquid interfaces also exist. These interfaces generate pressure, which gives birth to capillary contraction. The magnitude of capillary contraction depends on the specific energies across these interfaces, nature of the liquid, and pore size. This capillary contraction is responsible for producing the driving force to the collapse of the amorphous network. This driving force is proportional to the pore diameter. At this stage, the film might be prone to crack because of the pressure differential occurring due to the presence of pores with different diameters. The total stress due to drying,

capillary contraction, and network consolidation is normally around 100 MPa [95].

Gelation occurs due to the continuous removal of water and organics. During gelation, M–O–M network starts forming. With the heat treatment, a good number of M–O–M linkages form and densification proceeds. Skeletal densification occurs with the structural rearrangement of M–O–M bond. The structure approaches to the state of metastable liquid in the temperature range of 400–600°C. With the increase in temperature, viscous flow occurs [51], followed by crystallization above 600°C. However, crystallization starts after complete removal of organics. Crystallization kinetics depends on its own nature of crystallization (i.e., glass formers have slow rate of crystallization, while other oxides show

and their effects are summarized in **Table 3**.

330 Modern Technologies for Creating the Thin-film Systems and Coatings

**5.1. Physical changes occurring during heat treatment**

stage, while the final densification occurs after crystallization.

**5. Heat treatment**

After completion of pyrolysis, liquid film transforms to the metastable amorphous stage. With further heat treatment to the higher temperature, the amorphous film transforms to the crystalline stage via nucleation and growth process. Thermodynamic driving force plays a role behind this transformation. This force comes from the difference between the free energies of those two states. **Figure 3** describes the phenomena. The driving force for transformation to the final stage depends on the crystallization temperature and the free energy associated with both stages of the films.

**Figure 3.** Thermodynamic driving force associated with phase transformation.

Crystallization kinetics starts with the nucleation, which is quite similar to the transformation from amorphous glassy to crystalline phase. The different features of CSD-derived film are associated with the presence of residual hydroxyl group, excess surface area associated with the porosity, and skeletal density. However, the nucleation and growth theory applicable for glass-ceramic science also applies to the transformation of the CSD-derived film [96]. Gibb's free energy (*ΔG*) associated with the driving force for crystallization is expressed by the following equation:

$$
\Delta G = -\frac{4}{3}\pi r^3 (\Delta G\_v) + 4\pi r^2 \sigma \tag{10}
$$

where *ΔGv* and *σ* are Gibb's free energies associated with unit volume and surface, respectively, and *r* is the radius of the nucleus formed during nucleation. The equation gives rise to the concept of critical radius. A critical radius (*r\**) is the minimum required radius for the formation of stable nucleus. The derivative of the above equation with respect to radius gives the relationship between the critical radius (*r\**) and the energy barrier (*ΔG\**) required to be overcome to form a stable nucleus, which is described by the following equation:

$$
\Delta G^{\star} = \frac{16\pi\sigma^3}{\Im(\Delta G\_{\upsilon})^2} \tag{11}
$$

This equation holds good for homogeneous nucleation where the amorphous film does not encounter any nucleation site. Heterogeneous nucleation occurs when the amorphous material can rest on nucleation site, i.e., any surface such as impurity, substrate, grain boundaries, etc. In case of heterogeneous nucleation, the above equation is modified by the contact angle term, *f*(*θ*), associated with the substrate surface (roughness) and the crystal:

$$
\Delta G\_{hetero}^{\*} = \frac{16\pi\sigma^3}{3(\Delta G\_v)^2} f(\theta) \tag{12}
$$

where

$$f(\theta) = \frac{2 - 3\cos\theta + \cos^3\theta}{4} \tag{13}$$

where *θ* is the contact angle between the substrate surface and the crystal. Heterogeneous nucleation is always energetically favorable because of the lower energy barrier due to the presence of the preferential nucleation sites.

Rates of nucleation and growth with respect to the temperature coordinate are important factors for tailoring microstructure of the film. Higher nucleation rate gives finer microstructure, while higher rate of growth with lower nucleation gives coarse microstructure. The relationship among temperature, nucleation density (number of nuclei in a cubic meter volume), and free energy barrier is governed by the following equation:

free energy (*ΔG*) associated with the driving force for crystallization is expressed by the

where *ΔGv* and *σ* are Gibb's free energies associated with unit volume and surface, respectively, and *r* is the radius of the nucleus formed during nucleation. The equation gives rise to the concept of critical radius. A critical radius (*r\**) is the minimum required radius for the formation of stable nucleus. The derivative of the above equation with respect to radius gives the relationship between the critical radius (*r\**) and the energy barrier (*ΔG\**) required to be

*v*

This equation holds good for homogeneous nucleation where the amorphous film does not encounter any nucleation site. Heterogeneous nucleation occurs when the amorphous material can rest on nucleation site, i.e., any surface such as impurity, substrate, grain boundaries, etc. In case of heterogeneous nucleation, the above equation is modified by the contact angle term,

where *θ* is the contact angle between the substrate surface and the crystal. Heterogeneous nucleation is always energetically favorable because of the lower energy barrier due to the

Rates of nucleation and growth with respect to the temperature coordinate are important factors for tailoring microstructure of the film. Higher nucleation rate gives finer microstructure, while

*πσ <sup>G</sup> G* 3 2

overcome to form a stable nucleus, which is described by the following equation:

<sup>16</sup> <sup>Δ</sup> 3( ) <sup>=</sup> <sup>D</sup>

å

*f*(*θ*), associated with the substrate surface (roughness) and the crystal:

presence of the preferential nucleation sites.

(10)

(11)

(12)

(13)

following equation:

332 Modern Technologies for Creating the Thin-film Systems and Coatings

where

$$m^\* \propto \exp\left(-\Delta \frac{G^\*}{RT}\right) \tag{14}$$

As the temperature is raised, higher energy is provided to overcome the nucleation barrier. Therefore, the rate of nucleation increases. After nucleation rate reaches the maximum height, it declines with further increase of temperature. Driving force for crystallization gets reduced as the material approaches its melting temperature [94] and barrier height to nucleation increases. This type of phenomena gives rise to a bell-shaped curve. The same kind of curve is also applicable for the growth rate. A typical curve is presented in **Figure 4**.

**Figure 4.** Schematic of nucleation and growth locus in temperature coordinate for homogeneous nucleation (ideal case).

The nucleation curve is followed by the growth rate curve on the temperature coordinate, and they overlap to some extent. The area of the overlapped region depends on chemical composition of the solution, fabrication procedure, and prior heat treatment history. The representation in **Figure 4** is applicable for the homogeneous nucleation [97] as the apex of the growth rate curve lies at higher temperature than that of the nucleation rate curve. Based on the calculation, the maximum nucleation rate may be located at higher temperature than the maximum growth rate in case of heterogeneous nucleation. The schematic representation of these phenomena is depicted in **Figure 5**.

**Figure 5.** Schematic of nucleation and growth locus in temperature coordinate for heterogeneous nucleation and homogeneous nucleation with low interfacial energy.

In case of the film on the substrate, crystallization is affected due to the presence of two separate nucleation events. One is the surface nucleation of the film which is homogeneous in nature, and the other one is the interfacial nucleation on the substrate. Nucleation occurs on both the substrate and film surfaces, as represented in **Figure 6**.

**Figure 6.** Kinetic competition between two nucleation events on the interface and on the surface of the film: (a) nucleation at the interface only and (b) nucleation at both interface and surface of the film.

Heterogeneous nucleation occurs at the substrate first, while homogeneous nucleation and growth event occur on the surface of the film; growth direction is opposite to each other. The density of nucleus is higher at the interface than at the surface of the film. The dominance of one process over the other depends on the crystallization temperature of the film. The film with lower crystallization temperature shows smaller difference in the kinetics of two nucleation events, than the film with higher crystallization temperature [98].

### **5.3. Salient features**

on the calculation, the maximum nucleation rate may be located at higher temperature than the maximum growth rate in case of heterogeneous nucleation. The schematic representation

**Figure 5.** Schematic of nucleation and growth locus in temperature coordinate for heterogeneous nucleation and ho-

In case of the film on the substrate, crystallization is affected due to the presence of two separate nucleation events. One is the surface nucleation of the film which is homogeneous in nature, and the other one is the interfacial nucleation on the substrate. Nucleation occurs on both the

**Figure 6.** Kinetic competition between two nucleation events on the interface and on the surface of the film: (a) nuclea-

Heterogeneous nucleation occurs at the substrate first, while homogeneous nucleation and growth event occur on the surface of the film; growth direction is opposite to each other. The

of these phenomena is depicted in **Figure 5**.

334 Modern Technologies for Creating the Thin-film Systems and Coatings

mogeneous nucleation with low interfacial energy.

substrate and film surfaces, as represented in **Figure 6**.

tion at the interface only and (b) nucleation at both interface and surface of the film.

Heating rate also has impact over the kinetics of these processes. Various heating schedules with several heating rates are used to keep control over the microstructure of the film. Keddie and Giannelis experimented on the effect of heating rate (0.2–8000°C/min) forthe densification of TiO2 film as model system [99] and found that the thinnest film was obtained with the highest heating rate. Rapid thermal annealing (RTA) and isothermal heat treatment at high temperature are two useful processes for thin-film fabrication [50, 99].

Effect of the substrate is another important issue for the orientation of thin film. A highly oriented phase-pure barium zirconate (BaZrO3) film fabricated via sol-gel route was epitaxially grown on (100) plane on strontium titanate (SrTiO3) at 800°C, while the same solution deposited on LaAlO3 substrate had produced film with random structure [35].

### **6. Conclusion**

The CSD method combined with right sintering strategy is the blessing in the sector in SOFC manufacturing. The growth of the SOFC technologies is restricted mostly to the R&D sector because suitable position in the market of clean energy sources is yet to be achieved. The hurdle behind commercialization of intermediate to low-temperature SOFCs arises from its high manufacturing cost. The CSD method has proved to be a potential technique in this field. Although this method did not find application for over a century after its discovery, now it is progressing exponentially with time in all the sectors of thin film. The fabrication of crystalline YSZ and YDSZ film has initiated a new era in the manufacturing of SOFCs, which ensures the rapid commercialization of SOFCs in the near future. This technology requires low investment cost while demanding a good command over chemistry. Therefore, CSD technology is facing several difficulties related to the fabrication strategies. The major issues are preparation of homogeneous solution for multicomponent system, preservation of the solution for a considerable duration without aging, handling with highly reactive precursors, controlling the hydrolysis-condensation reaction, gelation kinetics, solvent evaporation, choice of environment-friendly chemicals without compromising necessary properties, selection of appropriate deposition technique, right heating schedule and environment, determination of critical thickness limit, etc. Many challenges have been successfully dealt with, while a good number of issues need more attention. However, recent success stories predict the bright future of this technology. Presently, effort is also given to the water-based CSD method.

### **Author details**

Mridula Biswas and Pei-Chen Su\*

\*Address all correspondence to: peichensu@ntu.edu.sg

School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore, Singapore

### **References**


[11] Isenberg A. Energy conversion via solid oxide electrolyte electrochemical cells at high temperatures. Solid State Ionics. 1981;3:431–437.

**Author details**

Singapore, Singapore

**References**

Mridula Biswas and Pei-Chen Su\*

tions. 2014;47:1–4.

Society. 2007;154(1):B20–B24.

Technology. 1976;13(1):72–75.

1977;48(2):775–780.

\*Address all correspondence to: peichensu@ntu.edu.sg

336 Modern Technologies for Creating the Thin-film Systems and Coatings

School of Mechanical and Aerospace Engineering, Nanyang Technological University,

[1] Hong S, Bae J, Koo B, Kim YB. High-performance ultra-thin film solid oxide fuel cell using anodized-aluminum-oxide supporting structure. Electrochemistry Communica-

[2] Su PC, Chao CC, Shim JH, Fasching R, Prinz FB. Solid oxide fuel cell with corrugated

[3] Shim JH, Chao CC, Huang H, Prinz FB. Atomic layer deposition of yttria-stabilized zirconia for solid oxide fuel cells. Chemistry of Materials. 2007;19(15):3850–3854.

[4] Huang H, Nakamura M, Su P, Fasching R, Saito Y, Prinz FB. High-performance ultrathin solid oxide fuel cells for low-temperature operation. Journal of The Electrochemical

[5] Huang H, Gür TM, Saito Y, Prinz F. High ionic conductivity in ultrathin nanocrystalline

[6] Su PC, Prinz FB. Nanoscale membrane electrolyte array for solid oxide fuel cells.

[7] Chao CC, Hsu CM, Cui Y, Prinz FB. Improved solid oxide fuel cell performance with

[8] Choi H, Cho GY, Cha SW. Fabrication and characterization of anode supported YSZ/GDC bilayer electrolyte SOFC using dry press process. International Journal of

[9] Greene J, Wickersham C, Zilko J, Welsh L, Szofran F. Morphological and electrical properties of rf sputtered Y2O3-doped ZrO2 thin films. Journal of Vacuum Science &

[10] Croset M, Schnell JP, Velasco G, Siejka J. Composition, structure, and ac conductivity of rf-sputtered calcia-stabilized zirconia thin films. Journal of Applied Physics.

Precision Engineering and Manufacturing-Green Technology. 2014;1(2):95–99.

gadolinia-doped ceria films. Applied Physics Letters. 2006;89(14):3107.

thin film electrolyte. Nano Letters. 2008;8(8):2289–2292.

Electrochemistry Communications. 2012;16(1):77–79.

nanostructured electrolytes. ACS Nano. 2011;5(7):5692–5696.


[40] Biswas M, Xie H, Su PC. Low temperature synthesis of sub-micrometer yttria-doped barium zirconate thin films by modified chemical solution deposition technique. ECS Transactions. 2015;68(1):481–488.

[26] Dislich H. Sol-gel: Science, processes and products. Journal of Non-Crystalline Solids.

[27] Schroeder H. Properties and applications of oxide layers deposited on glass from

[28] Dislich H, Hussmann E. Amorphous and crystalline dip coatings obtained from organometallic solutions: Procedures, chemical processes and products. Thin Solid

[29] Kueper TW, Visco SJ, De Jonghe L. Thin-film ceramic electrolytes deposited on porous and non-porous substrates by sol-gel techniques. Solid State Ionics. 1992;52(1):251–259.

[30] Mehta K, Xu R, Virkar AV. Two-layer fuel cell electrolyte structure by sol-gel processing.

[31] Oh EO, Whang CM, Lee YR, Park SY, Prasad DH, Yoon KJ, et al. Extremely thin bilayer electrolyte for solid oxide fuel cells (SOFCs) fabricated by chemical solution deposition

[32] Courtin E, Boy P, Piquero T, Vulliet J, Poirot N, Laberty-Robert C. A composite sol-gel process to prepare a YSZ electrolyte for solid oxide fuel cells. Journal of Power Sources.

[33] Bailly N, Georges S, Djurado E. Elaboration and electrical characterization of electrosprayed YSZ thin films for intermediate temperature-solid oxide fuel cells (IT-SOFC).

[34] Chen YY, Wei WCJ. Processing and characterization of ultra-thin yttria-stabilized zirconia (YSZ) electrolytic films for SOFC. Solid State Ionics. 2006;177(3):351–357. [35] Paranthaman M, Shoup SS, Beach DB, Williams RK, Specht ED. Epitaxial growth of BaZrO3 films on single crystal oxide substrates using sol-gel alkoxide precursors.

[36] Schneller T, Schober T. Chemical solution deposition prepared dense proton conducting Y-doped BaZrO3 thin films for sofc and sensor devices. Solid State Ionics.

[37] Groß B, Engeldinger J, Grambole D, Herrmann F, Hempelmann R. Dissociative water vapour absorption in BaZr 0.85Y0.15O2.925/H2O: Pressure-compositions isotherms in terms of fermi-dirac statistics. Physical Chemistry Chemical Physics. 2000;2(2):297–301. [38] Xie H, Su PC. Fabrication of yttrium-doped barium zirconate thin films with submicrometer thickness by a sol-gel spin coating method. Thin Solid Films. 2015;

[39] Rupp JL, Infortuna A, Gauckler LJ. Thermodynamic stability of gadolinia-doped ceria thin film electrolytes for micro-solid oxide fuel cells. Journal of the American Ceramic

Journal of Sol-Gel Science and Technology. 1998;11(2):203–207.

(CSD). Advanced Materials. 2012;24(25):3373–3377.

Materials Research Bulletin. 1997;32(12):1697–1704.

1986;80(1):115–121.

Films. 1981;77(1):129–140.

2012;206:77–83.

2003;164(3):131–136.

584:116–119.

Society. 2007;90(6):1792–1797.

Solid State Ionics. 2012;222:1–7.

organic solutions. Optica Acta. 1962;9:249.

338 Modern Technologies for Creating the Thin-film Systems and Coatings


[68] Barrow D, Petroff T, Sayer M. Thick ceramic coatings using a sol gel based ceramicceramic 0–3 composite. Surface and Coatings Technology. 1995;76:113–118.

[53] Tuttle B, Schwartz R. Solution deposition of ferroelectric thin films. MRS Bulletin.

[54] Schwartz RW, Reichert TL, Clem PG, Dimos D, Liu D. A comparison of diol and methanol-based chemical solution deposition routes for PZT thin film fabrication.

[55] Ramamurthi SD, Payne DA. Structural investigations of prehydrolyzed precursors used in the sol-gel processing of lead titanate. Journal of the American Ceramic Society.

[56] Schwartz RW. Chemical solution deposition of perovskite thin films. Chemistry of

[57] Brinker CJ, Scherer GW. Sol → gel → glass: I. Gelation and gel structure. Journal of Non-

[58] Esquivias L, Kallala M, Sanchez C, Cabane B. Advanced materials from gels SAXS study of gelation and precipitation in titanium-based systems. Journal of Non-Crystalline

[59] Brinker CJ, Scherer GW. Sol-Gel Science: The Physics and Chemistry of Sol-Gel

[60] Veldhuis SA, George A, Nijland M, ten Elshof JE. Concentration dependence on the shape and size of sol-gel-derived yttria-stabilized zirconia ceramic features by soft

[61] Doeuff S, Henry M, Sanchez C, Livage J. Hydrolysis of titanium alkoxides: Modification of the molecular precursor by acetic acid. Journal of Non-Crystalline Solids. 1987;89(1):

[62] Chatry M, Henry M, Livage J. Synthesis of non-aggregated nanometric crystalline

[63] Courtin E, Boy P, Rouhet C, Bianchi L, Bruneton E, Poirot N, et al. Optimized sol-gel routes to synthesize yttria-stabilized zirconia thin films as solid electrolytes for solid

[64] Joshi V, Dacruz C, Cuchiaro J, Araujo C, Zuleeg R. Analysis of C-V and I-V data of BST

[65] Vest RW, Jiejie X. PbTiO3/films from metalloorganic precursors. IEEE Transactions on

[66] Cui T, Markus D, Zurn S, Polla D. Piezoelectric thin films formed by MOD on cantilever beams for micro sensors and actuators. Microsystem Technologies. 2004;10(2):137–141.

[67] Agarwal V, Liu M. Preparation of barium cerate-based thin films using a modified

Ultrasonics, Ferroelectrics, and Frequency Control. 1988;35(6):711–717.

pechini process. Journal of Materials Science. 1997;32(3):619–625.

1996;21(06):49–54.

1990;73(8):2547–2551.

Solids. 1992;147:189–193.

206–216.

Materials. 1997;9(11):2325–2340.

Crystalline Solids. 1985;70(3):301–322.

Processing. Academic Press, Inc.; New York. 2013.

lithographic patterning. Langmuir. 2012;28(42):15111–15117.

zirconia particles. Materials Research Bulletin. 1994;29(5):517–522.

oxide fuel cells. Chemistry of Materials. 2012;24(23):4540–4548.

thin films. Integrated Ferroelectrics. 1997;14(1–4):133–140.

Integrated Ferroelectrics. 1997;18(1–4):275–286.

340 Modern Technologies for Creating the Thin-film Systems and Coatings


lization behavior of sol-gel derived lead zirconate titanate thin films. Journal of Materials Research. 1997;12(2):444–456.

[99] Keddie JL, Giannelis EP. Effect of heating rate on the sintering of titanium dioxide thin films: Competition between densification and crystallization. Journal of the American Ceramic Society. 1991;74(10):2669–2671.

[83] Somroop K, Pornprasertsuk R, Jinawath S. Fabrication of Y2O3-doped BaZrO3 thin films by electrostatic spray deposition. Thin Solid Films. 2011;519(19):6408–6412.

[84] Perednis D, Gauckler L. Solid oxide fuel cells with electrolytes prepared via spray

[85] Takahashi Y, Matsuoka Y. Dip-coating of TiO2 films using a sol derived from Ti(Oi-Pr)4-diethanolamine-H2O-i-PrOH system. Journal of Materials Science. 1988;23(6):

[86] Haas D, Birnie D, Zecchino M, Figueroa J. The effect of radial position and spin speed on striation spacing in spin on glass coatings. Journal of Materials Science Letters.

[87] Bradley D, Mehrotra R, Gaur D. Metal Alkoxides. Academic Press, London; 1978.

deposition. Journal of Materials Research. 2007;22(10):2868–2874.

Symposium Proceedings. Cambridge University Press; USA. 1986.

Oxide Thin Films. Springer-Verlag Wien; Heidelberg. 2013.

and Engineering. John Wiley & Sons; New York. 1997.

Journal of Non-Crystalline Solids. 2015;429:24–32.

[88] Brennecka GL, Tuttle BA. Fabrication of ultrathin film capacitors by chemical solution

[89] Tu Y, Milne S. Processing and characterization of Pb(Zr,Ti)O3 films, up to 10 μm thick, produced from a diol sol-gel route. Journal of Materials Research. 1996;11(10):2556–

[90] Brinker CJ, Clark DE, Ulrich DR. Better ceramics through chemistry II. In: MRS

[91] Glaser P, Pantano CG. Effect of the H2O/TEOS ratio upon the preparation and nitridation of silica sol/gel films. Journal of Non-Crystalline Solids. 1984;63(1):209–221.

[92] Garino TJ. The cracking of sol-gel films during drying. In: MRS Proceedings. Cambridge

[93] Schmidt H. Chemistry of material preparation by the sol-gel process. Journal of Non-

[94] Schneller T, Waser R, Kosec M, Payne D. Chemical Solution Deposition of Functional

[95] Garino TJ, Harrington M. Residual stress in PZT thin films and its effect on ferroelectric properties. In: MRS Proceedings. Cambridge University Press; USA. 1991.

[96] Chiang YM, Kingery WD, Birnie DP. Physical Ceramics: Principles for Ceramic Science

[97] Schmelzer JW, Abyzov AS, Fokin VM, Schick C, Zanotto ED. Crystallization of glassforming liquids: Maxima of nucleation, growth, and overall crystallization rates.

[98] Schwartz R, Voigt J, Tuttle B, Payne D, Reichert T, DaSalla R. Comments on the effects of solution precursor characteristics and thermal processing conditions on the crystal-

pyrolysis. Solid State Ionics. 2004;166(3):229–239.

342 Modern Technologies for Creating the Thin-film Systems and Coatings

2259–2266.

2564.

2001;20(19):1763–1766.

University Press; USA. 1990.

Crystalline Solids. 1988;100(1):51–64.

### **Modern Technologies for Creating Nanostructures in Thin‐Film Solar Cells Modern Technologies for Creating Nanostructures in Thin**‐**Film Solar Cells**

Yang Tang Yang Tang

Additional information is available at the end of the chapter Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64611

#### **Abstract**

Most photovoltaic devices (solar cells) sold in the market today are based on silicon wafers, so‐called first‐generation technology. There is an argument whether the market at present is on the verge of switching to a "second generation" of thin‐film solar cell technology. Thin‐film photovoltaic device technology relies on light management to enhance light absorption in thin absorber layers. The use of the ZnO nanorods in the thin‐film solar cells is an effective way to decrease the reflection. The variation of the geometrical parameters of the ZnO nanorods, such as the diameter, the height and the density, can lead to an optimum, which results in the maximal absorption in the absorber.

**Keywords:** thin film, solar cell, nanostructure, electrodeposition, efficiency

### **1. Introduction**

The solar energy as one of the new energy sources and a regenerated energy is abundant and pollution free. Most photovoltaic devices (solar cells) sold in the market today are based on silicon wafers, so‐called first‐generation technology. There is an argument that whether the market at present is on the verge of switching to a "second generation" of thin‐film solar cell technology. Nowadays three types of the thin‐film solar cells have realized industrialization. They are copper indium gallium selenide (CIGS) solar cells, CdTe solar cells and a‐Si solar cells. CIGS‐based thin‐film photovoltaic devices show the highest efficiency among the various thin‐film technologies, having recently reached a record value of 22.6% for the laboratory scale [1]. CdTe‐based thin‐film solar cells have got a record lab efficiency of 22.1% [2]. Some of the technologies have already entered the stage of mass production with commercial modules that provide stable efficiencies in the 13–14% range. There is still a large gap between the lab‐scale

© 2016 The Author(s). Licensee InTech. This chapter is 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. © 2017 The Author(s). Licensee InTech. This chapter is 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.

efficiency and the module efficiency indicating the great potential of the technologies. It is vitally important to further develop the innovative thin‐film technologies using environmentally friendly and sustainable approaches with lower costs and higher efficiencies.

### **2. Nanostructured thin‐film solar cells**

This section starts with a brief introduction on ZnO fundamental properties and the definition of nanostructures. The structures of the thin‐film photovoltaic devices are depicted. Then the nanorod meets the solar cell. The applications of the ZnO nanorods in thin‐film photovoltaic devices are given. It is followed by the results on the efficiency boost of the thin‐film solar cells implanted with ZnO nanorods.

### **2.1. Implantation of the nanostructures**

ZnO is a wide direct bandgap II–VI semiconductor. In last decade, the ZnO material ranging from its thin films to nanostructures has been widely investigated for their applications in various electronic and optoelectronic devices. ZnO is not really a newly discovered material. Research on ZnO has continued for many decades with interest. In terms of its characterization, reports go back to 1935 or even earlier. For example, lattice parameters of ZnO were investigated for many decades [3–7].

Most of the group II–VI binary compound semiconductors crystallize in either cubic zinc blende or hexagonal wurtzite structure where each anion is surrounded by four cations at the corners of a tetrahedron and vice versa. Three kinds of crystal structures are shared by ZnO, which are rocksalt, zinc blende and wurtzite [8]. The zinc blende ZnO structure can be stabilized only by growth on cubic substrates, and the rocksalt structure may be obtained at relatively high pressures [8].

In a variant of the cell structure the nontransparent rear metal contact can be replaced by a transparent conductive oxides (TCO) film. For conventional CIGS thin‐film solar cells, metallic Mo back electrodes are commonly used, making it impossible for light to pass through the metal electrode layer. It is possible to reverse the cell structure by starting with the deposition of the transparent contact (superstrate configuration). ZnO nanorod arrays are embedded between the TCO layer and the absorber layer serving as a buffer role. Optionally, an additional buffer layer can be inserted between the ZnO nanorods and the absorber in the superstrate solar cells. The light enters the cell through the superstrate, which has the advantage that the module can be encapsulated with nontransparent material of lower mass and lower cost. Moreover, if the other contact electrode were replaced by a TCO contact, the cell would be illuminated by both sides. In addition, the ZnO nanorod arrays have been incorporated into a superstrate or a bifacial cell structure of the other thin‐film photovoltaic devices such as dye‐ sensitized solar cells [9], quantum dye‐sensitized solar cells [10] and organic solar cells [11].

#### **2.2. Efficiency boost of the nanostructured thin‐film solar cells**

The ZnO nanorods electrodeposited on fluorine doped tin oxide (FTO) substrates have a typical bottom diameter of 220 nm and a top diameter of 120 nm. The ZnO nanorod arrays' density is 6.8 × 10<sup>8</sup> 1/cm<sup>2</sup> . The simulation of the nanostructured structure started with the ZnO nanorod arrays possessing this typical morphology and geometry. The modeling results including reflectance, transmission and absorption were weighted with the AM 1.5 solar spectrum. The optical modeling results on the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga) Se2 . The optical constants of the Cu(In,Ga)Se2 is derived from the layer. At first, The ZnO nanorod arrays' densities were varied while the other parameters such as the diameters and lengths are fixed. **Table 1** and **Figure 1** show the modeling results. The densities are varied from 4.2 × 10<sup>8</sup> to 3.2 × 109 1/cm<sup>2</sup> . Correspondingly, the percentage site coverages of the ZnO nanorods on the FTO surface are varied from 13.2 to 100%. As shown in the table, an increase in the ZnO nanorods' density leads to a considerable decrease in the reflection. The reflection is decreased from 12.07 to 5.60% by increasing the ZnO nanorods' density from 4.2 × 10<sup>8</sup> 1/cm<sup>2</sup> (site coverage 13.2%) to 3.2 × 109 1/cm<sup>2</sup> (site coverage 100%). An increase in the ZnO nanorods' density results in an increase in the transmission. With increasing the ZnO nanorods' density from 4.2 × 10<sup>8</sup> 1/cm<sup>2</sup> (site coverage 13.2%) to 3.2 × 109 1/cm<sup>2</sup> (site coverage 100%), the transmission is increased from 3.33 to 4.36%. The reason for the increase of the transmission in the range between 900 and 1200 nm is that ZnO nanorods work as a waveguide for the infrared light. As a result of the rise in the ZnO nanorods' density from 4.2 × 10<sup>8</sup> 1/cm<sup>2</sup> (site coverage 13.2%) to 3.1 × 109 1/cm<sup>2</sup> (site coverage 97.5%), the absorption in SnO2 :F descends from 12.15 to 11.09%. However, the absorption in the SnO2 :F ascends from 11.09 to 11.58% with increasing the ZnO nanorods' density from 3.1 × 109 1/cm<sup>2</sup> (site coverage 97.5%) to 3.2 × 109 1/cm<sup>2</sup> (site coverage 100.0%). An increase in the ZnO nanorods' density leads to a continuous increase of the absorption in ZnO nanorods. Owing to the decrease of the reflection and the SnO2 :F absorption with increasing the ZnO nanorods' density from 4.2 × 10<sup>8</sup> 1/cm<sup>2</sup> (site coverage 13.2%) to 3.1 × 109 1/cm<sup>2</sup> (site coverage 97.5%), the absorption in Cu(In,Ga)Se2 is boosted from 63.30 to 68.13%. An increase in the ZnO nanorods' density over 3.1 × 109 1/cm<sup>2</sup> results in a reduction in the absorption of Cu(In,Ga)Se2 . Therefore, the structure with the ZnO nanorods' density of 3.1 × 109 1/cm<sup>2</sup> (site coverage 97.5%) has the maximum absorption in Cu(In,Ga)Se2 .

efficiency and the module efficiency indicating the great potential of the technologies. It is vitally important to further develop the innovative thin‐film technologies using environmen-

This section starts with a brief introduction on ZnO fundamental properties and the definition of nanostructures. The structures of the thin‐film photovoltaic devices are depicted. Then the nanorod meets the solar cell. The applications of the ZnO nanorods in thin‐film photovoltaic devices are given. It is followed by the results on the efficiency boost of the thin‐film solar cells

ZnO is a wide direct bandgap II–VI semiconductor. In last decade, the ZnO material ranging from its thin films to nanostructures has been widely investigated for their applications in various electronic and optoelectronic devices. ZnO is not really a newly discovered material. Research on ZnO has continued for many decades with interest. In terms of its characterization, reports go back to 1935 or even earlier. For example, lattice parameters of ZnO were

Most of the group II–VI binary compound semiconductors crystallize in either cubic zinc blende or hexagonal wurtzite structure where each anion is surrounded by four cations at the corners of a tetrahedron and vice versa. Three kinds of crystal structures are shared by ZnO, which are rocksalt, zinc blende and wurtzite [8]. The zinc blende ZnO structure can be stabilized only by growth on cubic substrates, and the rocksalt structure may be obtained at relatively high pressures [8].

In a variant of the cell structure the nontransparent rear metal contact can be replaced by a transparent conductive oxides (TCO) film. For conventional CIGS thin‐film solar cells, metallic Mo back electrodes are commonly used, making it impossible for light to pass through the metal electrode layer. It is possible to reverse the cell structure by starting with the deposition of the transparent contact (superstrate configuration). ZnO nanorod arrays are embedded between the TCO layer and the absorber layer serving as a buffer role. Optionally, an additional buffer layer can be inserted between the ZnO nanorods and the absorber in the superstrate solar cells. The light enters the cell through the superstrate, which has the advantage that the module can be encapsulated with nontransparent material of lower mass and lower cost. Moreover, if the other contact electrode were replaced by a TCO contact, the cell would be illuminated by both sides. In addition, the ZnO nanorod arrays have been incorporated into a superstrate or a bifacial cell structure of the other thin‐film photovoltaic devices such as dye‐ sensitized solar cells [9], quantum dye‐sensitized solar cells [10] and organic solar cells [11].

The ZnO nanorods electrodeposited on fluorine doped tin oxide (FTO) substrates have a typical bottom diameter of 220 nm and a top diameter of 120 nm. The ZnO nanorod arrays'

. The simulation of the nanostructured structure started with the

**2.2. Efficiency boost of the nanostructured thin‐film solar cells**

1/cm<sup>2</sup>

tally friendly and sustainable approaches with lower costs and higher efficiencies.

**2. Nanostructured thin‐film solar cells**

346 Modern Technologies for Creating the Thin-film Systems and Coatings

implanted with ZnO nanorods.

**2.1. Implantation of the nanostructures**

investigated for many decades [3–7].

density is 6.8 × 10<sup>8</sup>


**Table 1.** The optical modeling results on the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga)Se2 . The simulated ZnO nanorod has a bottom diameter of 220 nm and a top diameter of 120 nm. The density of the nanorods is 3.1 × 109 1/cm<sup>2</sup> with the site coverage of 97.5%. The heights of the ZnO nanorods are varied.

**Figure 1.** The simulated optical spectra of the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga)Se2 . The simulated ZnO nanorod has a bottom diameter of 220 nm and a top diameter of 120 nm. The nanorod's height is 300 nm. The densities of the ZnO nanorod arrays are varied. Correspondingly the percentage site coverages of the ZnO nanorods on the SnO2 :F surface are varied.

Since the ZnO nanorods' density has been optimized to be 3.1 × 109 1/cm<sup>2</sup> (site coverage 97.5%), the other morphological parameters of the ZnO nanorods would be varied to achieve the maximum absorption in Cu(In,Ga)Se2 . Thus, the ZnO nanorods' heights are varied while the other parameters such as the diameters and densities are fixed. **Table 2** and **Figure 2** show the modeling results. The nanorods' heights are varied from 50 to 1000 nm. By increasing the ZnO nanorods' height from 50 to 110 nm, the reflection of the structure is decreased from 8.30 to 5.86%, which is diminished by 29.4%. An increase in the ZnO nanorods' heights from 110 to 1000 nm does not induce a significant change of the structures' reflections and the reflections vary in a small range between 5.65 and 5.93%. As shown in **Table 2**, an increase in the ZnO nanorods' height leads to a considerable increase of the structure's transmission. The transmission is boosted from 3.57 to 6.02% by increasing the ZnO nanorods' height from 50 to 1000 nm, which is enlarged by 68.6%. As shown in **Figure 2**, the increase in the transmission is in the range from 900 (1.38 eV) to 1200 nm (1.03 eV), because ZnO nanorods work as a waveguide enhancing the infrared transmission. An increase in the ZnO nanorods' height from 50 to 700 nm induces a slight decrease of the absorption in SnO2 :F from 11.75 and 10.88%, which is diminished by 7.4%. However, the absorption in the SnO2 :F ascends from 10.88 to 10.95% with increasing the ZnO nanorods' height from 700 to 1000 nm. An increase in the ZnO nanorods' height leads to a continuous increase of the absorption in ZnO nanorods. The absorption in ZnO nanorods ascends from 1.01 to 2.58% as a result of the increase in the nanorods' height from 50 to 1000 nm. Owing to the decrease of the reflection and the SnO2 :F absorption with increasing the ZnO nanorods' height from 50 to 100 nm, the absorption in Cu(In,Ga)Se2 is boosted from 66.52 to 68.65%. As a consequence of the boost in the transmission and the ZnO nanorod absorption, an increase in the ZnO nanorods' height over 100 nm results in a reduction in the absorption of Cu(In,Ga)Se2 . Therefore, the nanostructured thin‐film solar cells with the ZnO nanorods' height of 100 nm has the maximum absorption in Cu(In,Ga)Se2 .


**Table 2.** The optical modeling results on the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga)Se2 . The simulated ZnO nanorod has a top diameter of 120 nm and a length of 100 nm. The density of the ZnO nanorod arrays is 3.1 × 109 1/ cm<sup>2</sup> . The bottom diameters of the ZnO nanorod are varied. Correspondingly the percentage site coverages of the ZnO nanorods on the SnO2 :F surface are varied.

Since the ZnO nanorods' density has been optimized to be 3.1 × 109

height from 50 to 700 nm induces a slight decrease of the absorption in SnO2

and 10.88%, which is diminished by 7.4%. However, the absorption in the SnO2

from 10.88 to 10.95% with increasing the ZnO nanorods' height from 700 to 1000 nm. An

the maximum absorption in Cu(In,Ga)Se2

surface are varied.

**Figure 1.** The simulated optical spectra of the structure of glass/SnO2

348 Modern Technologies for Creating the Thin-film Systems and Coatings

97.5%), the other morphological parameters of the ZnO nanorods would be varied to achieve

nanorod has a bottom diameter of 220 nm and a top diameter of 120 nm. The nanorod's height is 300 nm. The densities of the ZnO nanorod arrays are varied. Correspondingly the percentage site coverages of the ZnO nanorods on the SnO2

the other parameters such as the diameters and densities are fixed. **Table 2** and **Figure 2** show the modeling results. The nanorods' heights are varied from 50 to 1000 nm. By increasing the ZnO nanorods' height from 50 to 110 nm, the reflection of the structure is decreased from 8.30 to 5.86%, which is diminished by 29.4%. An increase in the ZnO nanorods' heights from 110 to 1000 nm does not induce a significant change of the structures' reflections and the reflections vary in a small range between 5.65 and 5.93%. As shown in **Table 2**, an increase in the ZnO nanorods' height leads to a considerable increase of the structure's transmission. The transmission is boosted from 3.57 to 6.02% by increasing the ZnO nanorods' height from 50 to 1000 nm, which is enlarged by 68.6%. As shown in **Figure 2**, the increase in the transmission is in the range from 900 (1.38 eV) to 1200 nm (1.03 eV), because ZnO nanorods work as a waveguide enhancing the infrared transmission. An increase in the ZnO nanorods'

1/cm<sup>2</sup>

. Thus, the ZnO nanorods' heights are varied while

:F/ZnO nanorods/Cu(In,Ga)Se2

(site coverage

. The simulated ZnO

:F

:F from 11.75

:F ascends

First, the ZnO nanorods' bottom diameters are varied while the other parameters such as the top diameter, length and density are fixed. **Table 3** and **Figure 3** show the modeling results. The nanorods' bottom diameters are varied from 120 to 220 nm. Correspondingly, the percentage site coverage of the ZnO nanorods on the FTO surface is varied from 29.9 to 97.5%. As shown in the table, the ZnO nanorod possessing the same size of the bottom and top diameters has the hexagonal prism morphology and shows a high reflection of 10.31%. An increase in the ZnO nanorods' bottom diameter leads to a considerable decrease in the reflection. By increasing the ZnO nanorods' bottom diameter from 120 (site coverage 29.9%) to 220 nm (site coverage 97.5%), the reflection is decreased from 10.31 to 5.88%, which is reduced by 43.0%. With increasing the ZnO nanorods' bottom diameter from 120 (site coverage 29.9%) to 220 nm (site coverage 97.5%), the transmission is slightly increased from 3.66 to 3.83%. As a result of the rise in the ZnO nanorods' bottom diameter from 120 (site coverage 29.9%) to 220 nm (site coverage 97.5%), the absorption in SnO2 :F descends from 11.81 to 11.28%. An increase in the ZnO nanorods' bottom diameter leads to a continuous increase of the absorption in ZnO nanorods. Owing to the decrease of the reflection and the SnO2 :F absorption with increasing the ZnO nanorods' bottom diameter from 120 (site coverage 29.9%) to 220 nm (site coverage 97.5%), the absorption in Cu(In,Ga)Se2 is boosted from 64.91 to 68.65%. Therefore, the nanostructured thin‐film solar cells with the ZnO nanorods' bottom diameter of 220 nm (site coverage 97.5%) has the maximum absorption in Cu(In,Ga)Se2 .

**Figure 2.** The simulated optical spectra of the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga)Se2 . The simulated ZnO nanorod has a bottom diameter of 220 nm and a top diameter of 120 nm. The density of the nanorods is 3.1 × 109 1/cm<sup>2</sup> with the site coverage of 97.5%. The heights of the ZnO nanorod are varied.

Since the ZnO nanorods' bottom diameter has been optimized to be 220 nm, the top diameter of the ZnO nanorods would be varied to achieve the maximum absorption in Cu(In,Ga)Se2 . **Table 4** and **Figure 4** show the modeling results. The nanorods' top diameters are varied from 80 to 160 nm. With increasing the ZnO nanorods' top diameter from 80 to 150 nm, the reflection of the structure is decreased from 6.85 to 5.64%, which is diminished by 17.7%. As shown in **Table 4** an increase in the ZnO nanorods' top diameter over 150 nm results in a slight increase in the reflection. The reflection of the structure with the ZnO nanorods possessing a top diameter of 160 nm is 5.65%. By increasing the ZnO nanorods' top diameter from 80 to 160 nm, the transmission is increased from 3.67 to 3.96%. As a result of the rise in the ZnO nanorods' top diameter from 80 to 160 nm, the absorption in SnO2 :F descends from 11.48 to 11.18%. An increase in the ZnO nanorods' top diameter leads to a continuous increase of the absorption in ZnO nanorods. Owing to the decrease of the reflection and the SnO2 :F absorption with increasing the ZnO nanorods' top diameter from 80 to 140 nm, the absorption in Cu(In,Ga)Se2 is boosted from 67.84 to 68.76%. By increasing the ZnO nanorods' top diameter from 140 to 150 nm, the absorption in Cu(In,Ga) Se2 keeps constant. An increase in the ZnO nanorods' top diameter over 150 nm results in a reduction in the absorption of Cu(In,Ga)Se2 . Therefore, the nanostructured thin‐film solar cells with the ZnO nanorods' top diameter of 150 nm has the maximum absorption in Cu(In,Ga)Se2 . Compared with the structure without using the ZnO nanorods, the absorption of Cu(In,Ga)Se2 in the nanorod‐integrated structure is boosted from 62.20 to 68.76%, which is enlarged by 9.5%.


**Table 3.** The optical modeling results on the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga)Se2 . The simulated ZnO nanorod has a bottom diameter of 220 nm and a height of 100 nm. The density of the ZnO nanorod arrays is 3.1 × 109 1/ cm<sup>2</sup> . The top diameters of the ZnO nanorods are varied.

Since the ZnO nanorods' bottom diameter has been optimized to be 220 nm, the top diameter of

and **Figure 4** show the modeling results. The nanorods' top diameters are varied from 80 to 160 nm. With increasing the ZnO nanorods' top diameter from 80 to 150 nm, the reflection of the structure is decreased from 6.85 to 5.64%, which is diminished by 17.7%. As shown in **Table 4** an increase in the ZnO nanorods' top diameter over 150 nm results in a slight increase in the reflection. The reflection of the structure with the ZnO nanorods possessing a top diameter of 160 nm is 5.65%. By increasing the ZnO nanorods' top diameter from 80 to 160 nm, the transmission is increased from 3.67 to 3.96%. As a result of the rise in the ZnO nanorods' top diameter from

nanorods' top diameter leads to a continuous increase of the absorption in ZnO nanorods. Owing

By increasing the ZnO nanorods' top diameter from 140 to 150 nm, the absorption in Cu(In,Ga)

with the ZnO nanorods' top diameter of 150 nm has the maximum absorption in Cu(In,Ga)Se2

Compared with the structure without using the ZnO nanorods, the absorption of Cu(In,Ga)Se2 in the nanorod‐integrated structure is boosted from 62.20 to 68.76%, which is enlarged by 9.5%.

keeps constant. An increase in the ZnO nanorods' top diameter over 150 nm results in a

:F descends from 11.48 to 11.18%. An increase in the ZnO

:F/ZnO nanorods/Cu(In,Ga)Se2

:F absorption with increasing the ZnO nanorods'

. Therefore, the nanostructured thin‐film solar cells

is boosted from 67.84 to 68.76%.

. **Table 4**

1/cm<sup>2</sup>

. The simulated ZnO

.

the ZnO nanorods would be varied to achieve the maximum absorption in Cu(In,Ga)Se2

nanorod has a bottom diameter of 220 nm and a top diameter of 120 nm. The density of the nanorods is 3.1 × 109

80 to 160 nm, the absorption in SnO2

Se2

to the decrease of the reflection and the SnO2

**Figure 2.** The simulated optical spectra of the structure of glass/SnO2

350 Modern Technologies for Creating the Thin-film Systems and Coatings

with the site coverage of 97.5%. The heights of the ZnO nanorod are varied.

reduction in the absorption of Cu(In,Ga)Se2

top diameter from 80 to 140 nm, the absorption in Cu(In,Ga)Se2

**Figure 3.** The simulated optical spectra of the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga)Se2 . The simulated ZnO nanorod has a top diameter of 120 nm and a height of 100 nm. The density of the ZnO nanorod arrays is 3.1 × 109 1/ cm<sup>2</sup> . The bottom diameters of the ZnO nanorod are varied. Correspondingly the percentage site coverages of the ZnO nanorods on the SnO2 :F surface are varied.


**Table 4.** The optical modeling results on the structure of glass/SnO2 :F/ZnO nanorods/Cu(In,Ga)Se2 . The simulated ZnO nanorod has a bottom diameter of 220 nm and a height of 100 nm. The density of the ZnO nanorod arrays is 3.1 × 109 1/ cm<sup>2</sup> . The top diameters of the ZnO nanorods are varied.

**Figure 4.** Current versus time transients obtained on FTO substrate at potential of −1.39 V.

### **3. Growth of ZnO nanorod arrays**

**Sample ZnO nanorods'** 

cm<sup>2</sup>

**top diameter (nm)**

352 Modern Technologies for Creating the Thin-film Systems and Coatings

**Table 4.** The optical modeling results on the structure of glass/SnO2

**Figure 4.** Current versus time transients obtained on FTO substrate at potential of −1.39 V.

. The top diameters of the ZnO nanorods are varied.

**Reflection (%) Transmission (%)**

 – 13.18 3.47 12.27 – 62.20 60 6.85 3.67 11.48 1.28 67.84 80 6.48 3.73 11.41 1.34 68.16 100 6.15 3.78 11.35 1.40 68.44 120 5.88 3.83 11.28 1.48 68.65 130 5.77 3.87 11.25 1.51 68.72 140 5.69 3.90 11.22 1.55 68.76 150 5.64 3.93 11.20 1.59 68.76 160 5.65 3.96 11.18 1.62 68.71

nanorod has a bottom diameter of 220 nm and a height of 100 nm. The density of the ZnO nanorod arrays is 3.1 × 109

**Absorption in** 

**Absorption in ZnO nanorods**  **Absorption in Cu(In,Ga)Se<sup>2</sup> (%)**

. The simulated ZnO

1/

**(%)**

:F/ZnO nanorods/Cu(In,Ga)Se2

**Sn2 O3 :F (%)**

> The solution‐based fabrication routes including hydrothermal method and electrochemical deposition (ECD) method are the ways to grow ZnO nanostructures at a low temperature down to the range between 60 and 90°C [12–23]. Meanwhile the growth can be achieved over large areas up to 30 × 30 cm. A two‐stage electrodeposition method for tailoring the ZnO nanostructures is presented.

#### **3.1. Nucleation and crystal growth**

For the electrolyte recipe of ZnO nanostructures, there are two choices for the Zn source in the solution: ZnCl2 [12] or Zn(NO3 )2 [13]. In order to provide the oxygen source, one would have to bubble O2 into the electrolyte during electrodeposition if ZnCl2 were used [12]. The oxygen source is from the reduction of NO3 for the electrolyte using Zn(NO3 )2 . The formations of ZnO by NO3 precursors were described in Eqs. (1)–(4) as follows [13].

$$\text{Zn} \left( \text{NO}\_3 \right)\_2 \rightarrow \text{Zn}^{2+} + 2\text{NO}\_3^{-} \tag{1}$$

$$\rm NO\_3^- + 2e^\cdot + H\_2O \rightarrow 2OH^- + NO\_2^- \tag{2}$$

$$\text{Zn}^{2+} + 2\text{OH}^{-} \rightarrow \text{Zn(OH)}\_{2} \tag{3}$$

$$\text{Zn} \left( \text{OH} \right)\_2 \rightarrow \text{ZnO} + \text{H}\_2\text{O} \tag{4}$$

During the ECD process, the electrodeposited products are deposited on one of the electrodes. The electrode with the deposited materials is the working electrode (WE). However, the WE is not adequate for the complete ECD process. At least another electrode should be installed for allowing current to flow. In the simplest case, a two‐electrode cell is used for ECD. In order to measure the accurate WE potential, a three‐electrode cell containing a WE, a counter electrode (CE) and a reference electrode (RE) is more common. A current flows between the WE and CE, while the potential of the WE is measured against the RE. The glass substrates are used as the WE. The electrochemical cell was placed in a thermoregulated bath. The liquid electrolyte contains the salts regarding the recipes. A schematic illustration of the set up for the ECD process is shown in **Figure 5**. The electrochemical process is controlled and recorded by a potentiostat/galvanostat.

**Figure 5.** The scheme of the electrochemical deposition system.

**Figure 6** shows a typical curve for current vs transit time obtained in the electrodeposition process of the ZnO nanorods. At the beginning of the potential step, there is a rapid surge of the current. The potential difference between the FTO surface and the electrolyte leads to the accumulation of excess charges near the surface. As a result, the electric double layer is formed at the phase boundary [24]. The initial surge corresponds to the electric double layer charging at the onset of the potential step [25]. After the surge, current decays abruptly as cations are reduced and the anions are oxidized in the close vicinity of the electrode [26, 27]. The nucleation forms in this stage. As fresh cations and anions diffuse, the current began to increase. The internal hexagonal structure of ZnO favors the anisotropic growth along the c‐axis direction, which leads to the formation of the nanorods [28, 29]. During the nanorods' growth process, the surface area of the nanorods is increasing continuously. Therefore, the transient current density cannot be determined except for monitoring the in situ growth of the nanorods and collecting the information of the nanorods' morphological changes during the electrodeposition process.

**Figure 6.** Current versus time transients obtained on FTO substrate at potential of −1.39 V.

#### **3.2. Two‐stage electrodeposition**

Since the growth of ZnO nanostructures greatly depends on the nucleation process, it is possible to control the ZnO nanostructures' growth and properties by adjusting the amperage in the nucleation process. We developed a two‐stage electrodeposition method to control the electrodeposited ZnO nanostructures' morphology and geometry. The samples were prepared from an aqueous solution: 7 mM Zn(NO3 )2 .6H2 O and 7mM NH4 NO3 . The method was divided into two stages. A galvanic current (galvanic control) was applied during the first stage and then it was switched to a potentiostatic mode (−1.39V) (potentiostatic control) for the following growth stage. In order to investigate the influence of the galvanic current on the ZnO nanorods' growth and properties, we prepared one sample by using potentiostatic method (−1.39V) and the other samples by using two‐stage electrodeposition method with the galvanic current ranging from −0.2 to −2 mA. All the FTO substrates were cut into small 2.5 cm<sup>2</sup> rectangles and then cleaned in an ultrasonic bath of acetone and ethanol with subsequent rinsing in distilled water. After preparation, the samples were washed with distilled water to remove any residual salt. By using this new method, the packing density of ZnO nanorods on bare FTO can be adjusted.

### **3.3. Control of ZnO nanorod arrays' density**

**Figure 6** shows a typical curve for current vs transit time obtained in the electrodeposition process of the ZnO nanorods. At the beginning of the potential step, there is a rapid surge of the current. The potential difference between the FTO surface and the electrolyte leads to the accumulation of excess charges near the surface. As a result, the electric double layer is formed at the phase boundary [24]. The initial surge corresponds to the electric double layer charging at the onset of the potential step [25]. After the surge, current decays abruptly as cations are reduced and the anions are oxidized in the close vicinity of the electrode [26, 27]. The nucleation forms in this stage. As fresh cations and anions diffuse, the current began to increase. The internal hexagonal structure of ZnO favors the anisotropic growth along the c‐axis direction, which leads to the formation of the nanorods [28, 29]. During the nanorods' growth process, the surface area of the nanorods is increasing continuously. Therefore, the transient current density cannot be determined except for monitoring the in situ growth of the nanorods and collecting the information of the nanorods' morphological changes during the electrodeposition process.

354 Modern Technologies for Creating the Thin-film Systems and Coatings

Since the growth of ZnO nanostructures greatly depends on the nucleation process, it is possible to control the ZnO nanostructures' growth and properties by adjusting the amperage in the nucleation process. We developed a two‐stage electrodeposition method to control the electrodeposited ZnO nanostructures' morphology and geometry. The samples were pre-

**Figure 6.** Current versus time transients obtained on FTO substrate at potential of −1.39 V.

)2 .6H2

 rectangles and then cleaned in an ultrasonic bath of acetone and ethanol with subsequent rinsing in distilled water. After preparation, the samples were washed with distilled water to remove any residual salt. By using this new method, the packing density of ZnO nanorods on

divided into two stages. A galvanic current (galvanic control) was applied during the first stage and then it was switched to a potentiostatic mode (−1.39V) (potentiostatic control) for the following growth stage. In order to investigate the influence of the galvanic current on the ZnO nanorods' growth and properties, we prepared one sample by using potentiostatic method (−1.39V) and the other samples by using two‐stage electrodeposition method with the galvanic current ranging from −0.2 to −2 mA. All the FTO substrates were cut into small 2.5

O and 7mM NH4

NO3

. The method was

**3.2. Two‐stage electrodeposition**

cm<sup>2</sup>

bare FTO can be adjusted.

pared from an aqueous solution: 7 mM Zn(NO3

Since the growth of ZnO nanorods greatly depends on the nucleation process, it is possible to control the ZnO nanostructures' growth and properties by adjusting the current density in the nucleation process. We developed a new method, that is, two‐stage electrodeposition method to control the packing density and diameter of the electrodeposited ZnO nanorods. The two‐stage eletrodeposition method was divided into two stages. A galvanic current was applied during the first stage and then it was switched to a potentiostatic mode for the following growth stage. In order to investigate the influence of the galvanic current density on the ZnO nanorods' growth and properties, we prepared one sample by using potentiostatic method with a potential of −1.39V and the other samples by using galvanic current ranging from −0.2 to −2 mA/cm2 .

The average density, average diameter the average length in five ZnO nanorods samples were estimated from a statistical evaluation and are summarized in **Table 5**. The diameter and height dispersion in five ZnO nanorod samples are shown in **Figure 7**. It shows that the ZnO nanorods' density in samples prepared by using two‐stage eletrodeposition method with a high galvanic current density (−0.5 to −2 mA/cm2 ) is larger than that of sample prepared under a low galvanic current density (−0.2 mA/cm2 ). Therefore, it is clear that an increase in galvanic current density leads to a considerable increase of the ZnO nanorods' density. By increasing the galvanic current density from −0.2 to −2 mA/cm2 , the density of ZnO nanorods was boosted from 4.2 × 10<sup>8</sup> to 1.3 × 109 1/cm<sup>2</sup> , which is enlarged by ∼200%. According to the formations of ZnO by NO3 precursors described in Eqs. (1)–(4), an increase in a current raises the yields of OHˉ ions. As a result, the nucleation sites were boosted in the initial nucleation process. The mechanism is illustrated in **Figure 8**. The ZnO nanorods' density was increased in the following process under the potentiostatic mode.


**Table 5.** Growth parameters of two stages, average diameter, average height and density of the ZnO nanorods.

In order to investigate the nucleation process at the galvanic stage, four samples were prepared under a galvanic mode with the galvanic current density ranging from −0.2 to −2 mA/cm2 , and the process was stopped before the following potentiostatic stage. The densities of the nucleation sites in these samples were shown in **Table 6**. The size of nuclei sites in the sample prepared under −0.2 mA/cm2 was too small to be observed in the measurements. By increasing the galvanic

**Figure 7.** The diameter and height dispersion of the ZnO nanorods prepared by using two‐stage electrodeposition method with the galvanic current density of 0 mA/cm2 (a), (b); −0.2 mA/cm2 (c), (d); −0.5 mA/cm2 (e), (f); −1.2 mA/cm2 (g), (h) and −2 mA/cm2 (i), (j).

**Figure 8.** The sketch of the ZnO nanorods' nucleation mechanism under different galvanic current densities.

current density, the density of the nucleation sites was boosted from 1.5 × 109 to 2.4 × 109 1/cm<sup>2</sup> . These nucleation sites assisted the following growth of ZnO nanorods with a higher density.

An increase in the galvanic current density, on the other hand, induced a significant decrease of the ZnO nanorods' average diameter. The average diameter of ZnO nanorods was reduced from 317 to 74 nm as the galvanic current density was increased from −0.2 to −2 mA/cm2 . The size of ZnO nanorods' diameter depended on the nucleation sites at the initial stage. Since the yields of OHˉ ions is higher than Zn2+ ions in the initial process under a higher galvanic current density, the fact that the rate of OHˉ generation is larger than the diffusion rate of Zn2+ ions to the cathode suppresses the lateral growth of the nucleation sites. As a consequence, the size of the nanorods' diameter was decreased by using a higher galvanic current density. The galvanic current did not lead to a significant variation of the ZnO nanorods' average length and the ZnO nanorods keep a virtually constant average length around 620 nm.


**Table 6.** Growth parameters and density of the samples prepared under a galvanostatic mode.

### **4. Conclusion**

**Figure 7.** The diameter and height dispersion of the ZnO nanorods prepared by using two‐stage electrodeposition

**Figure 8.** The sketch of the ZnO nanorods' nucleation mechanism under different galvanic current densities.

(a), (b); −0.2 mA/cm2

(c), (d); −0.5 mA/cm2

(e), (f); −1.2 mA/cm2

method with the galvanic current density of 0 mA/cm2

356 Modern Technologies for Creating the Thin-film Systems and Coatings

(i), (j).

(g), (h) and −2 mA/cm2

Thin‐film photovoltaic device technology relies on light management to enhance light absorption in thin absorber layers. The use of the ZnO nanorods in the thin‐film solar cells is an effective way to decrease the reflection. The variation of the geometrical parameters of the ZnO nanorods, such as the diameter, the height and the density can lead to an optimum, which results in the maximal absorption in the absorber. An approach of a rigorous modeling is presented to simulate and optimize the light absorption in the Cu(In,Ga)Se2 absorbers with ZnO nanorods.

ZnO nanorod arrays were fabricated by using an electrochemical deposition method. A two‐ stage electrodeposition method for adjusting the ZnO nanorods' density on bare SnO2 :F substrates has been described and analyzed. The ZnO nanorods' density arrays were adjusted from 4.2 × 10<sup>8</sup> to 1.3 × 109 1/cm<sup>2</sup> by increasing the galvanic current density from −0.2 to −2 mA/ cm<sup>2</sup> . An increase in the galvanic current density induced a significant decrease of the ZnO nanorods' average diameter.

In this chapter, the modeling results show that the use of the ZnO nanorod arrays lead to an increase of the absorption in the absorber layer. After that, the growth of the ZnO nanorod arrays is tailored towards the optimized shape and geometry of the nanorod arrays. The opening questions and further research suggestions are outlined as follows.

The nanorods in the simulation are vertically aligned on the substrate. However, the real ZnO nanorods grown on FTO substrates are randomly oriented. It is suggested to run simulation on a tilted nanorod, which reflects the average orientation situation of the nanorod arrays. It is suggested to develop more effective ways to tailor the morphology and geometry of the ZnO nanorod arrays. Besides the use of the nanorods in the traditional thin‐film solar cells, there are other possible methods for improving the light management in the thin‐film solar cells. For example, some metal nanoparticles such as silver and gold nanodots can be incorporated into the solar cells. Due to the surface plasmonic effects of the nanoparticles, light will be efficiently coupled into the solar cells. In addition, a nanostructured antireflective coating layer can be coated on the surface of the cover glass in the thin‐film solar modules. The antireflective coating layer will decrease the reflection between the glass and the air resulting in the boost of the light harvesting in the internal solar cells. The modern technology of the light management of the thin‐film solar cells is now developing rapidly and the combination of the nanostructures and thin films will show the power.

### **Acknowledgements**

This work was supported by the National Natural Science Foundation of China under Grant No. 61404007 and the Beijing Talents Fund (2015000021223ZK38).

### **Author details**

Yang Tang

Address all correspondence to: tangy118@hotmail.com

National Institute of Clean‐and‐Low‐Carbon Energy, Future Science & Technology City, Beijing, People's Republic of China

### **References**


[7] R. R. Reeber, J. Appl. Phys. 41, 5063, 1970.

The nanorods in the simulation are vertically aligned on the substrate. However, the real ZnO nanorods grown on FTO substrates are randomly oriented. It is suggested to run simulation on a tilted nanorod, which reflects the average orientation situation of the nanorod arrays. It is suggested to develop more effective ways to tailor the morphology and geometry of the ZnO nanorod arrays. Besides the use of the nanorods in the traditional thin‐film solar cells, there are other possible methods for improving the light management in the thin‐film solar cells. For example, some metal nanoparticles such as silver and gold nanodots can be incorporated into the solar cells. Due to the surface plasmonic effects of the nanoparticles, light will be efficiently coupled into the solar cells. In addition, a nanostructured antireflective coating layer can be coated on the surface of the cover glass in the thin‐film solar modules. The antireflective coating layer will decrease the reflection between the glass and the air resulting in the boost of the light harvesting in the internal solar cells. The modern technology of the light management of the thin‐film solar cells is now developing rapidly and the combination of the

This work was supported by the National Natural Science Foundation of China under Grant

National Institute of Clean‐and‐Low‐Carbon Energy, Future Science & Technology City,

nanostructures and thin films will show the power.

358 Modern Technologies for Creating the Thin-film Systems and Coatings

No. 61404007 and the Beijing Talents Fund (2015000021223ZK38).

Address all correspondence to: tangy118@hotmail.com

Beijing, People's Republic of China

[1] ZSW, Press release, June 15, 2016.

[2] First Solar, Press release, February 23, 2016.

[5] T. J. Gray, J. Am. Ceram. Soc. 37, 534, 1954.

[3] C. W. Bunn, Proc. Phys. Soc. London 47, 835, 1935.

[4] R. B. Heller, J. McGannon, and A. H. Weber, J. Appl. Phys. 21, 1283, 1950.

[6] G. P. Mohatny and L. V. Azaroff, J. Chem. Phys. 35, 1268, 1961.

**Acknowledgements**

**Author details**

Yang Tang

**References**


### **Close‐Spaced Sublimation (CSS): A Low‐Cost, High‐ Yield Deposition System for Cadmium Telluride (CdTe) Thin Film Solar Cells Close**‐**Spaced Sublimation (CSS): A Low**‐**Cost, High**‐**Yield Deposition System for Cadmium Telluride (CdTe) Thin Film Solar Cells**

Nowshad Amin and Kazi Sajedur Rahman Nowshad Amin and Kazi Sajedur Rahman

Additional information is available at the end of the chapter Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/66040

#### **Abstract**

Semiconductors are the key materials in many of our modern day devices, such as sensors, integrated circuits, energy harvesting devices, optoelectronics and so on. However, apart from two known elemental semiconductors that are silicon and germanium, we have been using many of the synthesized ones since the microelectronic revolution known as invention of transistor. Numerous compound semiconductors since then have been synthesized, grown, deposited or simply fabricated by numerous processes in the scientific community. To avoid associated chemical disposals or keep safe from toxic or combustible gas usages in any semiconductor fabrication facilities, many researchers choose physical vapor deposition as the simplest method. One of such processes is called Close-Spaced Sublimation (CSS), which is a kind of thermal evaporation by nature. This chapter would give a comprehensive outline on CSS as one of the most advantageous semiconductor deposition processes for many compound semiconductors having relatively low evaporation temperature. Cadmium telluride (CdTe) is one of the examples utilized for solar cell absorber materials since the early 1980s using CSS technique. Therefore, growth of CdTe thin films by CSS and its utilization in thin film solar cells will be discussed to comprehend the ultimate benefits of the close-spaced sublimation (CSS) process.

**Keywords:** semiconductors, thin films, close-spaced sublimation, temperature profile, CdTe thin film solar cells

© 2017 The Author(s). Licensee InTech. This chapter is 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. © 2017 The Author(s). Licensee InTech. This chapter is 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. Introduction**

Close-spaced sublimation (CSS) is one of the simplest methods in physical vapor deposition. Materials especially semiconductors that evaporate below 800°C can be coated on substrates like glass in both vacuum and atmospheric pressure. The target materials have to be in the form of solid in either chunk or powder form [1–3]. As for example, compound semiconductor like cadmium telluride (CdTe) can be deposited at around 600*°*C with a thickness of 1–10 μm within 10 min of deposition time, which is one of the fastest deposition times among other physical vapor deposition (PVD) methods. Needless to mention, CSS and binary compound material cadmium telluride (CdTe) are densely interrelated due to extensive usage of CSS in the growth of CdTe thin film [4, 5]. The binary compound CdTe has been recognized as one of the promising thin film photovoltaic materials owing to its near optimum bandgap of 1.44 eV and high absorption coefficient over 105 /cm. CdTe, therefore, absorbs over 90% of available photons (hν > 1.44 eV) in 1 μm thickness, and hence, films of only 1–3 μm are sufficient for solar cells [6, 7]. Several types of CdTe solar cells such as Schottky barrier, homojunction, heterojunction, and p-i-n have been investigated to date [8– 10]. Among all, the most successful configurations are the heterojunctions where a wide bandgap semiconductor can be used as the heterojunction partner or "window." However, cadmium sulfide (CdS) has been the most widely studied and most appropriate window material for CdTe solar cells to date. Most recent development in CdTe thin film solar cells has found noteworthy improvements in small area conversion efficiencies. A number of techniques such as atomic layer epitaxy, spraying, electrodeposition (ED), and close-spaced sublimation (CSS) have been employed for the fabrication of CdTe thin film solar cells with significant efficiencies of over 20% with various configurations. It is quite notable that even though important dissimilarities exist, the performances achieved are independent to processing demonstrating the versatility of CdTe and its superior status in the photovoltaic technologies.

The deposition method for CdTe thin films differs widely and can considerably affect the material properties and device performance. Since CdTe has high absorption coefficient hence thicknesses for CdTe thin films are limited within 2–10 μm [11, 12]. There are various methods to deposit CdTe, which includes close-spaced sublimation (CSS), vapor transport deposition (VTD), electrodeposition (ED), physical vapor deposition (PVD), sputtering, etc. [13–15]. Among all the deposition methods, the highest efficient CdTe thin film solar cell was obtained by close-spaced sublimation (CSS). The substrate temperature is one of the crucial parameters for CdTe deposition as it could be observed that most of the deposition techniques demonstrated has substrate heating. Higher growth temperature not only enhances the deposition rate, but it also determines the quality of junction formation. Moreover, some research studies illustrated that CdTe deposited at higher temperature exhibits better performance. Therefore, the resistivity of the CdTe films decreases with an increase in substrate temperature.
