**3. Ni/PEN films used as electrodes in QC devices**

#### **3.1 Cross-sectional TEM images of Ni/PEN films**

572 Recent Advances in Nanofabrication Techniques and Applications

Ni/Ni QC devices are fabricated, two Ni thin films are directly contacted with their edges crossed, as shown in Fig. 2(a). For the fabrication of Ni/NiO/Ni QC devices, after two sets of polished PMMA/Ni/PEN/PMMA structures were prepared, the Ni edge in one of the two samples was oxidized by O2 plasma at a power of 5 W (= 5 mA and 1 kV) and then the oxidized and the unoxidized Ni edges were attached together with their edges crossed, as shown in Fig. 2(b). Also, for the fabrication of Ni/P3HT:PCBM/Ni QC devices, after two sets of polished PMMA/Ni/PEN/PMMA structures were prepared, P3HT:PCBM organic molecule blend was sandwiched between two sets of PMMA/Ni/PEN/PMMA structures whose edges were crossed, as shown in Fig. 2(c). P3HT and PCBM organic molecules were separately dissolved in monochlorobenzene, then blended together with a weight ratio of 1:1 to form a 20 mg/ml solution. This P3HT:PCBM solution was dripped on one of the polished PMMA/Ni/PEN/PMMA structure, then sandwiched by the other of the polished PMMA/Ni/PEN/PMMA structure. These three types of QC devices are shown in Fig. 2.

Fig. 2. Fabrication process of (a) Ni/Ni QC devices, (b) Ni/NiO/Ni QC devices and (c)

The Ni thickness was measured by a mechanical method using the stylus surface profiler DEKTAK and an optical method using the diode pumped solid state (DPSS) green laser at a wavelength of 532 nm and the photodiode detector. The surface morphologies of Ni/PEN samples were analyzed by atomic force microscope (AFM) Nanoscope IIIa. The microstructures as well as the Ni/PEN interfacial structures were examined using a JEOL JEM-3000F transmission electron microscope (TEM) operating at 300 kV. The cross-sectional TEM samples were prepared by a combination of mechanical polishing and Ar ion thinning. To reduce the beam-heating effects during ion thinning, the sample stage was cooled to -160

Ni/P3HT:PCBM/Ni QC devices.

**2.2 Evaluation methods of Ni/PEN films and QC devices** 

Figs. 3(a) and (b) show the cross-sectional TEM images for Ni (20 nm)/PEN films. It can be seen that there is no diffusion of Ni into the PEN layer, resulting in clear and smooth formation of the Ni/PEN interface. Here, it should be noted that some researchers have reported that metal atoms diffuse into organic layers in the process of the metal evaporation onto organic layers (Tarlov, 1992; Hirose et al., 1996; Ito et al., 1999; Dürr et al., 2002). For example, the metastable atom electron spectroscopy (MAES) spectra of Au on the p-sexiphenyl (6P)/Au system shows that the features of 6P remain even though Au was deposited to about 20 nm thickness (Ito et al., 1999). This indicates that Au atoms or clusters penetrate into the 6P films. The soft x-ray photoemission spectroscopy (SXPS) investigation of the interface between evaporated indium and perylenetetracarboxylic dianhydride (PTCDA) also demonstrates that the interfacial region is very wide, ranging from 7 to 60 nm, and this means that the metal atoms of indium diffuse into PTCDA organic layers (Hirose et al., 1996). Moreover, according to studies on the interaction between evaporated Ag and octadecanethiol (ODT) on Au films using XPS, Ag deposited at 300 K migrates through the ODT layer and resides at the ODT/Au interface (Tarlov, 1992). As compared with above results, such a metal diffusion into organic layers does not occur in Ni/PEN interface. This indicates that Ni thin films on PEN organic substrates are suitable for metal/organic films used in QC devices. It can also be confirmed that the surface of Ni films is smooth, and this smoothness is in good agreement with the results of the AFM observation, where the surface roughness *R*a is 1.1 nm. Fig. 3(c) shows the electron diffraction (ED) pattern for the same specimen. Ni thin films on PEN films have been shown to be face-centered-cubic structures, which are equal to those in bulk Ni structures. The ED pattern also shows that Ni thin films have polycrystalline structures, which can be recognized from the cross-sectional TEM image of Fig. 3(a). Thus, Ni /PEN films are suitable for QC devices from the viewpoint of the Ni/PEN interfacial and internal structures.

Fig. 3. (a) Cross-sectional TEM image, (b) high-resolution cross-sectional TEM image and (c) ED pattern for Ni (20 nm)/PEN films.

Lithography-Free Nanostructure Fabrication Techniques Utilizing Thin-Film Edges 575

in a junction resistance. These experimental results indicate that Ni/PEN films are suitable for

Fig. 4. 3-D surface images obtained from AFM observation for (a) PEN, (b) Ni (16 nm)/PEN

Fig. 5. Surface roughness as a function of the metal film thickness for Ni/PEN and Au/PEN.

Fig. 6. Scaling properties of the surface roughness *R*a for PEN and Ni/PEN. The inset

represents the scaling properties of the RMS surface roughness *R*q.

QC devices from the viewpoint of the surface, interface and internal structures.

and (c) Au (14 nm)/PEN.

#### **3.2 AFM surface morphology of Ni/PEN films**

Fig. 4 shows the three-dimensional (3-D) surface images obtained from AFM observation for (a) PEN, (b) Ni (16 nm) /PEN and (c) Au (14 nm)/PEN. From the 3-D images, which are 500×500 nm2 in area, mound-like surfaces are observed in Ni (16 nm)/PEN and Au (14 nm)/PEN, and are classified by the surface roughness *R*a. Here, the surface roughness *R*a is defined by

$$R\_a = \frac{1}{L\_x L\_y} \int\_0^{L\_x} \int\_0^{L\_y} |h(\mathbf{x}, y)| \, d\mathbf{x} dy \,\tag{1}$$

where *h* (*x*,*y*) is the height profile as a function of *x* and *y* and *Lx*(*<sup>y</sup>*) is the lateral scanning size in the *x* (*y*) direction. *R*a of PEN is 1.3 nm, which is smaller than that of widely-used organic films, such as polyethylene terephthalate (PET) and polyimide. *R*a of Ni (16 nm)/PEN is also as small as 1.22 nm. In contrast, *R*a of Au (14 nm)/PEN is as large as 2.53 nm. Fig. 5 shows the surface roughness as a function of the metal film thickness for Ni/PEN and Au/PEN. *R*a increases up to 3.8 nm for a film thickness of 21 nm for Au/PEN. In comparison, *R*a decreases slightly down to 1.1 nm with increasing the film thickness for Ni/PEN. Here, we consider the growth mode of Ni/PEN, and discuss their feasibility in QC devices from the viewpoint of the surface roughness. Fig. 6 shows the scaling properties of *R*a for PEN and Ni/PEN. The inset represents the scaling properties of the root mean square (RMS) surface roughness *R*q. *R*q obeys a scaling law, *R*q = *w*(*L*) *L* , where *w*(*L*) is the interface width corresponding to the standard deviation of the surface height, *L* is the system size and is the growth scaling exponent. The growth scaling exponent for roughening, *w*(*L*) *L* , has been widely used to characterize the growth of a solid from a vapor, such as the epitaxial growth of Fe/Si (111) (Chevrier et al., 1991), growth of evaporated Ag/quartz (Palasantzas et al., 1994) and molecular beam epitaxial growth of CuCl/CaF2 (111) (Tong et al., 1994), as described by the Kardar-Parisi-Zhang (KPZ) equation (Kardar et al., 1986). As for PEN and Ni/PEN, shows the almost constant value of 0.67-0.68, as seen from the similar roughness slope in any sample. This indicates that the surface morphology of Ni/PEN exhibits almost the same behaviour as that of PEN and these results are consistent with the 3-D AFM observation in Fig. 4. We have also found that the surface is described as self-affine due to ≠ , where is the dynamical exponent in a scaling law, *R*q = *w*(*L*) <sup>h</sup> *t* . Here, *t*h is a growth thickness. As one can see from Fig. 5, is the negative value since the surface roughness slightly decreases with increasing the thickness for Ni/PEN. This results in ≠ , which shows the self-affine growth and it can also be seen in sputtered copper films (Ohkawa et al., 2002) and evaporated silver films on silicon substrates (Thompson et al., 2004). The growth process itself of Ni thin films on PEN organic substrates is of great interest and is rich in physics, so detailed work including the dynamic physical mechanism, such as the random deposition and ballistic deposition, will be reported elsewhere. Here, we consider their feasibility in QC devices from the viewpoint of the surface roughness. Since the junction area in QC devices is determined by the film thickness, we need to clarify the surface roughness in the same scanning scale as the thickness size. As shown in Fig. 6, *R*a's of Ni (16 nm)/PEN and PEN are 0.34 nm and 0.44 nm, respectively, which correspond to 2-3 atomic layers, in the scanning scale of 16 nm. This result suggests that the number of molecules sandwiched between two metal thin films in QC devices can be strictly determined in a high resolution of 2-3 atoms by controlling the thickness of Ni thin films and it leads to a high product yield of memory devices and switching devices due to the reduction of the fluctuation

Fig. 4 shows the three-dimensional (3-D) surface images obtained from AFM observation for (a) PEN, (b) Ni (16 nm) /PEN and (c) Au (14 nm)/PEN. From the 3-D images, which are 500×500 nm2 in area, mound-like surfaces are observed in Ni (16 nm)/PEN and Au (14 nm)/PEN, and are classified by the surface roughness *R*a. Here, the surface roughness *R*a is

0 0

where *h* (*x*,*y*) is the height profile as a function of *x* and *y* and *Lx*(*<sup>y</sup>*) is the lateral scanning size in the *x* (*y*) direction. *R*a of PEN is 1.3 nm, which is smaller than that of widely-used organic films, such as polyethylene terephthalate (PET) and polyimide. *R*a of Ni (16 nm)/PEN is also as small as 1.22 nm. In contrast, *R*a of Au (14 nm)/PEN is as large as 2.53 nm. Fig. 5 shows the surface roughness as a function of the metal film thickness for Ni/PEN and Au/PEN. *R*a increases up to 3.8 nm for a film thickness of 21 nm for Au/PEN. In comparison, *R*a decreases slightly down to 1.1 nm with increasing the film thickness for Ni/PEN. Here, we consider the growth mode of Ni/PEN, and discuss their feasibility in QC devices from the viewpoint of the surface roughness. Fig. 6 shows the scaling properties of *R*a for PEN and Ni/PEN. The inset represents the scaling properties of the root mean square (RMS) surface roughness *R*q. *R*q obeys a scaling

of a solid from a vapor, such as the epitaxial growth of Fe/Si (111) (Chevrier et al., 1991), growth of evaporated Ag/quartz (Palasantzas et al., 1994) and molecular beam epitaxial growth of CuCl/CaF2 (111) (Tong et al., 1994), as described by the Kardar-Parisi-Zhang (KPZ)

0.67-0.68, as seen from the similar roughness slope in any sample. This indicates that the surface morphology of Ni/PEN exhibits almost the same behaviour as that of PEN and these results are consistent with the 3-D AFM observation in Fig. 4. We have also found that the

*t* . Here, *t*h is a growth thickness. As one can see from Fig. 5,

value since the surface roughness slightly decreases with increasing the thickness for Ni/PEN.

copper films (Ohkawa et al., 2002) and evaporated silver films on silicon substrates (Thompson et al., 2004). The growth process itself of Ni thin films on PEN organic substrates is of great interest and is rich in physics, so detailed work including the dynamic physical mechanism, such as the random deposition and ballistic deposition, will be reported elsewhere. Here, we consider their feasibility in QC devices from the viewpoint of the surface roughness. Since the junction area in QC devices is determined by the film thickness, we need to clarify the surface roughness in the same scanning scale as the thickness size. As shown in Fig. 6, *R*a's of Ni (16 nm)/PEN and PEN are 0.34 nm and 0.44 nm, respectively, which correspond to 2-3 atomic layers, in the scanning scale of 16 nm. This result suggests that the number of molecules sandwiched between two metal thin films in QC devices can be strictly determined in a high resolution of 2-3 atoms by controlling the thickness of Ni thin films and it leads to a high product yield of memory devices and switching devices due to the reduction of the fluctuation

 ≠ , where

*x y*

*a*

<sup>1</sup> | ( , )| , *L L x y*

, where *w*(*L*) is the interface width corresponding to the standard deviation

, which shows the self-affine growth and it can also be seen in sputtered

is the growth scaling exponent. The growth

shows the almost constant value of

is the dynamical exponent in a scaling

is the negative

, has been widely used to characterize the growth

*<sup>R</sup> h x y dxdy L L* (1)

**3.2 AFM surface morphology of Ni/PEN films** 

defined by

law, *R*q = *w*(*L*) *L*

law, *R*q = *w*(*L*) <sup>h</sup>

This results in

of the surface height, *L* is the system size and

equation (Kardar et al., 1986). As for PEN and Ni/PEN,

scaling exponent for roughening, *w*(*L*) *L*

surface is described as self-affine due to

≠  in a junction resistance. These experimental results indicate that Ni/PEN films are suitable for QC devices from the viewpoint of the surface, interface and internal structures.

Fig. 4. 3-D surface images obtained from AFM observation for (a) PEN, (b) Ni (16 nm)/PEN and (c) Au (14 nm)/PEN.

Fig. 5. Surface roughness as a function of the metal film thickness for Ni/PEN and Au/PEN.

Fig. 6. Scaling properties of the surface roughness *R*a for PEN and Ni/PEN. The inset represents the scaling properties of the RMS surface roughness *R*q.

Lithography-Free Nanostructure Fabrication Techniques Utilizing Thin-Film Edges 577

Fig. 7. Ni thickness dependence of the electric resistivity for Ni thin films on PEN substrates.

Fig. 8. (a) Ni electrode resistance as a function of the linewidth *l*, which corresponds to the Ni thickness *d*, in QC devices. Schematic illustration of (b) QC devices and (c) conventional

cross-bar structures.

#### **3.3 Electric resistivity of Ni/PEN films**

Fig. 7 shows the Ni thickness dependence of the electric resistivity for Ni thin films on PEN substrates. The electric resistivity Ni increases with decreasing the Ni thickness *d*. In order to explain this experimental result quantitatively, we have calculated the electric resistivity using Mayadas-Shatzkes model (Mayadas et al., 1970). According to Mayadas-Shatzkes model, the electric resistivity Ni is expressed by

$$\rho\_{\rm Ni} \;/\ \rho\_0 = \left[1 - \frac{3}{2}a + 3a^2 - 3a^3 \ln(1 + \frac{1}{a})\right]^{-1} \,. \tag{2}$$

$$\alpha = \frac{\mathcal{A}}{D} \frac{R\_{\mathcal{g}}}{1 - R\_{\mathcal{g}}},$$

where is the electron mean free path, *D* is the average grain diameter, *R*g is the reflection coefficient for electrons striking the grain boundary and 0 is the electric resistivity for bulk Ni. The electron mean free path is 11 nm for bulk Ni. The average grain diameter *D* is 3 nm, which has been obtained from the high-resolution TEM image and the ED pattern. The reflection coefficient *R*g is 0.71-0.95, which is the extrapolation value obtained from *R*g in Ni thin films with the thickness of 31-115 nm (Nacereddine et al., 2007). From Fig. 7, the experimental result shows good agreement with the calculation result quantitatively. This means that the main contribution to the electric resistivity comes from the electron scattering at grain boundaries in Ni thin films on PEN substrates. Here, we discuss the use of Ni thin films on PEN substrates for electrodes of QC devices. As can be seen from Fig. 7, the electric resistivity of Ni thin films on PEN substrates is 1-2 orders larger than that of bulk Ni. This large resistivity could produce high-resistance electrodes in QC devices. However, as stated in the introduction of this chapter, the electrode resistance can be reduced down since the film width can be controlled to the one as long as ~mm. Fig. 8(a) shows the Ni electrode resistance as a function of the linewidth *l*, which corresponds to the Ni thickness *d*, in QC devices. The schematic illustration of QC devices is shown in Fig. 8(b). In Fig. 8(a), Ni electrode resistances in the conventional cross-bar structures are also shown. The black solid line, dashed line and dotted line represent Ni electrode resistances estimated in conventional cross-bar structures with aspect ratios of 1:1, 3:1 and 5:1, respectively, where the resistivity is assumed to be the reported value in Ni thin films on glass substrates (Vries, 1987). The schematic illustration of conventional cross-bar structures is shown in Fig. 8(c). From Fig. 8(a), Ni electrode resistances in the conventional cross-bar structures are larger than 0.74, 1.2 and 3.7 M in aspect ratios of 5:1, 3:1 and 1:1, respectively, for a linewidth of less than 20 nm. In contrast, QC devices show electrode resistances as small as 0.3-2 kfor a linewidth of 10-20 nm. This resistance reduction in Ni electrodes makes it possible to detect the resistance of sandwiched materials between two edges of Ni thin films very strictly and precisely and also this result indicates that QC devices have potential application in high-frequency devices. Thus, Ni/PEN films are suited for electrodes in QC devices from the viewpoint of the electrical properties as well as the surface, interface and internal structures.

Fig. 7 shows the Ni thickness dependence of the electric resistivity for Ni thin films on PEN

to explain this experimental result quantitatively, we have calculated the electric resistivity using Mayadas-Shatzkes model (Mayadas et al., 1970). According to Mayadas-Shatzkes

2 3

 

3 1 / 1 3 3 ln(1 ) , <sup>2</sup>

, <sup>1</sup> *g g*

is the electron mean free path, *D* is the average grain diameter, *R*g is the reflection

*R D R* 

is 3 nm, which has been obtained from the high-resolution TEM image and the ED pattern. The reflection coefficient *R*g is 0.71-0.95, which is the extrapolation value obtained from *R*g in Ni thin films with the thickness of 31-115 nm (Nacereddine et al., 2007). From Fig. 7, the experimental result shows good agreement with the calculation result quantitatively. This means that the main contribution to the electric resistivity comes from the electron scattering at grain boundaries in Ni thin films on PEN substrates. Here, we discuss the use of Ni thin films on PEN substrates for electrodes of QC devices. As can be seen from Fig. 7, the electric resistivity of Ni thin films on PEN substrates is 1-2 orders larger than that of bulk Ni. This large resistivity could produce high-resistance electrodes in QC devices. However, as stated in the introduction of this chapter, the electrode resistance can be reduced down since the film width can be controlled to the one as long as ~mm. Fig. 8(a) shows the Ni electrode resistance as a function of the linewidth *l*, which corresponds to the Ni thickness *d*, in QC devices. The schematic illustration of QC devices is shown in Fig. 8(b). In Fig. 8(a), Ni electrode resistances in the conventional cross-bar structures are also shown. The black solid line, dashed line and dotted line represent Ni electrode resistances estimated in conventional cross-bar structures with aspect ratios of 1:1, 3:1 and 5:1, respectively, where the resistivity is assumed to be the reported value in Ni thin films on glass substrates (Vries, 1987). The schematic illustration of conventional cross-bar structures is shown in Fig. 8(c). From Fig. 8(a), Ni electrode resistances in the conventional cross-bar structures are larger than 0.74, 1.2 and 3.7 M in aspect ratios of 5:1, 3:1 and 1:1, respectively, for a linewidth of less than 20 nm. In contrast, QC devices show electrode resistances as small as 0.3-2 kfor a linewidth of 10-20 nm. This resistance reduction in Ni electrodes makes it possible to detect the resistance of sandwiched materials between two edges of Ni thin films very strictly and precisely and also this result indicates that QC devices have potential application in high-frequency devices. Thus, Ni/PEN films are suited for electrodes in QC devices from the viewpoint of the

Ni increases with decreasing the Ni thickness *d*. In order

1

(3)

is 11 nm for bulk Ni. The average grain diameter *D*

0 is the electric resistivity for

(2)

Ni is expressed by

 

electrical properties as well as the surface, interface and internal structures.

coefficient for electrons striking the grain boundary and

0

Ni

 

**3.3 Electric resistivity of Ni/PEN films** 

substrates. The electric resistivity

bulk Ni. The electron mean free path

model, the electric resistivity

where 

Fig. 7. Ni thickness dependence of the electric resistivity for Ni thin films on PEN substrates.

Fig. 8. (a) Ni electrode resistance as a function of the linewidth *l*, which corresponds to the Ni thickness *d*, in QC devices. Schematic illustration of (b) QC devices and (c) conventional cross-bar structures.

Lithography-Free Nanostructure Fabrication Techniques Utilizing Thin-Film Edges 579

<sup>0</sup> <sup>20000</sup> <sup>40000</sup> <sup>60000</sup> <sup>80000</sup> <sup>0</sup>

In this section, we present current density-voltage (*J-V*) characteristics in Ni/NiO/Ni QC devices, which consist of NiO tunnel barriers sandwiched between two Ni thin films whose edges are crossed. First, we introduce the background of sub-micrometer scale tunnel junctions and the motivation for fabricating QC devices with tunnel barriers. Then, we show the derivation of a formula for *J-V* characteristics of QC devices with tunnel barriers and

Sub-micrometer scale tunnel junctions have attracted much interest due to their potential application in magnetic random access memories (MRAMs), fast detectors of terahertz (THz) and IR radiation and superconducting quantum interference devices (SQUIDs). Magnetic tunnel junctions (MTJs), which consist of two ferromagnetic metals separated by thin insulators, can be expected as ultrahigh-density MRAM devices because of the giant TMR effect at room temperature (Miyazaki et al., 1995; Moodera et al., 1995; Yuasa et al., 2004; Parkin et al., 2004). CoFeB/MgO/CoFeB MTJs with a small junction area of 0.02 m2 exhibit a large TMR ratio of 98 %, where a clear current-induced magnetization switching (CIMS) with a low switching current density of 3.6 MA/cm2 have been observed (Hayakawa et al., 2008). Antenna-coupled tunnel junction devices, which consist of metal/insulator/metal tunnel junctions coupled to a thin-film metal antenna, can also be expected as fast detectors of THz and IR radiation (Sanchez et al., 1978; Kale, 1985; Hobbs et al., 2005). Ni/NiO/Ni tunnel junctions with a junction area of 0.16m2 coupled to thin-film metal antennas can serve as IR detectors and frequency mixers in the 10 m band (Wilke et al., 1994; Fumeaux et al., 1996). Moreover, much effort has been devoted to the development of sub-micrometer scale SQUIDs, which are very promising devices with a high magnetic flux sensitivity (Rugar et al., 2004; Troeman et al., 2007; Huber et al., 2008; Kirtley et al., 2009). Aluminum SQUIDs with an effective area of 0.034m2 display a high flux sensitivity

Fig. 10. Aging properties for Ni/Ni QC devices with a junction area of 17×17 nm2.

finally demonstrate experimental results of their *J-V* characteristics.

**4.2.1 Background of sub-micrometer scale tunnel junctions** 

Time (s)

0.1 0.2 0.3

0.5

0.7

*I* = 1.0 A

**4.2 J-V characteristics of Ni/NiO/Ni QC devices** 

Voltage (V)
