Effect of Sub 1-nm Pt Nanoparticle on the ... - ECS Transactions

ECS Transactions, 61 (39) 1-11 (2014) 10.1149/06139.0001ecst ©The Electrochemical Society

Effect of Sub 1-nm Pt Nanoparticle on the Conduction Properties of Graphene Based Field Effect Transistor Haisheng Zheng, Somik Mukherjee, Keshab Gangopadhyay, Shubhra Gangopadhyay Department of Electrical and Computer Engineering, University of Missouri, Columbia, Missouri 65201, U.S.A.

We have developed a low-power tilted-target sputtering process for achieving sub-1 nm metal nanoparticles (NPs) with controlled sizes and number densities and narrow size distributions on graphene. In this report, the unique properties of Pt NPs and their applications in graphene-based field-effect transistor by modifying the conduction channel with sub 1-nm Pt NPs has been studied. By using ultra-small size Pt NPs with quantized energy levels, we influence the electrical properties of large-area single-layer graphene by both electron transfer, carrier scattering and strain, resulting in substantial Dirac point shifting in the I-V characteristics and change in conductivity, respectively, which could be helpful to better understand the metal NP-graphene interaction. We also show that the plasma damage created on the graphene surface from the sputter deposition of metal NPs can be removed by 250 °C annealing in hydrogen.

Introduction Nanoelectronic devices such as graphene-based transistors are attractive platforms due to their promising electric properties. However, most electronic applications of graphene are handicapped by the absence of an intrinsic band gap in the as-produced material (1). Various approaches have been developed in the past decade in order to generate an energy gap in graphene (2–4). For CMOS compatible memory application, it is also essential to be able to fabricate both the n and p type device and control the threshold voltage precisely. Different methods of doping on graphene has been studied to alter the semiconducting properties of graphene (5–7), which is a process typically used to tailor the electrical properties of intrinsic traditional semiconductors such as Si. Introduced dopant atoms modify the electronic band structure of graphene and open up an energy gap between the valence and conduction bands. It is well known that graphene exhibits ballistic transport on a submicron scale and can be doped heavily using a multitude of techniques without significant loss of mobility (8). Previous experiments with graphene showed the possibility of inducing charge carriers to this material by the adsorption of various gases including NH3, H2O, and NO2 (9). Along with using molecular adsorbents, ion irradiation has been attempted as a more traditional route towards doping graphene (10). However, direct adoption of the traditional technology by ion implantation has so far met with limited success. The strong covalent C−C bonds in the honeycomb lattice structure of graphene render a substitutional doping by diffusion less efficient and less viable. Recently, researchers reported a two-step process as an efficient way to dope graphene with single atoms (11). The idea involved creating vacancies by high-energy

1

ECS Transactions, 61 (39) 1-11 (2014)

atom/ion bombardment and subsequently filling these vacancies with desired dopants. Using this technique, different elements (Pt, Co, and In) have been successfully doped in the single-atom form (11). The authors report that the high binding energy of the metalvacancy complex ensures its stability. The two-step process includes creating desired vacancies in graphene followed by dopant deposition. This technique by far is one of the most efficient atomic doping techniques for graphene as it relies on controllable vacancy generation by a high energy atom/ion source and dopant deposition by a less energetic atom/ion source. The resulting doping density, while not very high, can theoretically be controlled accurately and result in predictable device characteristics if embedded in a device architecture. The use of thermal evaporated Au NPs and thin film on doping graphene has been studied (12). The Au NPs deposited by this method, however, are larger than 10 nm with broad size variation and low number density, making precise control of the doping concentration difficult. The role of fine controlled, ultra-small metal NPs in 0.5-2 nm regime on doping graphene remains unexplored. While studying the effects of Pt NPs induced doping and strain on graphene, it is important to understand that for graphene lying on a substrate, the electronic properties of graphene alter and the charge carrier mobility can drop accordingly by orders of magnitude (13). This typically occurs due to extrinsic scattering from the surrounding medium, which could, in turn, effectively reduce the carrier mean free path (14). It is also essential to note that the electrical characteristics of graphene are also heavily linked to the quality of the 2D crystal structure (15). Raman spectroscopy is a powerful tool for monitoring the properties of graphene, damages to the crystal lattice, doping effects, mechanical strain etc. Extensive analysis has been performed in order to understand how a defect-indicating D-peak can be informative regarding the nature of the 2-D crystal structure in graphene. Raman spectroscopy can also monitor doping in graphene. Recently, researchers reported the change in Raman signatures as a function of doping by directly controlling the doping concentration by applying a gate voltage in a top or backgate configuration, which produces a resulting shift of the Fermi energy from the Dirac point. The Raman spectrum of graphene shows the following variations with doping (16): (1) The G peak position shifts to higher wavenumbers with increasing doping level and saturates for high doping levels; (2) The G peak full width at half maximum decreases with increasing doping level and saturates when the electron-hole gap becomes higher than the phonon energy; (3) The 2D peak position shifts to higher wavenumbers with increasing doping level with p doping, but shifts to lower wavenumbers with n doping; and (4) The ratio between peak intensities I2D/IG decreases with increasing doping level. Note that when there are more than one effects induced by the NPs, the changes of Raman signatures will be a combination of all the effects. It was reported that the effects of strain and charges can be optically separated from each other by correlation analysis of the two modes, enabling simple quantification of both (17). In this work, we report how ultra-small size Pt NPs influence the electrical and optical properties of large-area single-layer graphene, which could be helpful to better understand the metal NP-graphene interaction. Pt NPs were deposited using the tiltedtarget sputtering (TTS) deposition technique on top of a graphene sheet. The Fermi level of graphene can be modulated by the electric gating effect and charge transfer between the Pt NP and the graphene layer. The corresponding energy band diagram for our graphene field-effect transistor (FET) is shown in Figure 1. Electrical characterization has

2


been carried out to probe their effects on the conduction properties of graphene-based FETs.

Figure 1 Schematic of changes in energy band diagram for graphene-based FET before and after 0.7 nm Pt NP deposition. See the section of electrical characterization for more discussion.

Device fabrication Chemical vapor deposition (CVD) grown graphene on P type heavily doped Si wafer with 300 nm SiO2 was purchased from ACS MATERIAL. P-type heavily doped Si wafers were employed as the substrate with Cr-Au bottom gate contacts. A 2 nm Cr adhesion layer was deposited prior to the 80 nm Au deposition. The 300 nm thermally grown silicon dioxide (SiO2) on top of the Si substrate was used as the blocking layer. Cr-Au electrodes were thermally deposited atop the graphene layer through a shadow mask with 50 µm channel length and 63:1 width:length ratio, respectively. The samples were then annealed for 2 hours at 400 °C and 10-7 Torr working pressure to remove the PMMA residue, which is left behind by the graphene transfer process and is a suspected contaminant that alters the intrinsic properties of graphene. Pt nanoparticles (NPs) were sputtered on top of the graphene films using tilted-target sputter (TTS) deposition at 30 mW power and 23.8° target angle for either 5 or 20 seconds. Raman spectroscopy was performed on 5 s Pt NP-graphene substrates before and after 250 °C H2 annealing to understand the effect of H2 annealing on graphene electrode quality after Pt deposition. The final device structure is shown in Figure 2.

3


Figure 2. Device schematics of graphene FET (a) without and (b) with Pt nanoparticles deposited on top of graphene.

Results and discussion Pt Nanoparticle Deposition A number of fabrication methods have been developed to produce metallic nanoclusters (MNCs) (e.g. quantum dots) including electron beam lithography (EBL)(18), thermal dewetting(19), and chemical synthesis methods(20). However, the large variation in MNC size and number density as observed in these studies is undesirable for applications focusing on size-dependent behavior(21–29). We have developed a CMOScompatible TTS technique to achieve MNCs with controllable sizes, particle densities, and narrow size distributions(30–37). These MNCs have been utilized in Si- and GaAsbased single-layer(30,34) and multi-layer(32,38) non-volatile memory (NVM) devices, dye-sensitized solar cells (DSSCs)(33), trace vapor chemical sensors(39), hydrogen spillover(36), as well as room temperature two-terminal single-electron tunneling devices(31). Among these studies, we have demonstrated the utility of MNCs as discrete charge storage nodes; their ultra-small sizes allow observation and controlled charge storage and transfer down to the single electron level at ambient temperature (27 °C) (30,32,34). Sub-1 nm Pt NPs deposited using room temperature TTS process on different supporting structures in our previous study showed remarkably narrow size distributions, homogeneity, and high number density(40). Figure 3 shows the transmission electron microscopy (TEM) images of 0.7 nm Pt NPs on top of a graphene sheet and their size distribution. Due to the strong surface interaction, these particles are very stable on the graphene sheet even after prolonged TEM beam exposure, which is known to cause NP coalescence (35). Pt NPs with diameter of 0.7 nm were used for this study.

4


Figure 3. TEM image of 0.7 nm Pt NPs on top of a graphene sheet. Inset displays the narrow size distribution of these Pt NPs

Electrical characterization All I-V measurements of these devices were carried out in a triaxial guarded and electromagnetic shielded Signatone probe station in a dark, nitrogen environment at room temperature (~300 K). During the measurement, the source electrode was fixed at 0 V while the drain and gate electrodes were biased at different values. To insure that property changes were due to Pt NP deposition rather than sample variation, samples were first characterized as a control before Pt NP deposition atop the same sample and another characterization. Figure 4 shows a comparison of resistance-electric field characteristics for a graphene FET with and without conduction channel modification, with the gate voltage sweeping from -120 V to 120 V and drain voltage of 0.01 V. The relatively high gate bias is used due to the relatively small gate capacitance of a thick (300 nm) SiO2 blocking layer. The maximum resistance peak indicates the charge neutralization point, which corresponds to the Dirac point of the graphene FET. In the gate voltage range from -120 V to the Dirac point, the Fermi energy is raised above the Dirac point by the gate voltage, resulting in hole conducting transport; whereas, in the gate voltage range from 120 V to the Dirac point, the Fermi energy is lowered below the Dirac point by the gate voltage, resulting in electron conducting transport. The graphene layer of the control device (without Pt NPs) showed p-type behavior, with the Dirac point at ~55 V (0.18 MV/m). We attribute this to the type of metal in contact with graphene and band alignment between the work functions of the Cr-Au electrodes and graphene. Song et al. (41) and Park et al. (42) have shown that the work function of graphene depends on the type of metal used as the contact, with a Dirac point of more than 80 V for graphene with Cr-Au electrodes (41). Significant Dirac point shifting and change in conductivity was observed for the device with Pt NPs (e.g. from ~ 0.1 kΩ to 0.3 kΩ at gate voltage of -75 V), presumably due to the charge scattering effect with the presence of the Pt NPs in the conduction channel. For the sample with 0.7 nm Pt NPs, the Dirac point shifted in the negative direction to about 40.8 V, which is an indication of n-type

5


behavior. We attribute this shift to the difference in Fermi energy of the 0.7 nm Pt NP and the graphene layer, resulting in alignment and energy band bending of the graphene layer under thermal equilibrium. The Fermi level of Pt NPs is expected to be between 1.5 to 5.3 eV depending on the actual number of atoms of the 0.7 nm Pt NPs, meaning it could potentially be far below the 5.3 eV of bulk material (43). Assuming that the deposited NPs are spherical with a mean diameter d, the number of Pt atoms N present in a given NP can be estimated by

= ƒ

= ƒ

×

[1]

where r is the atomic radius which is 0.139 nm for Pt and f is the “packing fraction” equal to 0.67 for Body-Centered Cubic (BCC) structure and 0.74 for Face-Centered Cubic (FCC) structure(33,39). The number of atoms for 0.7 nm Pt NPs in this range of f is estimated to be 10 to 12, resulting in a relatively smaller Fermi energy for this size of Pt NPs than the 4.3 eV of graphene when in contact with Cr-Au as electrodes (41,43). Thus, the Pt NPs in this size can act as electron donor to the graphene layer, resulting n-type behavior and Dirac point shift. The doping concentration can be estimated by =

×∆

!

[2]

where Cox is the oxide capacitance (~ 11.5 nF/cm2 for 300 nm SiO2), q is the elementary charge, and ∆Vth = 40.8 eV is the Dirac point shifting due to the n-type behavior of graphene caused by Pt NPs. The doping concentration is calculated to be about 2.93×1012 cm-2, which is comparable to the 1.90×1012 cm-2 number density of the 0.7 nm Pt NPs determined by the TEM study. As the actual number density could be under estimated due to the fact that NP size smaller than 0.5 nm can hardly be resolved in the TEM image, we believe that each of the Pt NPs donated approximately one electron to the graphene conduction channel. Drain current vs. drain voltages at different gate bias were plotted as 2-D stability plots to better see the conduction properties of these n-type graphene FETs (Figure 5). 2D stability measurements have been routinely used for determination of semiconductor band gap in field effect transistors (44,45), whereas the transfer characteristics (drain current vs. gate voltage) are used to determine the mobility gap. Figure 5 (a) shows no energy band gap for the control graphene sample, whereas Figure 5 (b) shows the observed shift in Dirac point, which is in good agreement with the result observed above in Figure 4. Note that the Dirac point in the 2-D stability plot for the control sample is larger than the gate voltage (~ 120 V) and for the 0.7 nm Pt NP sample is ~ 48 V, both of which are positively shifted when compared to the value obtain from Figure 4. We attribute this to be the bias stressing effect for sample being measured under a long period of time under high bias. Note that there is a low conductance region with size of about 10 meV at the Dirac point evident in Figure 5 (b). This value is below the room temperature thermal energy of 25.9 meV. It is unlikely that there is any band gap opening with the presence of 0.7 nm Pt NPs in the conduction channel. A future low temperature measurement will be conducted to further confirm about this.

6


Gate Voltage (V) -100

-50

0

50

100

Vd= 0.01V

Ω) Resistance (kΩ

1.0 Vg= -120V to 120V 0.8 0.6 0.4 0.2 0.0 -0.4

Control With 0.7 nm Pt NPs

-0.2

0.0

0.2

0.4

E-field of Gate (M V/m) Figure 4. Comparison of Resistance-Electric field characteristics for device with and without conduction channel modification. Significant Dirac point shifting, increasing of hysteresis window, and change in conductivity was observed for the device with conduction channel modification.

Figure 5. (a) 2-D drain current-gate voltage plot showing no energy band gap for the control graphene sample; (b) A shift of Dirac point is observed for the device with 0.7 nm Pt NPs.

Raman Spectroscopy Raman spectroscopy study was carried out to further confirm the effects of doping, plasma damage of the Pt NPs to the graphene layer, and H2 annealing. Although TTS Pt NPs leads to some defect formation on the graphene sheet, the graphene sheet is defect free after depositing 0.7 nm Pt NPs with subsequent 1 hr H2 annealing at 250 °C. Annealing in inert gases in order to restore the graphene lattice has been employed before, where the intensity of the D-peak induced by defects due to ion irradiation is significantly reduced by post-annealing in N2, indicating the restoration of the damaged lattice, albeit

7


at much higher temperatures (>900 °C) (10). Given the H2 molecule’s smaller size and higher mobility compared to N2, annealing in H2 at 250 °C could be enough to recover the sample from a small degree of plasma-induced damage. According to Malard et al. (46), the most prominent features in the Raman spectra of monolayer graphene using laser excitation at 2.41 eV are the so-called G-band appearing at ~ 1582 cm-1 (graphitic peak) and the G’-band at ~ 2700 cm-1. In the case of a disordered sample or at the edge of a graphene sample, we can also see the so-called disorder-induced D-band at about half of the frequency of the G’-band (~ 1350 cm-1). The first observation from Figure 6 curve (a) is the absence of D-band peak in the graphene sample, which indicates the absence of defects in the sample. The IG/IG’ ratio ranges from below 0.5 to above 0.5. CVD-derived few-layer graphene (FLG) exhibits characteristics similar to that of turbostratic graphene, that is, lack of long-range order in the z-direction (i.e. perpendicular to the substrate). In contrast, most of the reported Raman data using few-layer exfoliated Highly Ordered Pyrolytic Graphene (HOPG) tend to keep the highly ordered structure of graphite (47). Single-layer HOPG can be differentiated from bilayer and trilayer graphene by the shape of the 2D band. In exfoliated bilayer graphene, the 2D band can be fitted with four Lorentzians, while monolayer graphene has a single, sharp Lorenztian peak (48,49). On the other hand, one, two, and three layer CVD-derived FLG on SiC and metal surfaces all exhibit a single and sharp Lorentzian-shaped peak comparable to turbostratic graphite (47). The purported reason for this phenomenon is the lack of order along the z-axis in FLG compared with the ordered AB stacking in HOPG crystals and inhomogeneous electronic coupling between the layers of FLG, having areas with strong and weak coupling. Thus, the line shape of the 2D peak alone is not enough to identify the number of layers in CVD FLG. Instead, the IG/IG’ ratio is the most relevant graphical property to use for ascertaining the number of graphene layers in CVD FLG. IG/IG’ < 0.5 is supposed to indicate monolayer graphene while IG/IG’ in the range 0.5-1 suggests 2-3 layers of graphene (46,47,50,51). To understand the effects of Pt deposition on graphene (i.e. plasma damage), graphene substrates were analyzed by Raman after 5 s Pt deposition (Figure 6 curve (b)). The appearance of D-band peaks indicating plasma-induced defects can be observed in the Raman spectra. Thus, it can be concluded here that sputtering Pt NPs introduce disorders in the sp2 carbon order characteristic of graphene. The position of the G and 2D peaks can also be used to determine combine effect of Pt NPs induced doping and strain to the graphene layer. Figure 6 shows a representative RAMAN spectra comparing the G and 2D peaks position of graphene layer under different process. After sputtering 0.7 nm Pt nanoparticle, the shifting of G band (from 1582.9 to 1591.9 cm-1) and 2D band (from 2699.6 to 2695.7 cm-1) in the RAMAN spectra indicates an n-type doping of graphene by the 0.7nm Pt nanoparticles (16). Although this collaborates with our electrical measurements, the shift in 2D peak position (~4 cm-1) is within the expected error bar (±6.3) and thus is inconclusive. However, the shift in G band location is significantly larger than the error bars and hint towards a successful ntype doping of the graphene sample. Apart from the observed shifts in G and 2D band positions, there is also the introduction of a small defect indicative D peak, which disappears after annealing in H2 at 250oC for an hour. Despite significant reduction in the D peak, the 2D band positioning stays red shifted compared to defect free graphene and the G peak remains blue shifted relative to the defect free graphene sample. This

8


indicated successful n-type doping by charge transfer and/or strain of the graphene by the Pt NPs. The average G and 2D peak position for multiple sampling are summarized in Table I. More intensive RAMAN analysis to understand the effect of nanoparticle induced doping of graphene and other 2-D crystals would be performed in the future.

Normalized Intensity

Defect free graphene 5s Pt sputtered H 2 annealed 0.7 nm Pt sputtered 1588.9 c. Removal of defects by H2 anneal

2697

1591.9

b. Appearence of D peak after sputtering

2695.7

1582.9

a. Defect free Graphene sheet

2699.6

1400 1600 1800 2000 2200 2400 2600 2800 3000

RAMAN Shift (cm-1) Figure 6. Raman spectroscopy study of graphene sheet before (curve a) and after (curve b) depositing 0.7 nm Pt NPs on top and after 250 °C H2 annealing (curve c). Note the shifting of G-band and 2D-band indicates an n-type doping of graphene by the 0.7 nm Pt NPs. TABLE I. Summary of average G and 2D peaks position for each condition in Figure 6 Column Header Goes Here G band (cm-1) 2D band (cm-1) Before Pt NPs deposition 1583.8±1.8 2696.5±4.9 After 5sec Pt NPs deposition 1591.1±4.1 2700.1±4.1 5sec sample after 250 ̊C 1587.8±2.6 2697.1±6.3 Hydrogen annealing

Conclusion To conclude, we have demonstrated the unique properties of graphene-based FETs with conduction channel modification with sub-nm Pt NPs. Using ultra-small Pt NPs with quantized energy levels, we influenced the electrical properties of large-area single-layer graphene, resulting in n-type doping behavior and substantial Dirac point shifting in the IV characteristics. We have also shown that the graphene surface plasma damage created by the TTS deposition of metal NPs can be removed by 250 °C H2 annealing. This study offers a wealth of information toward understanding the properties of MNC-graphene interaction. Future study including strain measurement, low temperature measurement and first principle simulation will further reveal these change of electronic structure of graphene due to these sub nm Pt NPs.

9


Acknowledgments This work was supported by the National Science Foundation grants GOALI-0802157 and GOALI-1232178, as well as the Nano Technology Enterprise Consortium (NTEC) contract W15QKN-11-9-0001-RPP1. We thank Charles M. Darr for the helpful suggestion on the manuscript. References 1. T. O. Wehling et al., Nano Lett., 8, 173–177 (2008) . 2. S. Peng and K. Cho, Nanotechnology, 11, 57 (2000) . 3. H. Chang, J. Do Lee, S. M. Lee, and Y. H. Lee, Appl. Phys. Lett., 79, 3863–3865 (2001) . 4. J. Zhao, A. Buldum, J. Han, and J. P. Lu, Nanotechnology, 13, 195 (2002) . 5. X. Fan, Z. Shen, A. Q. Liu, and J.-L. Kuo, Nanoscale, 4, 2157–2165 (2012) . 6. P. A. Denis, Chem. Phys. Lett., 492, 251–257 (2010). 7. P. P. Shinde and V. Kumar, Phys. Rev. B, 84, 125401 (2011) . 8. F. Schedin et al., Nat. Mater., 6, 652–5 (2007) . 9. K. S. Novoselov et al., Science., 306, 666–669 (2004) . 10. B. Guo et al., Nano Lett., 10, 4975–4980 (2010) . 11. H. Wang et al., Nano Lett., 12, 141–144 (2012) . 12. Y. Wu et al., Small, 8, 3129–36 (2012) . 13. Y.-C. Lin et al., Nano Lett., 12, 414–9 (2012) . 14. L. A. Ponomarenko et al., Phys. Rev. Lett., 102 (2009) . 15. Y.-C. Lin, C.-Y. Lin, and P.-W. Chiu, Appl. Phys. Lett., 96, 133110 (2010) . 16. C. Casiraghi, Phys. Rev. B, 80, 233407 (2009) . 17. J. E. Lee, G. Ahn, J. Shim, Y. S. Lee, and S. Ryu, Nat. Commun., 3, 1024 (2012) . 18. H. C. George et al., Appl. Phys. Lett., 96, 042114 (2010). 19. J. Dufourcq et al., Appl. Phys. Lett., 92, 073102 (2008) . 20. S. Wakamatsu, J. Nakada, S. Fujii, U. Akiba, and M. Fujihira, Ultramicroscopy, 105, 26–31 (2005) . 21. M. Hirasawa, H. Shirakawa, H. Hamamura, Y. Egashira, and H. Komiyama, J. Appl. Phys., 82, 1404–1407 (1997) . 22. J. J. Brege, C. E. Hamilton, C. A. Crouse, and A. R. Barron, Nano Lett., 9, 2239– 2242 (2009) . 23. A. R. Canário, E. A. Sanchez, Y. Bandurin, and V. A. Esaulov, Surf. Sci., 547, L887– L894 (2003) . 24. J. Fu and Y. Zhao, Nanotechnology, 21, 175303 (2010) . 25. S. P. Park, S. S. Kim, J. H. Kim, C. N. Whang, and S. Im, Appl. Phys. Lett., 80, 2872–2874 (2002) . 26. Z.-Q. Tian, B. Ren, and D.-Y. Wu, J. Phys. Chem. B, 106, 9463–9483 (2002) . 27. Y. Volokitin et al., Nature, 384, 621–623 (1996) . 28. J. Dufourcq et al., Mater. Sci. Eng. C, 27, 1496–1499 (2007) . 29. Z. Liu, C. Lee, V. Narayanan, G. Pei, and E. C. Kan, IEEE Trans. Electron Devices, 49, 1606–1613 (2002) . 30. M. Yun, D. W. Mueller, M. Hossain, V. Misra, and S. Gangopadhyay, IEEE Electron Device Lett., 30, 1362–1364 (2009) . 31. M. Yun, B. Ramalingam, and S. Gangopadhyay, Nanotechnology, 22, 465201 (2011) .

10


32. M. Yun, B. Ramalingam, and S. Gangopadhyay, J. Electrochem. Soc., 159, H393 (2012) . 33. S. Mukherjee et al., Nanotechnology, 23, 485405 (2012) . 34. R. C. Jeff et al., Appl. Phys. Lett., 99, 072104 (2011) . 35. B. Ramalingam, S. Mukherjee, C. J. Mathai, K. Gangopadhyay, and S. Gangopadhyay, Nanotechnology, 24, 205602 (2013) . 36. S. Mukherjee, B. Ramalingam, and S. Gangopadhyay, J. Mater. Chem. A, 2, 3954 (2014) . 37. S. Mukherjee, B. Ramalingam, K. Gangopadhyay, and S. Gangopadhyay, J. Electrochem. Soc., 161, F493–F499 (2014). 38. B. Ramalingam, H. Zheng, and S. Gangopadhyay, Appl. Phys. Lett., 104, 143103 (2014) . 39. H. Zheng, B. Ramalingam, V. Korampally, and S. Gangopadhyay, Appl. Phys. Lett., 103, 193305 (2013) . 40. S. Mukherjee, B. Ramalingam, K. Gangopadhyay, and S. Gangopadhyay, J. Electrochem. Soc., 161, F493–F499 (2014) . 41. S. M. Song, J. K. Park, O. J. Sul, and B. J. Cho, Nano Lett., 12, 3887–92 (2012) . 42. N. Park, B.-K. Kim, J.-O. Lee, and J.-J. Kim, Appl. Phys. Lett., 95, 243105 (2009) . 43. A. Nie et al., Int. J. Quantum Chem., 107, 219–224 (2007). 44. M. Han, B. Özyilmaz, Y. Zhang, and P. Kim, Phys. Rev. Lett., 98, 206805 (2007). 45. Y.-C. Chen et al., ACS Nano, 7, 6123–6128 (2013) . 46. E. F. Holby and D. Morgan, J. Electrochem. Soc., 159, B578–B591 (2012) . 47. P. T. Yu, W. Gu, R. Makharia, F. T. Wagner, and H. A. Gasteiger, ECS Trans., 3, 797–809 (2006) . 48. C. Solliard and M. Flueli, Surf. Sci., 156, Part 1, 487–494 (1985) . 49. Y. Shao-Horn et al., Top. Catal., 46, 285–305 (2007) . 50. Y.-G. Guo, J.-S. Hu, and L.-J. Wan, Adv. Mater., 20, 2878–2887 (2008) . 51. D. Choudhury, B. Das, D. D. Sarma, and C. N. R. Rao, Chem. Phys. Lett., 497, 66–69 (2010) .

11