Formation of high conductive nano-crystalline silicon embedded in amorphous silicon-carbide films with large optical band gap Yang Ji, Dan Shan, Mingqing Qian, Jun Xu, Wei Li, and Kunji Chen Citation: AIP Advances 6, 105107 (2016); View online: https://doi.org/10.1063/1.4965922 View Table of Contents: http://aip.scitation.org/toc/adv/6/10 Published by the American Institute of Physics Articles you may be interested in Evaluation of microstructures and carrier transport behaviors during the transition process from amorphous to nanocrystalline silicon thin films Journal of Applied Physics 105, 054901 (2009); 10.1063/1.3087500 Pulse voltage induced phase change characteristics of the ZnxSbyTez phase-change prototype device AIP Advances 6, 105211 (2016); 10.1063/1.4966909 Characterization of the Si:H network during transformation from amorphous to micro- and nanocrystalline structures Journal of Applied Physics 100, 103701 (2006); 10.1063/1.2384812 Determining the material structure of microcrystalline silicon from Raman spectra Journal of Applied Physics 94, 3582 (2003); 10.1063/1.1596364 Crystal size and temperature measurements in nanostructured silicon using Raman spectroscopy Journal of Applied Physics 90, 4175 (2001); 10.1063/1.1398601 Raman spectroscopy of amorphous and microcrystalline silicon films deposited by low-pressure chemical vapor deposition Journal of Applied Physics 78, 6999 (1998); 10.1063/1.360468 AIP ADVANCES 6, 105107 (2016) Formation of high conductive nano-crystalline silicon embedded in amorphous silicon-carbide films with large optical band gap Yang Ji, Dan Shan, Mingqing Qian, Jun Xu,a Wei Li, and Kunji Chen National Laboratory of Solid State Microstructures and School of Electronic Science and Engineering and Collaborative Innovation Center of Advanced Microstructures, Nanjing University, Nanjing 210093, China (Received 21 August 2016; accepted 8 October 2016; published online 17 October 2016) High conductive phosphorus-doped nano-crystalline Si embedded in SiliconCarbide (SiC) host matrix (nc-Si:SiC) films were obtained by thermally annealing doped amorphous Si-rich SiC materials. It was found that the room conductivity is increased significantly accompanying with the increase of doping concentrations as well as the enhanced crystallizations. The conductivity can be as high as 630 S/cm for samples with the optical band gap around 2.7 eV, while the carrier mobility is about 17.9 cm2 / V·s. Temperature-dependent conductivity and mobility measurements were performed which suggested that the carrier transport process is strongly affected by both the grain boundaries and the doping concentrations. © 2016 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). [http://dx.doi.org/10.1063/1.4965922] I. INTRODUCTION Recently, un-doped and doped nano-crystalline Si (nc-Si) embedded in amorphous SiC host matrix (nc-Si:SiC) films have attracted much attention since they can be applied for many kinds of optoelectronic devices, such as the next generation of solar cells and light emitting devices.1–6 It was reported that the band offset both in conduction band and valence band for nc-Si:SiC system is the lowest compared with other conventionally used dielectric materials (such as SiO2 and SiNx ), which is helpful for the electron and hole transport and in turn to improve the device performance. For example, it was reported that, by using p-type nc-Si:SiC films to get hetero-junction solar cells with n-type crystalline-Si (c-Si), the spectral response in the visible light range can be enhanced compared with that of conventional c-Si p-n junction solar cells.7 Moreover, an intense visible light emission from nc-Si:SiC films deposited using advanced electron-cyclotron-resonance chemical vapor deposition (ECRCVD) technique was also observed at room temperature under laser excitation.8 X. Xu et al. used laser crystallization technique to get the p-i-n structures containing nc-Si:SiC films and they observed the improved electroluminescence due to the enhanced radiative recombination probability.9 Usually, nc-Si:SiC films can be obtained by growing Si-rich SiC films or amorphous Si/SiC multilayers with subsequently thermal annealing at high temperature.5–12 By controlling the Si/C ratio or the post-annealing conditions, the optical band gap can be tunable which brings convenience to the design of optoelectronic devices. However, the SiC matrix with a large band gap usually has a poor conductive property and in turn impedes the carrier transport process which will deteriorate the device performance.13,14 Lechner et al. fabricated the doped nc-Si films and they obtained the maximum conductivity up to about 5 S/cm with the very small conductivity activation energy.15 From the view point of device application, it is highly desired to get nc-Si:SiC films with high conductivity at the suitable optical band gap (>2.5 eV) so that they can act as the window layer aThe corresponding Author: Jun Xu Email: [email protected] Tel:+8613585174001 Fax: 86-25-83595535 2158-3226/2016/6(10)/105107/9 6, 105107-1 © Author(s) 2016 105107-2 Ji et al. AIP Advances 6, 105107 (2016) or the conductive layer in nano-crystalline SiC film-based optoelectronic devices without absorbing incident (or emitted) light too much.13,16 In our previous work, hydrogenated amorphous silicon carbide (a-SiC:H) films were prepared in a plasma-enhanced chemical vapor deposition (PECVD) system. Nc-Si:SiC films were fabricated after annealing at 900 ◦ C and 1000 ◦ C. It was found that for samples annealed at 1000 ◦ C, the average size of nc-Si is 9 nm and the crystallinity is reached to 70 %. The optical band gap is enlarged from 1.9 eV to 2.2 eV with annealing temperature increasing, which can be attributed to the increase of Si-C bonds. The conductivity reaches to 1.2×10-6 S/cm, which is 4 orders of magnitude higher than that of as-deposited sample.14 However, the conductivity is still too low to apply in the related devices such as thin film solar cells or light emitting devices. In the present work, nc-Si:SiC with various phosphorus (P) doping concentrations are prepared by annealing P-doped amorphous Si-rich SiC films. The micro-structures, optical and electronic properties are studied as a function of doping concentrations. It is found that with increasing the P doping concentrations, the conductivity of formed nc-Si:SiC is gradually increased and a very high conductive nc-Si:SiC film with conductivity of 630 S/cm can be obtained. Furthermore, the temperature-dependent behaviors of conductivity and mobility are studied, which reveal that the Fermi level indeed shifts to the conduction band with doping concentration increasing and the transport mechanism changes with increasing the P doping concentrations. II. EXPERIMENT Hydrogenated amorphous SiC films with various P doping concentrations were prepared on quartz glass plates and p-Si substrates in a PECVD system using gas mixtures of pure SiH4 , CH4 , PH3 and H2 . The flow rate of SiH4 and CH4 was kept at 5 sccm and 1.5 sccm respectively. The flow rate of PH3 (1 % diluted in H2 ) was changed from 0 sccm to 5 sccm for various doping concentrations. Here, we define the doping gas ratio R as the gas ratio of [PH3 ] to [SiH4 ] which is changed from 0 to 1. During the deposition process, the radio-frequency power, chamber pressure and substrate temperature were kept at 30 W, 10 mTorr and 250 ◦ C respectively. The deposition time was 40 min. Fig. 1(a) and (b) are the XPS spectra around 283 eV and 100 eV of as-deposited samples without doping, which is related to the C 1s and Si 2p signals, respectively. By integrating the C 1s peak (∼283 eV) and the Si 2p peak (∼100 eV), the composition ratio of Si/C is about 20, which implies that the as-deposited sample is Si-rich amorphous SiC film. It is also found that the Si 2p peak can be decomposed into two sub-bands, which is assigned to Si-Si state (EBSi2p =99.6±0.1 eV) and Si-C state (EBSi2p =100.4±0.1 eV), respectively.15 It is suggested from XPS spectra that the most of Si atoms are bonded to neighboring Si and only small portion of Si are bonded to C atoms. It is noted that the Si/C ratio is almost the same for un-doped and P-doped samples even after thermal annealing. XPS spectra also reveal that the oxygen atom content is 2 %∼3 % in all the doped samples (not shown in Fig. 1), which may be attributed to oxidation by air during both deposition and annealing process. FIG. 1. (a) Si 2p and (b) C 1s spectra of as-deposited un-doped samples. 105107-3 Ji et al. AIP Advances 6, 105107 (2016) The as-deposited samples were dehydrogenation at 450 ◦ C for 1 h in N2 ambient and annealing at 1000 ◦ C for 1 h in N2 atmosphere in sequence to form nc-Si:SiC films. It is found that the hydrogen is completely effused out of the films after above thermal treatments. The changes of film microstructures before and after annealing were characterized by Raman spectra using a Jobin Yvon Horiba HR800 spectrometer while the optical absorption was measured by a Shimadzu UV-3600 spectrophotometer for samples deposited on quartz plates. To measure electronic properties (mobility, doping type, sheet carrier concentration and sheet resistance) at room temperature and at higher temperature (from 300 K to 410 K), a LakeShore 8400 series Hall effect measurement system was utilized by using Van der Pauw configurations for samples on quartz substrates. The films deposited on p-Si substrates were used to prepare samples for cross-section transmission electron microscopy (TEM) observations. III. RESULTS AND DISCUSSION Fig. 2(a) is the P 2p XPS spectra for 1000 ◦ C annealed SiC films with doping gas ratio R=0, 0.6 and 1, respectively. There are no detectable signals related to P atoms for un-doped samples as shown in Fig. 2(a). However, a weak peak centered at 129.6 eV, which is attributed to the formation P-Si bonds,17 can be observed for P-doped samples with R=0.6 and the peak becomes stronger for samples with R=1. It is estimated that the P doping concentration is about 2.1 % and 2.3 % for samples with R= 0.6 and 1, respectively. We also measured the depth-profile XPS spectra as shown in Fig. 2(b), in which, each spectra was detected after Ar+ ion etching for various time (the etching speed was 0.1∼0.2 nm/s), which corresponds to the maximum detected position of about 120 nm from the surface. It is found that the P-related signals are not changed obviously with the depth, indicating atomic distribution is quite uniform in our samples. Fig. 3 is the Raman scattering spectra of as-deposited and 1000 ◦ C annealed samples with various P doping concentrations. A broad peak at 473 cm-1 can be found in as-deposited films, which is related to the transverse optical (TO) phonon mode of amorphous Si-Si phases. After annealing at 1000 ◦ C, a narrow peak appears at 516 cm-1 , indicating the transformation from a-Si to nc-Si.18 It is found that for annealed samples, Raman peak is almost unchanged with increasing the P doping concentrations, while the Raman intensity is gradually enhanced which indicates the enhanced crystallization. In order to estimate crystallinity of annealed films, three components are decomposed in Raman spectra (see the inset in Fig. 3), which is a crystalline component peaked at 516 cm-1 , an amorphous component peaked at 473 cm-1 , and an intermediate component peaked at 504 cm-1 corresponding to Si-Si bonds at grain boundaries.19–22 The crystallinity (X c ) can be estimated based on the following formula:23 Xc = Ic /(Ic + 0.88Ia ), (1) where, I a and I c are integrated intensities of the amorphous component peaked at 473 cm-1 and the crystalline component peaked at 504 cm-1 and 516 cm-1 respectively. The calculated results X c for the FIG. 2. P 2p spectra of annealed samples (a) with different R values and (b)with different depths. 105107-4 Ji et al. AIP Advances 6, 105107 (2016) FIG. 3. Raman spectra of samples with various P doping concentrations. The inset presents the deconvolution of the Raman spectrum of annealed un-doped sample. intrinsic samples is about 76 % after 1000 ◦ C annealing. With increasing the P doping concentrations, X c goes up and can be reached as high as 91 % for samples with doping ratio R=1. It seems that introduction P impurities are helpful for promoting the crystallization. The effect of P doping on the structures of nc-Si films was also discussed previously.24 It was reported that with increasing the P doping concentrations, the grain size of formed nc-Si is increased and the nc-Si film becomes more ordered which may be related to the reduced strain in the films. However, the further work is need to exactly understand the enhanced crystallinity due to the P doping. The microstructures after 1000 ◦ C annealing are further studied by cross-sectional TEM observations. Fig. 4(a) shows the cross-section TEM image of 1000 ◦ C annealed sample. It is seen that the film thickness is about 170 nm. Fig. 3(b) and (c) are the high resolution cross-section TEM images for annealed samples with doping gas ratio R=0.6 and 1, respectively. The formation of Si nano-crystals can be clearly identified in the figures and the lattice constant is 0.31 nm which corresponds to the Si (111) orientation. The average size of nano-dots is 12 nm for sample with doping ratio R=0.6, which is increased to 15 nm for sample with R=1. The TEM observations also suggest the crystallization is enhanced with increasing the P doping concentrations. This is consistent with the Raman scattering results. The optical band gaps E opt of annealed samples with various P doping ratio are deduced from Tauc formula:25 (αhv)1/2 = B(hv − Eopt ), (2) where, α is the absorption coefficient, hv is photon energy and B is a constant. As shown in Fig. 5(a), the optical band gap is gradually increased from 2.2 eV to 2.8 eV for 1000 ◦ C annealed samples with increasing the doping ratio R from 0 to 1. It is noticed that the optical band gap is enlarged after annealing compared with that of as-deposited films, which can be clearly seen in the inset of Fig. 5(b) that the sample becomes more transparent after annealing. It has been reported that optical band gap is enlarged after annealing in B-doped nc-SiOx Ny films,13 which was explained as the contribution both from the quantum confinement effects and the increased amount of SiOx composition due to formation of Si nano-crystals. In our previous work, we also found that the existence of grain boundary after annealing can influence the optical band gap obviously.26 Therefore, the increasing optical band gap with doping ratio R can be ascribed to the increased influences of grain boundaries due to enhanced crystallization as well as the increased amount of Si-C bonds due to the formation of nc-Si phases though we cannot completely rule out of the effect of oxygen contamination during 105107-5 Ji et al. AIP Advances 6, 105107 (2016) FIG. 4. Cross-section TEM image of 1000 ◦ C annealed sample with doping ratio R=0.6 (a) and high resolution TEM images of 1000 ◦ C annealed samples with doping ratio R=0.6 (b) and R=1 (c), respectively. the annealing process. However, it is argued that the oxygen content measured by XPS for all the samples is 2 %∼3 % both for as-deposited and annealed samples as we mentioned before. Fig. 5(b) shows the transmittance spectra of samples with R=0.22 before and after annealing. Though existence of the strong interference effect due to the film thickness (∼170 nm), it can be seen that the transmittance is improved in the visible light range, especially for the spectral region with the wavelength less than 550 nm after annealing. It is also observed that the transmittance can be FIG. 5. (a) Optical band gap of 1000 ◦ C annealed samples as a function of doping ratio R. (b) Transmittance spectra of samples (R=0.22) before and after annealing with a characteristic thickness of ∼170 nm. Insets are the pictures of samples (R=0.22) before and after annealing. 105107-6 Ji et al. AIP Advances 6, 105107 (2016) further improved in the samples with the larger R values (not shown here) due to their higher band gap as shown in Fig. 5(a). It is worth noting that the transmittance spectra shown in Fig. 5(b) are from the samples with thickness of ∼170 nm, which is far more beyond the demand thickness (∼10 nm) in actual thin film solar cells as a window layer.5 Based on the measured absorption coefficient, it is estimated that the transmittance will exceed 80% in the whole visible spectral range for annealed samples with a thickness of 10 nm. Fig. 6 shows the room temperature conductivity σ and mobility µ as a function of P doping ratio R for annealed samples. It is found that both the conductivity and mobility are gradually increased with increasing the doping ratio R from 0 to 0.6. The conductivity can be as high as 630 S/cm with the mobility of 17.9 cm2 /V·s, which is significantly higher than that of doped SiC and SiC powder compact reported previously.14,27 Further increasing doping ratio R results in the slight reduction of mobility but the conductivity still keeps its high level. The increasing of mobility with the doing concentrations can be attributed to the improved crystallinity due to the enhanced crystallization as discussed before which reduces scattering with the disordered amorphous structures. However, further increasing P doping concentrations will cause the strong free-electron scattering and/or grain boundary scattering which results in the slight decrease of mobility. In order to further understand the carrier transport process of annealed samples, the temperaturedependent conductivity and mobility behaviors are studied. As shown in Fig. 7, when doping ratio R is lower than 0.6, the relationship between dark conductivity σ and the temperature T can be well described as the formula, which usually used to describe the transport process in micro-crystalline and nano-crystalline Si-based materials:13,28 σ = σ0 exp(-Ea /kB T ), (3) where, σ0 , E a and k B are constant parameter, conductivity activation energy and Boltzmann constant, respectively. The activation energy E a can be deduced through Arrhenius plots as given in Fig. 7.28,29 It is found that E a is obviously decreased with increasing the P doping concentrations, from 400 meV of un-doped sample to 12 meV of sample with R=0.22, which means the Fermi level is very closed to the bottom of conduction band due to the P doping so that the conduction electron density can be increased significantly (n ∝ exp(-E a /k B T )). For sample with R>0.3, dark conductivity is nearly independent of temperature between 300 K and 410 K, which means that the Fermi level is located within the conduction band.30 With further increasing the doping gas ratio R, the more activated P atoms leads to the further increase of carrier concentrations to increase the conductivity. Recently, we experimentally demonstrated that the phosphorus dopants can occupy the inner sites of Si nc-Si substitutionally, which generate the impurity-related level in the gap to provide the conduction electrons.31 In fact, with increasing the P doping concentrations, the measured carrier concentration FIG. 6. Room temperature conductivity and mobility as a function of P doping ratio of 1000 ◦ C annealed samples. 105107-7 Ji et al. AIP Advances 6, 105107 (2016) FIG. 7. Temperature-dependent dark conductivity of 1000 ◦ C annealed samples with various P doping ratio. increases from 1.3×1014 cm-3 to 2.6×1020 cm-3 . On the other hand, thermal treatment can promote crystallization in our samples which leads to the increase of carrier mobility (i.e. form 2.3 cm2 /V·s for un-doped one to 17.9 cm2 /V·s for P-doped one with R=0.6) as we mentioned before. Since both of the mobility and carrier concentration are increased, the conductivity is consequently increased obviously as given in Fig. 6. Fig. 8 shows the temperature-dependent mobility for samples with various P doping concentrations. It is found that the mobility µ increases with the temperature ranged from 300 K to 410 K for intrinsic sample. At the present stage, the increase in mobility with temperature is not fully understood. The similar behaviors were previously observed in microcrystalline Si films. It was explained in term of the effect of grain boundaries since they can trap the carriers in the ionization defects states to form the potential barrier. At high temperature, the carriers gain much more energy to pass through the potential barrier which results in the increased mobility.26 Another possibility is the scattering process FIG. 8. Temperature-dependent mobility of 1000 ◦ C annealed samples with various P doping ratio. 105107-8 Ji et al. AIP Advances 6, 105107 (2016) via the ionized impurity centers which also induce the increased mobility with temperature. However, for samples with high P doping concentrations, they exhibit different temperature-dependent behaviors. The mobility is slightly decreased with the measured temperature. It is suggested that the impurity ionization carrier concentration is so high for samples with high doping concentrations that carriers can saturate the trapping states and reduce the potential barrier height.26,32 Consequently, grain boundary barriers are not serious obstacles for the carrier transport for heavily doped samples. The carrier transport process is then dominantly controlled by phonon scattering mechanism and the mobility decreases with increasing the temperature. As discussed before, for samples with P doping ratio R higher than 0.22, they are degenerate semiconductors with carrier concentration of 1020 cm-3 , it is reasonable that the carrier transport characteristics in those samples are similar with metal-like materials.33,34 IV. CONCLUSIONS In summary, P-doped nc-Si:SiC films with high conductivity can be achieved by annealing P-doped amorphous Si-rich SiC layer at 1000 ◦ C. It is found that the crystallinity of annealed samples can be enhanced with increasing the P doping concentrations which is responsible for the improved electronic properties. The obtained film exhibits the best conductivity of 630 S/cm with the optical band gap of 2.7 eV while the measured Hall mobility is about 17.9 cm2 /V·s. Moreover, both the temperature-dependent conductivity and mobility were studied. It is shown that with increasing the doping ratio, the activation energy is gradually decreased which indicates the up-shift of Femi level to the conduction band. For samples with low doping ratio, the mobility is increased with temperature implying the influences of grain boundaries and ionized impurities. However, for samples with high doping ratio, they show the metal-like behaviors that the mobility is decreased with temperature, which suggests that the photon scattering mechanism dominates the carrier process in the heavilydoped films. Our results demonstrate that the present high conductive nanocrystalline films with large optical band gap can act as the good window materials potentially used in Si-based optoelectronic devices. ACKNOWLEDGMENTS This work was supported by 973 Project (2013CB632101), NSFC (No. 11274155), 333 Project of Jiangsu Province (BRA2015284), PAPD and the Research Innovation Program for College Graduates of Jiangsu Province (SJLX15 0019). 1 C. Huh, T.-Y. Kim, C.-G. Ahn, and B. Y. Kim, Appl. Phys. Lett. 106, 211103 (2015). Conibeer, M. Green, R. Corkish, Y. Cho, E.-C. Cho, C.-W. Jiang, T. Fangsuwannarak, E. Pink, Y. Huang, T. Puzzer, T. Trupke, B. Richards, A. Shalav, and K. Lin, Thin Solid Films 511–512, 654–662 (2006). 3 W. W. Mu, P. Zhang, J. Xu, S. H. Sun, J. Xu, W. Li, and K. J. Chen, IEEE J. Sel. Top. Quant. 20, 8200106 (2014). 4 Y. Kurokawa, S. Yamada, S. Miyajima, A. Yamada, and M. Konagai, Curr. Appl. Phys. 10, S435–S438 (2010). 5 Y. Q. Cao, P. Lu, X. W. Zhang, J. Xu, L. Xu, and K. J. Chen, Nanoscale Res. Lett. 9, 634 (2014). 6 P. Loper, M. Canino, D. Qazzazie, M. Schnabel, M. Allegrezza, C. Summonte, S. W. Glunz, S. Janz, and M. Zacharias, Appl. Phys. Lett. 102, 033507 (2013). 7 D. Song, E.-C. Cho, G. Conibeer, C. Flynn, Y. Huang, and M. A. Green, Sol. Energy Mater. Sol. Cells 92, 474–481 (2008). 8 S. J. Xu, M. B. Yu, Rusli, S. F. Yoon, and C. M. Che, Appl. Phys. Lett. 76, 2550 (2000). 9 X. Xu, Y. Q. Cao, P. Lu, J. Xu, W. Li, and K. J. Chen, IEEE Photonics Journal 6, 2200207 (2014). 10 C. Summonte, M. Canino, M. Allegrezza, M. Bellettato, A. Desalvo, R. Shukla, I. P. Jain, I. Crupi, S. Milita, L. Ortolani, L. López-Conesa, S. Estradé, F. Peiró, and B. Garrido, Mater. Sci. Eng. B 178, 551–558 (2013). 11 Y. J. Rui, S. X. Li, J. Xu, C. Song, X. F. Jiang, W. Li, K. J. Chen, Q. M. Wang, and Y. H. Zuo, J. Appl. Phys. 110, 064322 (2011). 12 Y. Q. Cao, J. Xu, Z. Y. Ge, Y. Y. Zhai, W. Li, X. F. Jiang, and K. J. Chen, J. Mater. Chem. C 3, 12061 (2015). 13 M. Perani, N. Brinkmann, A. Hammud, D. Cavalcoli, and B. Terheiden, J. Phys. Chem. C 119, 1390713914 (2015). 14 C. Song, Y. J. Rui, Q. B. Wang, J. Xu, W. Li, K. J. Chen, Y. H. Zuo, and Q. M. Wang, J. Alloy. Compd. 509, 3963–3966 (2011). 15 R. Lechner, A. R. Stegner, R. N. Pereira, R. Dietmueller, M. S. Brandt, A. Ebbers, M. Trocha, H. Wiggers, and M. Stutzmann, J. Appl. Phys. 104, 053701 (2008). 16 A. V. Semenov, V. M. Puzikov, M. V. Dobrotvorskaya, A. G. Fedorov, and A. V. Lopin, Thin Solid Films 516, 2892903 (2008). 17 W. B. Ying, Y. Mizokawa, Y. Kamiura, K. Kawamoto, and W. Y. Yang, Appl. Surf. Sci. 181, 114 (2001). 18 I. H. Campbell and P. M. Fauchet, Solid State Commun. 58, 739 (1986). 2 G. 105107-9 19 S. Ji et al. AIP Advances 6, 105107 (2016) Veprek, F. A. Sarott, and Z. Iqbal, Phys. Rev. B 36, 3344 (1987). S. Mavi, A. K. Shukla, S. C. Abbi, and K. P. Jain, J. Appl. Phys. 66, 5322 (1989). 21 Y. He, Y. Wei, G. Zheng, M. Yu, and M. Liu, J. Appl. Phys. 82, 3408 (1997). 22 D. Das and K. Bhattacharya, J. Appl. Phys. 100, 103701 (2006). 23 P. Martı́n, A. Torres, J. Jiménez, A. Rodrı́guez, J. Sangrador, and T. Rodrı́guez, Mat. Res. Soc. Symp. Proc. 808, A4.39.1–A4.39.6 (2004). 24 M. H. Gullanar, Y. H. Zhang, H. Chen, W. S. Wei, G. Y. Xu, T. M. Wang, R. Q. Cui, and W. Z. Shen, J. Cryst. Growth. 256, 254–260 (2003). 25 J. Tauc, R. Grigorovici, and A. Vancu, Phys. Stat. Sol. 15, 627 (1966). 26 J. Y. W. Seto, J. Appl. Phys. 46, 5247 (1975). 27 Y. W. Kim, K. J. Kim, H. C. Kim, N. H. Cho, and K. Y. Lim, J. Am. Ceram. Soc. 94, 991–993 (2011). 28 D. Das and B. Sain, J. Appl. Phys. 114, 073708 (2013). 29 D. M. Tan, P. Zhang, J. Xu, Y. Q. Cao, P. Lu, W. Li, J. Xu, and K. J. Chen, J. Vac. Sci. Technol. B 33, 041205 (2015). 30 S. P. Lau, J. M. Marshall, and T. E. Dyer, Philos. Mag. B 72, 323–333 (1995). 31 P. Lu, W. W. Mu, J. Xu, X. W. Zhang, W. P. Zhang, W. Li, L. Xu, and K. J. Chen, Scientific Reports 6, 22888 (2016). 32 G. Baccarani, B. Riccò, and G. Spadini, J. Appl. Phys. 49, 5565–5570 (1978). 33 S. Faÿ, J. Steinhauser, N. Oliveira, E. Vallat-Sauvain, and C. Ballif, Thin Solid Films 515, 8558–8561 (2007). 34 N. Basescu, Z.-X. Liu, D. Moses, A. J. Heeger, H. Naarmann, and N. Theophilou, Nature 327, 403–405 (1987). 20 H.