lable at ScienceDirect

Current Applied Physics 15 (2015) 648e653

Contents lists avai

Current Applied Physics

journal homepage: www.elsevier .com/locate/cap

The effect of annealing in forming gas on the a-IGZO thin filmtransistor performance and valence band cut-off of IGZO on SiNx

Raj Kamal a, Piyush Chandravanshi a, Duck-Kyun Choi b, Santosh M. Bobade c, *

a Department of Electronics & Communication, Jaypee University of Engineering & Technology, Raghogarh, Guna, Madhya Pradesh, Indiab Division of Materials Science and Engineering, Heangdang-dong 17, Seoungdong-Ku, Seoul, South Koreac Department of Physics, Jaypee University of Engineering & Technology, Raghogarh, Guna, Madhya Pradesh, India

a r t i c l e i n f o

Article history:Received 26 August 2014Received in revised form30 January 2015Accepted 22 February 2015Available online 5 March 2015

Keywords:TFTIGZOTransparent TFTXPS

* Corresponding author.E-mail address: santoshbobade@gmail.com (S.M. B

http://dx.doi.org/10.1016/j.cap.2015.02.0171567-1739/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t

In this investigation, the carrier concentration gradient between channel and contact region is achievedto improve the Thin film Transistors (TFT) performance by employing annealing at 350 �C in forming gas(N2 þ 5% H2). The contact region is covered with Mo metal and the channel region is only exposed toforming gas to facilitate the diffusion controlled reaction. The TFT using a-IGZO active layer is fabricatedin ambient of Ar:O2 in ratio 60:40 and the conductivity of the order of 10�3 S/cm is measured for as-deposited sample. The electrical conductivity of an annealed sample is of the order of 102 S/cm. Thedevice performance is determined by measuring merit factors of TFT. The saturation mobility ofmagnitude 18.5 cm2V�1 s�1 has been determined for W/L (20/10) device at 15 V drain bias. Theextrapolated field effect mobility for a device with channel width (W) 10 mm is 19.3 cm2V�1 s�1. The on/off current ratio is 109 and threshold voltage is in the range between 2 and 3 V. The role of annealing onthe electronic property of a-IGZO is carried out using X-ray photoelectron spectroscopy (XPS). The va-lance band cut-off has been approximately shifted to higher binding energy by 1 eV relative to as-deposited sample.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

Recently, Hosono et al. and H.-H. Hsu et al. have demonstratedthe use of IGZO for the flexible display [1,2]. The applicability andease of fabricating a-IGZO at room temperature has attractedattention of many researchers. Several studies have been carriedout to improve the performance of TFT. The field effect mobility invarious range between 10 and 68.5 cm2V�1 s�1 has been reportedby several authors [2e17]. The various experimental conditionssuch as RF power and co-sputtering [5,10,18], oxygen partial pres-sure during deposition [12,19], gate oxides [1e6,9,11e14], annealing[20e23], substrate types [5,6,24] and device dimension[5e9,15,16,25e27] have been explored to improve the performanceof TFT. It is to be realized that the carrier concentration has beencontrolled by varying oxygen's partial pressure during deposition inalmost all the studies.

The device dimensions, are another important parameter thatneeds to be considered as it plays an important role in improving

obade).

the resolution of display. Inmany of the reported investigations, thedimensions are too big [5e8]. To improve it, short channel deviceshave been demonstrated. However, the variation in the field effectmobility is marginal [9,15,16]. Kim et al. has reported a noticeableperformance for small dimensional device. The high mobility of themagnitude of 35.8 cm2 V�1 s�1 has been reported for W/L (10/50)mm in [3]. However, the additional process step to form etchstopper has been employed to improve the characteristics. Thehighest field effect mobility of the magnitude of 68.5 cm2 V�1 s�1

has been reported for a-IGZO TFTs using the SiOx passivation andthermal-annealing treatment [10].

The performance of TFT has also been measured in term ofsaturation mobility as well. Moreover, the saturation mobility hasbeen observed to be significantly smaller than the field effectmobility [28]. The saturation mobility in the range between 1 and26.66 cm2 V�1 s�1 has been determined in various studies [28e34].

Summarizing the reported literature, reasonably good perfor-mance has been obtained for extremely higher channel length andwidth of TFT. Few exceptional studies have been carried out for asmall dimensional device including short channel devices as well.The channel length of 6 mm, 10 mm and 50 nm devices have also

Fig. 1. Schematic with dimension of exposed area to ambient for annealing thefabricated bottom gate transistor.

R. Kamal et al. / Current Applied Physics 15 (2015) 648e653 649

been demonstrated in [9,15,16]. However, in those investigation,the field effect mobility exhibits in the range between 7 and12 cm2 V�1 s�1. The high field effect mobility has been obtained fordevice using etch stopper of SiO2 [3], which is due to additionalprocess step in fabrication. Although, device dimension is relativelysmaller, additional of process makes it unattractive. Thus, it is verybeneficial to establish a simple process to fabricate a TFT with goodperformance. In this investigation, there is no addition of step, suchas forming a etch stopper. It is very simple and usual process. Thechannel region carrier concentration is exclusively altered byannealing in forming gas and device dimension is still small. Inpresence of some important literature, the effect of annealing informing gas ambient on electrical performance of a-IGZO TFT andvalance band alignment on SiNx along with mechanism has yet tobe studied.

In this study, the high performance transistor with very goodsaturation mobility, the on/off current ratio, and reasonable subthreshold swing for relatively smaller devices are proposed. Thefabrication by usual process and control carrier concentration bypost annealing are examined.

2. Experimental

The devices were fabricated on glass with dimensions25� 25 mm2 using usual lithography process. The gate electrode ofMo (100 nm) was deposited using DC magnetron sputter and laterwas patterned out with channel width of 10, 20, 50 and 100 mm andlength 10 mm followed by SiNx deposition at 300 �C using PlasmaEnhanced Chemical Vapour Deposition (PECVD). The active layer ofa-IGZO of 50 nmwas deposited using 80 W RF power. The ambientof Ar:O2 in the ratio of 60:40 was maintained at 5mTorr workingpressure. For forming drain and source, 100 nm Mo was againdeposited using DC magnetron sputtering and the electrodes wereformed using wet etching process. The device was then annealed at350 �C for an hour in forming gas (N2þ 5%H2) ambient. The transfercharacteristics were measured using main frame semiconductoranalyzer E5270B. The four probe conductivity measurements atroom temperature of as-deposited and annealed samples werecarried out. The valance band spectra for both the sample weremeasured using XPS. The valance band spectra for three different a-IGZO thicknesses on SiNx 200 nm thick layer were also recorded.

The variation in the concentration of different cations betweenas-deposited and annealed sample suggests that the carrier con-centration can be different in channel region and Source/Drain (S/D) region, as S/D is covered by 100 nm thickMo and 20 mm inwidth.The ratio of concentration for various cations is provided in Table 1.On the basis of cations ratio (Ga/Zn) and In/(In þ Ga þ Zn), theconcentration is approximately determined in the range between7 � 1019 cm�3 and 1 � 1019 cm�3 and the Hall mobility is in therange between 9 and 20 cm2 V�1 s�1 [21].

The schematic of fabricated exposed area and covered area with100 nm Mo metal is presented in Fig. 1 and the ideal flat bandenergy diagram of TFT is presented in Fig. 2.

Table 1Ratio of concentration for various cations.

SampleSiNx(100 nm)/IGZO

Ga/Zn(1.06) [21]

Ga/(In þ Ga þ Zn)(0.19) [21]

In/(InþGaþZn)(0.62) [21]

As-dep (50 nm) 1.30 0.37 0.34Annealed (50 nm) 0.96 0.30 0.38As-dep (7e8 nm) 1.18 0.37 0.33As-dep (21 nm) 1.24 0.37 0.33

3. Results and discussion

In the bottom gate TFT, the semiconducting active layer isdeposited on the gate insulator. Very identical structures of filmsare deposited for XPS analysis. However, the IGZO thickness hasbeen varied and valance band spectra for three different sampleshave been recorded. The valance spectra for the as-depositedsample (50 nm), annealed sample (50 nm), as deposited sampleswith thickness 7, 21 and 50 nm IGZO on SiNx are provided in Fig. 3(a) and (b). The variation in the valance band cut-off is clearlyvisible. In addition, the difference in the cut-off for the as-depositedand annealed sample is closed to 1 eV. Thus, it suggests thatannealing is required to form distinct and complete valance band.The knee in valance band spectra for as-deposited sample has alsobeen transformed to well define valance band on annealing. It hasnot been observed for 7 nm and 21 nm as-deposited samples. It hasbeen suggested that O2p band in UPS spectra appears in energyrange 3 eVe10 eV. However due to lower cross-sectional area, thesepeaks may not be seen in XPS spectra. On the other hand XPS peakfor vacancy appears in spectra [35e37]. Thus, the shift in valenceband cut-off seems to be due to the charged oxygen and vacancy atoxygen site which results in the mid gap energy level which isvisible in XPS spectra. On annealing, the unreacted oxygen or trapdensity decrease and the band cut-off shifted to the value of bulkIGZO. The variation in the valence band cut-off with differentchannel thickness is observed for as-deposited samples as shown inFig. 3(b). The dependence of complete valence band formation ona-IGZO thickness was being investigated. However, no clear trendhas been seen. The valance band of IGZO is built of oxygen andhence spectra for O (1s) peak has been analyzed for all the foursamples. The reference of C (1s peak at 284.6 eV) is provided asinset in Fig. 4(a). The detail spectra for O (1s) for as-deposited 7, 21and 50 nm a-IGZO and the sample 50 nm a-IGZO on annealing areprovided in Fig. 4(a, b and c). The two distinct peaks have beenobserved for all the samples. Although, for the case of 50 nmsamples, the DBE is equal to 1.6 eV, the nature of peak on annealingis distinguishable. These two peaks in case of IGZO can be attrib-uted two kinds of oxygen surrounding. The higher binding energypeak attributed to (Ga/Zn)eO binding [38], while the lower bindingenergy peak is associated with IneO binding [39].

The four probe conductivity of a-IGZO on annealing in air and inreducing ambient are observed 3 � 10�3 S/cm and 5 � 102 S/cm,respectively. The defect reaction as shown in Eq. (1) occurs inreducing ambient leading to higher carrier concentration.

Ox ¼ 12O2 þ V

00 þ 2e� (1)

Further, the electron mobility is also a function of carrier con-centration. The fabricated devices have also been annealed informing gas ambient. The active region of TFT is L þ 40 mm leaving

Fig. 2. Ideal Flat band Energy Diagram of TFT.

Fig. 3. (a) Valance band spectra for as-deposited IGZO on 200 nm SiNx and IGZO annealed in forming gas IGZO on 200 nm SiNx; (b) Valance band spectra for three differentthickness of as-deposited IGZO on 200 nm SiNx.

R. Kamal et al. / Current Applied Physics 15 (2015) 648e653650

20 mm on both sides covered by 100 nm thick Mo contacts. Thechannel region only exposed to ambient while other part of activeregion is covered by 100 nm Mo. The schematic of TFT device isalready provided in Fig. 1. Thus, the annealing can possibly alter thecarrier concentration in channel region. Secondly, Ga/Zn ratio andIn/In þ Ga þ Zn ratio also exhibits variation on annealing. Asdescribed, the covered area is less likely to lose oxygen relative toexposed area; the gradient of carrier concentration is likely to exist.The conductivity difference of the order of 5 has been observed forthe samples prepared using identical condition but annealed in twodifferent ambient. The sample which has been annealed inreducing ambient exhibits very high conductivity. Considering thescattering of carrier when charge carrier concentration is high, theincrease in the mobility that can change the conductivity has beenruled out. Thus, the only suitable reasoning to explain the increasein conductivity is the increase in charge carrier concentration. Theloss of oxygen from the IGZO can be compensated by increase inelectron concentration.

It has also been reported that the forming gas annealing in-creases the Hall mobility of GZO film [40]. Hall mobility of thesample in forming gas is higher than that for the as-deposited,annealed in N2 ambient and O2 ambient. Identical argument hasbeen extended to explain the effect of thermal annealing on IGZO.

The electrical performance of fabricated device is discussed infollowing text. As-fabricated device have not shown switching

characteristic. The devices annealed in forming gas exhibit excel-lent transfer curve and are analyzed below. The saturation mobilityand other performance factors of TFT has also been obtained fromsquare law as shown in Eq. (2),

Id ¼ WLd

mFEε0ε

�Vgs � Vth

�22

(2)

where, W and L are channel width and length and d is thickness ofgate insulator, respectively, ε and εo are permittivity of gate insu-lator and free space, mFE is the mobility and Vth is threshold voltage.The sub threshold swing is extracted using Eq. (3),

S ¼�

dVGS

dlogðIDSÞ�

(3)

The density of interface trap are extracted from the Eq. (4) givenbelow [13].

Nt ¼�S logðeÞkT=q

� 1�

Ciq2

(4)

The IDS-VGS curves for devices for various channel width,representative IDS-VGS for W/L (100 mm/10 mm), are shown in Fig 5(aeb). The on/off current ratio determined is in the range of 108 to109. The off and on current has been chosen at �5 V and 30 V gate

Fig. 4. XPS spectra for oxygen peak for as-deposited and annealed sample (a) Peak for annealed IGZO 50 nm; (b) as-deposited IGZO 50 nm; (c) as-deposited IGZO 7 nm and as-deposited IGZO 21 nm.

Fig. 5. (a) Transfer curves as a function channel width at applied drain bias of 5 V; (b) transfer curve for device with W/L (100 mm/10 mm), device at three different drain bias.

R. Kamal et al. / Current Applied Physics 15 (2015) 648e653 651

bias respectively.The transfer curve at 5 V drain bias shows marginal threshold

voltage variation in the range between 3 and 4 V while the sub-threshold swing decreases with channel width. Further, from thetransfer curve for device with 100 mm/10 mm device, the thresholdvoltage appears to be decreasing with applied drain bias. This kindof behavior is undesired. The drain induced barrier lowering (DIBL)effect is prominent. In case of MOSFET, it is observed for shorterchannel length. Shorter the channel length more the drain voltagemodulates the source and substrate (a-IGZO) barrier. The off cur-rent is also increasing as drain bias increases.

The saturation mobility obtained using Eq. (2) are plotted as afunction of channel width in Fig. 6. The saturation mobility exhibits

decreasing trend with channel width. The field effect mobilityinduced by trans-conductance increase with channel length and ithas been suggested that contact resistance plays a role [3]. Thereported saturation mobility has been lower than that of field effectmobility. The difference in the saturation mobility and trans-conductance induced filed effect mobility is approximately 4times for IGZO TFT [28]. In the reported literature, the trans-conductance induced mobility in the range between 10 and 35.8(cm2 V�1 s�1) has been reported [3,5e9,11e17]. However, satura-tion mobility cannot directly be compare with filed effect mobility.The saturation mobility of about 18.5 cm2 V�1 s�1 is still a higherthan most of the reported values [28,30,31,33].

The drift component of the drain current is given by

Fig. 6. Saturation mobility as a function of channel width (W) at 15 V drain bias, insetshows saturation mobility for drain bias 5, 10 and 15 V.

R. Kamal et al. / Current Applied Physics 15 (2015) 648e653652

Id ¼ Wmeff CoxL

ðVGS � VT ÞVDS (5)

The field effect mobility can be obtained using Eq. 6

mFE ¼ LgmWCoxVDS

(6)

Sometimes, for a device, ID-VD curves are used for calculatingmobility. The drain current for device in saturation can be obtainedusing Eq. 7

ID;sat ¼BWmnCox

2LðVGS � VTÞ2 (7)

where B is the body effect, and is usually assumed to be unity. Themobility obtained using Eq. (7) which is considered as saturationmobility is shown in Eq. 8

msat ¼2Lm2

BWCox(8)

where m is the slope of the ðID;satÞ1=2 versus (VGS-VT) plot. Thesaturation mobility value obtained from Eq. (8) is usually lowerthan meff because the gate voltage dependence of the mobility is

Fig. 7. Device merit factors a function of channel width and various drain bias (a) thresholdratio and sub-threshold swing with channel width.

neglected in the Eq. (8). The saturation mobility is only valid whenthe drain current is governed bymobility, not by velocity saturation[41]. Further, it has been suggested that the electron mobility de-creases with the increase in carrier concentration. In this case, morecarrier concentration is expected in the accumulation region withincrease in the gate voltage. Similar argument [42] can be extendedto explain the decrease in saturation mobility.

The maximum saturation mobility of the magnitude of18.5 cm2 V�1 s�1 has been observed for the device with 20/10 mm.As channel width increases, the saturation mobility decrease andfor W/L (100/10) mm device and it is reduced by 10%. It has beensuggested that the spread of channel width and length leads to overestimation of mobility [16]. However, the change in the saturationmobility with channel width (W) using VDS ¼ 15 V is insignificant.Thus, the contribution of fringing current seems to be less. Thevariation in mobility as a function of channel width suggests thatthe contact resistance is increasing with channel width. Secondly,the conductivity of IGZO on annealing in air ambient is 105 timeslower than that for the sample annealed in forming gas ambient. Asprovided in Fig. 1, the covered area by 100 nm Mo blocks theambient. Thus it can be assumed that the conductivity of contactarea is much lower and spreading channel laterally can beinsignificant.

The device merit factor as a function of channel width at variousapplied drain bias are shown in Fig. 7 (a) and (b). Considering themerit factor, the device performance is excellent. The thresholdvoltage for at VDS ¼ 5 V is in the range between 3 and 4 V for all thedevices. The on/off current ratio varies in the range between 108

and 109. The off current is approximately 10�13 A. The performanceof TFT and DIBL effect suggests that short channel effect isresponsible for such behavior.

4. Conclusions

The effect of forming gas annealing on device performance hasbeen demonstrated. Controlling the carrier concentration by postannealing can be achieved by annealing time and temperature.Thermal annealing seems to be required for band structure for-mation in a-IGZO. The annealing also exhibits variation in atomicconcentration which possibly seems to contributing by evapo-rating unreacted metallic species. The high performance transistorwith reasonably high saturation mobility by creating carrier con-centration gradient across channel and contact has beendemonstrated.

voltage and off current as a function of channel width (b) the variation of on/off current

R. Kamal et al. / Current Applied Physics 15 (2015) 648e653 653

References

[1] Kenji Nomura, Hiromichi Ohta, Akihiro Takagi, Toshio Kamiya,Masahiro Hirano, Hideo Hosono, Nature 432 (2004) 488e492.

[2] Hsiao-Hsuan Hsu, Chun-Yen Chang, Chun-Hu Cheng, Curr. Appl. Phys. 13(2013) 1459e1462.

[3] Minkyu Kim, Jong Han Jeong, Hun Jung Lee, Tae Kyung Ahn, Hyun Soo Shin,Jin-Seong Park, Jae Kyeong Jeong, Yeon-Gon Mo, Hye Dong Kim, Appl. Phys.Lett. 90 (2007) 212114.

[4] Jim-Long Her, Tung-Ming Pan, Jiang-Hung Liu, Hong-Jun Wang, Ching-Hung Chen, Keiichi Koyama, Thin Solid Films 569 (2014) 6e9.

[5] Jae Kyeong Jeong, Jong Han Jeong, Hui Won Yang, Jin-Seong Park, Yeon-Gon Mo, Hye Dong Kim, Appl. Phys. Lett. 91 (2007) 113505.

[6] Yeon-Keon Moon, Sih Lee, Woong-Sun Kim, Byung-Woo Kang, Chang-Oh Jeong, Dong-Hoon Lee, Jong-Wan Park, Appl. Phys. Lett. 95 (2009) 013507.

[7] Jong Han Jeong, Hui Won Yang, Jin-Seong Park, Jae Kyeong Jeong, Yeon-Gon Mo, Hye Dong Kim, Jaewon Song, Cheol Seong Hwang, Electrochem. Soc.Solid State Lett. 11 (6) (2008) H157.

[8] Byung Du Ahn, Woong Hee Jeong, Hyun Soo Shin, Dong Lim Kim, Hyun JaeKim, Jae Kyeong Jeong, Sung-Hwan Choi, Min-Koo Han, Electrochem. Soc.Solid State Lett. 12 (12) (2009) H430eH432.

[9] Wantae Lim, Seon Hoo Kim, Yu-Lin Wang, J.W. Lee, D.P. Norton, S.J. Pearton,F. Ren, I.I. Kravchenko, J. Electrochem. Soc. 155 (6) (2008) H383.

[10] Ching-Ting Lee, Yung-Hao Lin, Mu-Min Chang, Hsin-Ying Lee, J. Disp. Technol.10 (2014) 293e298.

[11] Santosh M. Bobade, Ji-Hoon Shin, Young-Je Cho, Jung-Sun You, Duck-Kyun Choi, Appl. Sur. Sci. 255e17 (2009) 7831e7833.

[12] Hisato Yabuta, Masafumi Sano, Katsumi Abe, Toshiaki Aiba, Tohru Den,Hideya Kumomi, Kenji Nomura, Toshio Kamiya, Hideo Hosono, Appl. Phys.Lett. 89 (2006) 112123.

[13] Hideya Kumomi, Kenji Nomura, Toshio Kamiya, Hideo Hosono, Thin SolidFilms 516 (2008) 1516.

[14] H.Na Jong, M. Kitamura, Y. Arakawa, Appl. Phys. Lett. 93 (2008) 063501.[15] Ihun Song, Sunil Kim, Huaxiang Yin, Chang Jung Kim, Jaechul Park,

Sangwook Kim, Hyuk Soon Choi, Eunha Lee, Youngsoo Park, IEEE Elect. Dev.29 (6) (2008) 549e552.

[16] Jae Kyeong Jeong, Hyun-Joong Chung, Yeon-Gon Mo, Hye Dong Kim,J. Electrochem. Soc. 155 (11) (2008) H873eH877.

[17] K. Nomura, H. Ohta, K. Ueda, T. Kamiya, M. Hirano, H. Hosono, Science 300(5623) (2003) 1269e1272.

[18] H.J. Chang, K.M. Huang, C.H. Chu, S.F. Chen, T.H. Huang, M.C. Wu, ECS Trans. 28(4) (2010) 137e148.

[19] Jiangbo Chen, Li Wang, Xueqiong Su, Rongping Wang, J. Non-Crystalline Solids358 (2012) 2466e2469.

[20] Hyun-Woo Park, Min-Jun Choi, Yongcheol Jo, Kwun-Bum Chung, Appl. Sur.Sci. 321 (2014) 520e524.

[21] Hideo Hosono, Kenji Nomura, Youichi Ogo, Tomoya Uruga, Toshio Kamiya,J. Non-Cryst. Sol. 354 (2008) 2796e2800.

[22] Yuanjie Li, Zilong Liu, Kai Jiang, Xiaofen Hu, J. Non-Crystalline Solids 378(2013) 50e54.

[23] Huang-Kai Lin, Liang-Yu Su, Chin-Chin Hung, Jian Jang Huang, Thin Solid Films540 (2013) 247e250.

[24] Chi-Yuan Lee, Chienliu Chang, Wen-Pin Shih, Ching-Liang Dai, Thin SolidFilms 518 (2010) 3992e3998.

[25] Y. Li, Y.L. Pei, R.Q. Hu, Z.M. Chen, Y. Zhao, Z. Shen, B.F. Fan, J. Liang, G. Wang,Curr. Appl. Phys. 14 (2014) 941e945.

[26] Jae-Sung Kim, Min-Kyu Joo, Ming Xing Piao, Seung-Eon Ahn, Yong-Hee Choi,Junhong Na, Minju Shin, Man-Joong Han, Ho-Kyun Jang, Gyu-Tae Kima, ThinSolid Films 558 (2014) 279e282.

[27] Edward Namkyu Cho, Jung Han Kang, Ilgu Yun, Curr. Appl. Phys. 11 (2011)1015e1019.

[28] Yeon-Keon Moon, Sih Lee, Woong-Sun Kim, Byung-Woo Kang, Chang-Oh Jeong, Dong-Hoon Lee, Jong-Wan Park, Appl. Phys. Lett. 95 (2009) 013507.

[29] Jun Li, Fan Zhou, Hua-Ping Lin, Wen-Qing Zhu, Jian-Hua Zhang, Xue-Yin Jiang,Zhi-Lin Zhang, Curr. Appl. Phys. 12 (2012) 1288e1291.

[30] Hisato Yabuta, Masafumi Sano, Katsumi Abe, Toshiaki Aiba, Tohru Den,Hideya Kumomi, Kenji Nomura, Toshio Kamiya, Hideo Hosono, Appl. Phys.Lett. 89 (2006) 112123.

[31] Wantae Lim, Jung Hun Jang, S.-H. Kim, D.P. Norton, V. Craciun, S.J. Pearton,F. Ren, H. Shen, Appl. Phys. Lett. 93 (2008) 082102.

[32] G.Z. Geng, G.X. Liu, F.K. Shan, A. Liu, Q. Zhang, W.J. Lee, B.C. Shin, H.Z. Wu, Curr.Appl. Phys. 14 (2014) S2eS6.

[33] Byung Du Ahn, Hyun Soo Shin, Hyun Jae Kim, Jin-Seong Park, JaeKyeong Jeong, Appl. Phys. Lett. 93 (2008) 203506.

[34] Hideo Hosono, J. Non-Crystalline Solids 352 (2006) 851.[35] Yan Chan, Woo Chul Jung, Zhuhua Cai, Jae Jin Kim, Harry L. Tuller, Bilge Yildiz,

Energy Environ. Sci. 5 (2012) 7979e7988.[36] S.A. Leontiev, S.V. Koshcheev, V.G. Devyatov, A.E. Cherkashin, E.P. Milkheeva,

J. Struct. Chem. 38 (No. 5) (1997).[37] Deok-Yong Cho, Jaewon Song, Cheol Seong Hwang, Woo Seok Choi, T.W. Noh,

J.-Y. Kim, H.-G. Lee, B.-G. Park, S.-Y. Cho, S.-J. Oh, Jong Han Jeong, JaeKyeong Jeong, Yeon-Gon Mo, Thin Solid Films 518 (4) (2009) 1079e1081.

[38] Yuske Mizokawa, J. Elect. Spectrosc. Relat. Phen 14 (1978) 129e141.[39] A. Gurlo, M. Ivanovskaya, A. Pfau, U. Weimar, W. Gopel, Thin Solid Films 307

(1997) 288e293.[40] Byung Du Ahn, Sang Hoon Oh, Choong Hee Lee, Gun Hee Kim, Hyun Jae Kim,

Sang Yeol Lee, J. Cryst. Growth 309 (2007) 128e133.[41] Dieter K. Schroder, Semiconductor Material and Device Characterization, third

ed., Wiley Interscience, New Jersey, 2006, p. 502.[42] Siegfried Selberherr, Analysis and Simulation of Semiconductor Devices,

Springer-Verlag/Wien, New York, 1984, p. 91.