Controllable spin-polarized electrical transport in wide-band-gap oxide ferromagnetic semiconductors

JOURNAL OF APPLIED PHYSICS 107, 033713 共2010兲

Controllable spin-polarized electrical transport in wide-band-gap oxide ferromagnetic semiconductors Y. F. Tian,1,2 Shi-shen Yan,1,a兲 M. W. Zhao,1 Y. Y. Dai,1 Y. P. Zhang,1 R. M. Qiao,1 S. J. Hu,1 Y. X. Chen,1 G. L. Liu,1 L. M. Mei,1 Y. Qiang,2 and J. Jiao3 1

School of Physics, and National Key Laboratory of Crystal Materials, Shandong University, Jinan, Shandong 250100, People’s Republic of China 2 Department of Physics, University of Idaho, Moscow, Idaho 83844-0903, USA 3 Department of Physics, Portland State University, Portland, Oregon 97207-0751, USA

共Received 9 October 2009; accepted 5 January 2010; published online 8 February 2010兲 A family of wide-band-gap ternary oxide ferromagnetic semiconductor films with high transition metal concentration was prepared. The resistivity of these films can be changed up to four orders of magnitude by varying the composition or the concentration of the oxygen vacancies. Moreover, all these films show common features in electrical transport, i.e., Mott variable range hopping 共VRH兲 in the lower resistivity range, Efros VRH in the middle resistivity range, and “hard gap” resistance in the higher resistivity range. The above phenomena are well understood by considering the relative magnitude of three characterization lengths, i.e., Coulomb screening length, localization length of the carriers, and optimal hopping distance. Furthermore, spin polarization ratio of these magnetic semiconductors was obtained by fitting the experimental results of electrical transport. Therefore, the wide gap oxide ferromagnetic semiconductors with controllable spin-polarized electrical transport are expected to have application in spintronics devices as a spin injection source. © 2010 American Institute of Physics. 关doi:10.1063/1.3305457兴 I. INTRODUCTION

Wide-band-gap magnetic semiconductors attract increasing attention not only because they enable the integration of magnetism into existing semiconductor devices but also because they are transparent for any light at visible wavelength, which make them good candidates for both spintronics devices and transparent optical devices.1 By replacing some of the cations of parent nonmagnetic semiconductors 共such as GaAs, ZnO, and TiO2, etc.兲 with transition metal 共TM兲 atoms 共such as Mn, Fe, and Co, etc.兲, ferromagnetic semiconductors 共FMSs兲 can be synthesized. The exciting progress in FMS is that room temperature ferromagnetism has been achieved in many materials, such as Ti1−xCoxO2,2–4 Zn1−xTMxO,5,6 and so on. Besides the high Curie temperature and spin polarization, tunable resistivity in magnetic semiconductors is beneficial for resistivity matching in spin injection junctions. In this sense, FMS based on wide-bandgap oxide hosts 共such as ZnO, TiO2, and In2O3, etc.兲 are fascinating candidates as spin injection source if their resistivity can be tuned in a wide range by controlling the concentration of oxygen vacancies7–9 which supply the carriers. However, comparing various experimental results from different research groups, it is noticed that the magnetic properties and electrical transport of FMS are very sensitive to not only the concentration, distribution, and valence of the doped TM elements but also the defect density which were associated with the synthesis methods and postprocesses.

Therefore, a systematical study on a series of wide-band-gap oxide FMS which have controllable magnetic and electrical transport properties is highly desirable. The magnetoresistance10–13 in oxide FMS have also been studied for better understanding electrical transport of spinpolarized carriers. It was found that not only the Coulomb interactions between the charges of the electrons but also the exchange interactions between the spins of the electrons play an important role in the electrical transport of oxide FMS with spin-polarized carriers.13 Besides the Coulomb interactions and exchange interactions, other electrical and magnetic energies, such as “hard gap” energy, correlation energy, Zeeman energy, and spin-orbital coupling energy, are also expected to have significant contributions to the electrical transport. However, a universal relation between the electrical transport mechanisms and various interactions is still not clear in the oxide FMS. Here, we report a family of wide-band-gap ternary oxide FMS films which may be suitable as spin injection source, since their resistivity can be changed up to four orders of magnitude by varying the composition or the concentration of the oxygen vacancies. Moreover, All these films show common features in electrical transport, i.e., Mott variable range hopping 共VRH兲 in the lower resistivity range, Efros VRH in the middle resistivity range, and hard gap resistance in the higher resistivity range. Furthermore, spin polarization ratio of the oxide magnetic semiconductors was obtained by fitting the experimental results of electrical transport. II. EXPERIMENTAL

a兲

Authors to whom correspondence should be addressed: Electronic address: [email protected]. Tel.: ⫹86-531-88375097. FAX: ⫹86-53188377032.

0021-8979/2010/107共3兲/033713/5/$30.00

All the ferromagnetic oxide semiconductors studied here have similar nominal structure 关TM x Å / oxide y Å兴z,

107, 033713-1

© 2010 American Institute of Physics

033713-2

J. Appl. Phys. 107, 033713 共2010兲

Tian et al.

TABLE I. Detailed sample information, including composition, the thickness of TM 共x兲 and oxide semiconductor 共y兲 during each period, and the periods. It should be pointed out that all the samples were prepared under the remnant oxygen atmosphere, except 共In0.26Co0.74兲2O3−V in Fig. 3/Fig. 5共b兲 and Fig. 4/Fig. 5共c兲, which were respectively prepared under oxygen partial pressure ratio of 0.096% and 0.136% for the total pressure of Ar and O2 fixed at 1 Pa.

Sample composition 共In0.39Fe0.61兲2O3−V 共In0.31Fe0.69兲2O3−V 共In0.26Co0.74兲2O3−V 共Al0.21Co0.79兲2O3−V Zn0.18Co0.82O1−v 共In0.26Co0.74兲2O3−V Zn0.52Co0.48O1−v Zn0.30Co0.70O1−v Zn0.42Fe0.58O1−v Ti0.24Co0.76O2−v Zn0.30Fe0.70O1−v 共Al0.39Co0.61兲2O3−V Sn0.34Co0.66O2−v 共Al0.34Co0.66兲2O3−V 共In0.26Co0.74兲2O3−V Sn0.25Co0.75O2−v 共Al0.28Co0.72兲2O3−V Zn0.14Fe0.86O1−v Zn0.23Fe0.77O1−v Zn0.31Co0.69O1−v

TM Oxide thickness x thickness y 共Å兲 共Å兲 Periods z 5.0 5.0 4.5 6.6 5.2 4.5 5.2 5.2 5.0 6.0 5.0 4.4 5.0 4.4 4.5 5.0 4.4 5.0 5.0 5.0

9.1 6.5 5.0 3.3 2.4 5.0 12.0 4.8 7.8 5.0 5.0 5.5 7.8 4.4 5.0 5.0 3.3 1.6 3.2 4.8

60 60 60 60 60 60 60 60 30 60 30 60 60 60 60 60 60 30 30 60

Position 共top-to-bottom兲 Figure 2 Figure 2/Fig. 1共a兲 Figure 2/Fig. 5共a兲 Figure 2 Figure 2 Figure 3/Fig. 5共b兲 Figure 3 Figure 3 Figure 3 Figure 3/Fig. 1共b兲 Figure 3 Figure 4 Figure 4 Figure 4 Figure 4/Fig. 5共c兲 Figure 4 Figure 4 Figure 4 Figure 4 Figure 6

which were prepared by alternatively sputtering very thin 共x = 2 – 6 Å兲 TM layers and 共y = 2 – 20 Å兲 oxide layers for z 共z = 30– 60兲 periods under precisely controlled mixtures of argon and oxygen at room temperature on water cooled glass substrates 共20 ° C兲. Detailed sample information can be found in Table I. During all the deposition process, the total pressure of Ar and O2 was fixed at 1 Pa. It should be noticed that the minimum oxygen partial pressure was the remnant oxygen atmosphere in the vacuum chamber 共1.0⫻ 10−5 Pa兲. Unless noted otherwise, the samples were deposited under Ar gas with the remnant oxygen atmosphere. Due to the atomic interdiffusion, the nominal multilayered structures formed single layer films of metastable ternary oxides with high TM concentration and controlled oxygen vacancies. Microstructures of these films have been characterized by x-ray diffraction and high resolution transmission electron microscopy,14 which indicate that the as-deposited films are nanocrystalline or amorphous without any detectable TM clusters. The magnetic properties were measured by superconducting quantum interference device and ferromagnetism above room temperature was observed. The electrical transport measurements were carefully performed on a Van der Pauw configuration using a Keithley 2400 as current source and a Keithley 2182 as voltage detector. In Fig. 1 the magnetoresistance and magnetic hysteresis loops of 共In0.31Fe0.69兲2O3−v and Ti0.24Co0.76O2−v films are shown as an example of the studied wide-band-gap materials. The saturation magnetization of sample reduces very slowly with increasing temperature. For example, the magnetization of Ti0.24Co0.76O2−v is

FIG. 1. Magnetization loops and magnetoresistance measured at 5 K for 共a兲 共In0.31Fe0.69兲2O3−v and 共b兲 Ti0.24Co0.76O2−v FMS films, respectively.

524 emu/ cm3 共1.29␮B / Co兲 at 5 K and is 503 emu/ cm3 共1.24␮B / Co兲 at room temperature. The room temperature coercivity of studied samples is around several tens of oersted. The magnetoresistance shows obvious magnetic hysteresis behavior which corresponds to the magnetic hysteresis loops. The resistance peaks are located at the magnetic coercivity, and the resistance shows a trend to saturation when the magnetization approaches saturation with increasing magnetic field. This clearly clarifies that the negative magnetoresistance is due to spin-dependent electrical transport.

III. RESULTS AND DISCUSSION A. Electrical transport properties

Figures 2–4 show the temperature dependence of resistivity 共in logarithmic scale兲 for various oxide FMS films measured without a magnetic field. In Fig. 2, the resistivity 共in the range of 5 ⫻ 10−5 – 2 ⫻ 10−3 ⍀ m兲 of 共In1−xFex兲2O3−v, 共In1−xCox兲2O3−v, 共Al1−xCox兲2O3−v, and Zn1−xCoxO1−v 共x value is different for different samples兲 oxide semiconductor films is relatively small, and a linear relationship between ln ␳ and T−1/4 was found in the low temperature range. This behavior is typical of Mott VRH 共Ref. 15兲 which means that there exists a constant density of states at the Fermi level due to the negligible Coulomb interactions between carriers. For the 共In1−xCox兲2O3−v, Zn1−xCoxO1−v, Zn1−xFexO1−v, and Ti1−xCoxO2−v oxide FMS films, as shown in Fig. 3, the resistivity 共in the range of 10−3 – 2 ⫻ 10−1 ⍀ m兲 is in a middle range as compared with experimental results in Figs. 2 and 4. In contrast with the linear relationship between ln ␳ and T−1/4 observed in Fig. 2, a linear relationship between ln ␳ and T−1/2 was found in the low temperature range, as shown in Fig. 3. This behavior is typical of Efros VRH 共Ref.

033713-3

Tian et al.

FIG. 2. 共Color online兲 Resistivity in logarithmic scale vs T−1/4 for 共In1−xFex兲2O3−v, 共In1−xCox兲2O3−v, 共Al1−xCox兲2O3−v, and Zn1−xCoxO1−v oxide FMS films. A linear relationship between ln ␳ and T−1/4 in low temperature range was guided by the solid straight lines which were well described by Eq. 共3兲 using EH = 0 and m = 3.

16兲 which means that there exists a “soft gap” density of states near the Fermi level due to the electron-electron Coulomb interaction between carriers. Figures 4共a兲 and 4共b兲 show the temperature dependence of resistivity for 共Al1−xCox兲2O3−v, Sn1−xCoxO2−v, 共In1−xCox兲2O3−v, and Zn1−xFexO1−v films. As compared with the data in Figs. 2 and 3, the resistivity in Fig. 4 shows the highest experimental values in the range of 10−3 – 5 ⫻ 101 ⍀ m. In this case, a hard gap resistance with linear relationship between ln ␳ and T−1 was found in the low temperature range, as shown in Fig. 4共a兲, while Efros VRH was observed in the relatively high temperature range as shown in Fig. 4共b兲 for the same samples. The complex electrical transport properties indicate that not only electron-electron

FIG. 3. 共Color online兲 Resistivity in logarithmic scale vs T−1/2 for 共In1−xCox兲2O3−v, Zn1−xCoxO1−v, Zn1−xFexO1−v, and Ti1−xCoxO2−v oxide FMS films. A linear relationship between ln ␳ and T−1/2 was guided by the solid straight lines which were well described by Eq. 共3兲 using EH = 0 and m = 1. The 共In0.26Co0.74兲2O3−v sample here was deposited under oxygen partial pressure ratio 0.096% for the total pressure of Ar and O2 fixed at 1 Pa.

J. Appl. Phys. 107, 033713 共2010兲

FIG. 4. 共Color online兲 The temperature dependence of resistivity for 共Al1−xCox兲2O3−v, Sn1−xCoxO2−v, 共In1−xCox兲2O3−v, and Zn1−xFexO1−v FMS films. Resistivity in logarithmic scale vs 共a兲 T−1 and 共b兲 T−1/2 for the same samples. A linear relationship between ln ␳ and T−n was guided by the solid straight lines which were well described by Eq. 共3兲 using 共a兲 TM = 0 and 共b兲 EH = 0 and m = 1. The 共In0.26Co0.74兲2O3−v sample here was deposited under oxygen partial pressure ratio of 0.136% for the total pressure of Ar and O2 fixed at 1 Pa.

Coulomb interaction between carriers but also the hard gap energy due to many-electron excitation17 plays an important role in the electrical transport. It is worth a mention that although the 共In0.26Co0.74兲2O3−v samples in Figs. 2–4 were deposited under the same conditions except the oxygen partial pressure 共they have the same In and Co compositions but different oxygen vacancy concentrations兲, they showed quite different resistivity and transport mechanisms. Figure 5 is a set of R-T curves of the 共In0.26Co0.74兲2O3−v samples which have been shown in Figs. 2–4 and redrawn to demonstrate the effect of oxygen partial pressure on the resistivity and transport mechanisms. It is clear that the resistivity quickly increases with increasing oxygen partial pressure during deposition. Correspondingly, the electrical transport mechanisms change from Mott VRH through Efros VRH to hard gap hopping with increasing the resistivity. A reasonable explanation for this effect is that high oxygen partial pressure resulted in low carrier concentration in the samples and hence high resistivity. On the other hand, as shown in Figs. 2–4, various samples with different compositions show the same transport mechanisms if their resistivity is in the same range though microstructures of these samples are expected to change from one material to another, from one composition to another, and from one oxygen partial pressure to another. This clearly indicates that transport mechanisms are controlled by the concentration of the oxygen vacancies rather than the

033713-4

J. Appl. Phys. 107, 033713 共2010兲

Tian et al.

FIG. 6. 共Color online兲 Resistivity in logarithmic scale vs T−1/2 for the Zn0.31Co0.69O1−v sample measured at zero and 5 T magnetic field. Solid straight lines are theoretical fittings to experimental results by Eq. 共3兲 with ␳0 = 3.550⫻ 10−4 ⍀ m, TM 共0兲 = 170.94 K, and ␳共5T兲 = 3.140⫻ 10−4 ⍀ m, T M 共5T兲 = 158.56 K.

␧ = EH +

FIG. 5. Resistivity in logarithmic scale vs T−n for 共In0.26Co0.74兲2O3−v FMS films deposited under different oxygen partial pressure ratio 共a兲 0% 共remnant oxygen atmosphere兲, 共b兲 0.096%, and 共c兲 0.136%, respectively. Solid straight lines are theoretical fittings to experimental results by Eq. 共3兲 using 共a兲 EH = 0, m = 3, ␳0 = 5.19⫻ 10−5 ⍀ m, and T M = 28.19 K; 共b兲 EH = 0, m = 1, ␳0 = 2.72⫻ 10−4 ⍀ m, and TM = 353.36 K; and 共c兲 TM = 0, ␳0 = 4.22 ⫻ 10−3 ⍀ m, and EH / kB = 75.99 K, respectively.

B. Theoretical analysis and discussions

In the wide-band-gap FMS with high TM concentration, oxygen vacancies play a key role in ferromagnetic origin and electrical transport. The calculation on Fe doped In2O3 FMS with oxygen vacancies indicates that the spin of weakly localized s, p carriers of oxygen vacancies 共mainly from partial Fe 3d and Fe 4s electrons trapped in oxygen vacancies兲 are coupled parallel to the minor 3d spin of its nearest localized TM dopant 共Fe兲 due to strong exchange coupling.18 As a result, the carriers should be polarized and can be affected by not only the local electrical potential but also the local magnetic potential. Electrical transport is established via spindependent VRH between different localized states. The resistivity of spin-dependent VRH is determined by the energy difference ␧ between the initial occupied i state and the final vacant j state of the hopping process, which is related to the effective interactions between carriers in the oxide FMS system. We assume that the total energy difference ␧ between the initial occupied i state and the final vacant j state of a hopping process in the distance r takes the following form

共1兲

In Eq. 共1兲, EH term is a constant hard gap energy, C / rm term describes the electrical interaction 共such as m = 1 represents Coulomb interaction and m = 3 represents correlation energy兲, J cos ␪ / rm term represents magnetic interaction 共such as exchange interaction兲, C and J are two constant coefficients, m is an integer, and ␪ is the angle between the spin Si in the initial state and S j in the final state of a hopping process. Similar to Ref. 13, we can derive the optimal hopping distance r = rh at a certain temperature T and minimize the hopping resistivity ␳, which can be described as the following: rh =

concentration of TM elements and the type of oxides, although these two factors can contribute by influencing the concentration of the oxygen vacancies.

C J cos ␪ − . rm rm

␳=

冋

册 冋 冉 冊 册

m␰共C − J cos ␪兲 2kBT

1/共m+1兲

␳0 EH 具T M 典 exp + 1 + P 具cos ␪典 k BT T 2

共2兲

,

1/m+1

,

共3兲

where 具TM 典 = 共m + 1 / m兲m+1关m2m共C − J具cos ␪典兲 / kB␰m兴. Here, ␳0 is a prefactor of the resistivity, P is the spin polarization ratio of carriers near the Fermi level, kB is the Boltzmann constant, ␰ is the localization length of carriers, and 具cos ␪典 means the average value of cos ␪. Figure 6 shows the temperature dependence of the resistivity for Zn0.31Co0.69O1−v sample measured at 0 and 5 T magnetic field. It can be seen that Efros VRH with linear relationship between ln ␳ and T−1/2, but different sloops and intercepts were observed for zero and saturation magnetic field. This means that the magnetic field does not change the transport mechanism but causes a negative magnetoresistance. Since Efros VRH is observed, the experimental results are simulated by Eq. 共3兲 using EH = 0 and m = 1, while 具cos ␪典 = 0 for zero magnetic field and 具cos ␪典 = 1 for the saturation magnetic field. Good agreement indicates that our theoretical model can well describe the rich and colorful transport behaviors observed in wide-band-gap oxide FMS. From the theoretical fitting intercepts, ␳0 = 3.550⫻ 10−4 ⍀ m 共zero field兲, ␳共5T兲 = ␳0 / 共1 + P2兲 = 3.140⫻ 10−4 ⍀ m 共saturation field兲, spin polarization ratio of carriers for the

033713-5

J. Appl. Phys. 107, 033713 共2010兲

Tian et al.

Zn0.31Co0.69O1−v is deduced to be 36.1%. In the same way, we obtained the spin polarization ratio of Ti0.24Co0.76O2−v is about 21.9%, and the spin polarization ratio of Zn1−xFexO1−v changes from 20.6% to 30.0% for different sample compositions. High spin polarization of Zn1−xCoxO1−v makes it a very promising material for spintronics applications. In fact, the experimental results shown in Figs. 2–4 can also be well explained by Eq. 共3兲. For example, as EH = 0 and m = 3, Eq. 共3兲 is reduced to Mott VRH to fit the experimental results in Fig. 2; as EH = 0 and m = 1, Eq. 共3兲 is reduced to Efros VRH to fit the experimental results in Figs. 3 and 4共b兲; as EH / kBT Ⰷ 共具T M 典 / T兲1/m+1, Eq. 共3兲 becomes the hard gap transport equation which can be used to describe Fig. 4共a兲. These different hopping mechanisms can be well understood by considering the three character lengths, i.e., the Coulomb screening length rs, the localization length ␰ of the carriers, and the optimal hopping distance rh. According to Ref. 19, rs = 兵␬kBT / 关e2共n0 + p0兲兴其1/2, where ␬ is the dielectric constant and n0 and p0 are electron and hole concentrations at thermal equilibrium, respectively. In Fig. 2, the film resistivity is very small and comparable to some “bad metal” materials which means that these samples have high carrier concentration. As the mobile carrier concentration is high and the temperature is low, the Coulomb screening length rs becomes very small. On the other hand, small fitting parameter T M 共T M ⬀ 1 / ␰3, rh ⬀ 冑4 ␰ / T for m = 3兲 in Fig. 5共a兲 means large localization length ␰ and large optimal hopping distance rh. As rs Ⰶ rh and rs Ⰶ ␰ are satisfied, the Coulomb screening effect is strong, and both Coulomb interaction and hard gap energy can be neglected. As a result, Mott VRH was observed in the low temperature range, as shown in Fig. 2. On the contrary, in Figs. 4共a兲 and 4共b兲, the high resistivity means low carrier concentration. For the case of low carrier concentration, the Coulomb screening length rs can become very large. Relatively large fitting parameter T M 共T M ⬀ 1 / ␰, rh ⬀ 冑␰ / T for m = 1兲 in Figs. 5共b兲 and 5共c兲 means relatively small localization length ␰ and short optimal hopping distance rh. Therefore, ␰ Ⰶ rs and rh Ⰶ rs can be satisfied and the Coulomb screening effect is negligible. At very low temperature, the optimal hopping distance can be larger than the localization length, i.e., ␰ ⬍ rh Ⰶ rs. In this case, not only electron-electron Coulomb interaction between carriers but also the hard gap due to many-electron excitation plays an important role in electrical transport. As a result, hard gap transport was observed in the low temperature range, as shown in Fig. 4共a兲. Figure 3 is similar to Fig. 4共b兲. At a relatively high temperature, or at the middle carrier concentration of the oxygen vacancies, the optimal hopping distance may be shorter than the localization length, i.e., rh ⬍ ␰ Ⰶ rs. Since the hopping length is within the localization length, no other mobile carrier can be found to relax to produce hard gap energy but electron-electron Coulomb interaction plays the crucial role in the electrical transport. Correspondingly, Efros VRH was

observed in the relatively high temperature range, as shown in Fig. 4共b兲, or in the middle resistivity range at low temperature, as shown in Fig. 3. IV. CONCLUSIONS

In summary, tuning the resistivity of wide-band-gap ternary oxide FMS up to four orders by controlling the oxygen deficiencies enables their promising application as spin injection source. Universal electrical transport features were observed, i.e., Mott VRH in the lower resistivity range, Efros VRH in the middle resistivity range, and hard gap resistance in the higher resistivity range. These electrical transport phenomena are well understood by considering the relative magnitude of three characterization lengths, i.e., the Coulomb screening length, localization length of the carriers, and optimal hopping distance. Finally, the spin polarization ratio can be derived from simple electrical transport measurements, which greatly facilitates future spintronics design. ACKNOWLEDGMENTS

This work was supported by the National Basic Research Program of China 共Grant Nos. 2007CB924903 and 2009CB929202兲 and NSF 共Grant No. 10974120兲. H. Ohno, Science 291, 840 共2001兲. Y. Matsumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura, M. Kawasaki, P. Ahmet, T. Chikyow, S. Y. Koshihara, and H. Koinuma, Science 291, 854 共2001兲. 3 S. A. Chambers, S. Thevuthasan, R. F. C. Farrow, R. F. Marks, J. U. Thiele, L. Folks, M. G. Samant, A. J. Kellock, N. Ruzycki, D. L. Ederer, and U. Diebold, Appl. Phys. Lett. 79, 3467 共2001兲. 4 S. R. Shinde, S. B. Ogale, S. Das Sarma, J. R. Simpson, H. D. Drew, S. E. Lofland, C. Lanci, J. P. Buban, N. D. Browning, V. N. Kulkarni, J. Higgins, R. P. Sharma, R. L. Greene, and T. Venkatesan, Phys. Rev. B 67, 115211 共2003兲. 5 K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 79, 988 共2001兲. 6 M. Venkatesan, C. B. Fitzgerald, J. G. Lunney, and J. M. D. Coey, Phys. Rev. Lett. 93, 177206 共2004兲. 7 J. Philip, N. Theodoropoulou, G. Berera, J. S. Moodera, and B. Satpati, Appl. Phys. Lett. 85, 777 共2004兲. 8 Y. K. Yoo, Q. Xue, H. Lee, S. Cheng, X.-D. Xiang, G. F. Dionne, S. Xu, J. He, Y. S. Chu, S. D. Preite, S. E. Lofland, and I. Takeuchi, Appl. Phys. Lett. 86, 042506 共2005兲. 9 G. Peleckis, X. L. Wang, and S. X. Dou, Appl. Phys. Lett. 88, 132507 共2006兲. 10 A. G. Petukhov and M. Foygel, Phys. Rev. B 62, 520 共2000兲. 11 A. Janotti, D. Segev, and C. G. Van de Walle, Phys. Rev. B 74, 045202 共2006兲. 12 G. Peleckis, X. L. Wang, S. X. Dou, P. Munroe, J. Ding, and B. Lee, J. Appl. Phys. 103, 07D113 共2008兲. 13 Y. F. Tian, S. S. Yan, Y. P. Zhang, H. Q. Song, G. Ji, G. L. Liu, Y. X. Chen, L. M. Mei, J. P. Liu, B. Altuncevahir, and V. Chakka, J. Appl. Phys. 100, 103901 共2006兲. 14 H. Q. Song, L. M. Mei, S. S. Yan, X. L. Ma, J. P. Liu, Y. Wang, and Z. Zhang, J. Appl. Phys. 99, 123903 共2006兲. 15 N. F. Mott, J. Non-Cryst. Solids 1, 1 共1968兲. 16 A. L. Efros and B. I. Shklovskii, J. Phys. C 8, L49 共1975兲. 17 R. Chicon, M. Ortuño, and M. Pollak, Phys. Rev. B 37, 10520 共1988兲. 18 S. J. Hu, S. S. Yan, X. L. Lin, X. X. Yao, Y. X. Chen, G. L. Liu, and L. M. Mei, Appl. Phys. Lett. 91, 262514 共2007兲. 19 Y. D. Kwon, Phys. Rev. B 73, 165210 共2006兲. 1 2