JOURNAL OF APPLIED PHYSICS 103, 123514 共2008兲
Photoluminescence study of AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 crystals grown by the horizontal Bridgman technique Yunlong Cui,1,a兲 Utpal N. Roy,1 Arnold Burger,1 and Jonathan T. Goldstein2 1
Physics Department, Fisk University, Nashville, Tennessee 37208, USA Air Force Research Laboratory, Wright Patterson Air Force Base (WPAFB), Ohio 45433, USA
2
共Received 24 January 2008; accepted 13 April 2008; published online 18 June 2008兲 AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 single crystals grown by the horizontal Bridgman technique were investigated using photoluminescence 共PL兲 at temperatures varied from 8 to 300 K. For the AgGaSe2 crystals, free exciton 共FE兲, exciton bound to neutral donor 共D0, X兲, and edge emissions including donor-acceptor pair 共DAP兲 and free electron to neutral acceptor 共e, A0兲 transitions were observed. Two donor levels with binding energies of 18 and 39 meV and two acceptor levels with 61 and 117 meV were observed. The FE peak positions of the AgGaSe2 were found to be blueshifted when the samples were illuminated with higher laser intensity. This behavior was more pronounced at higher temperature when the peaks were also significantly broadened. For the AgGa0.9In0.1Se2 crystals, three DAP emission peaks at 1.673, 1.570, and 1.545 eV were observed at 8 K. The excitonic peaks were not observed below 100 K because they were overshadowed by the 1.673 eV DAP emission. For the AgGa0.8In0.2Se2 crystal, the excitonic peak was barely resolved in the PL spectra at 9 K, and only two shallow defect levels were shown. The temperature coefficients of the band-gap energies of the crystals were measured. The thermal expansion effect of the AgGa0.8In0.2Se2 crystal was found to be much larger than that of AgGaSe2. The PL study showed that the AgGa0.8In0.2Se2 crystal had advantages over the AgGa0.9In0.1Se2 crystal for their use as potential radiation detectors. © 2008 American Institute of Physics. 关DOI: 10.1063/1.2940739兴 I. INTRODUCTION
AgGaSe2 and AgGaxIn1−xSe2 crystals have been studied as promising materials for applications in nonlinear optics, solar cells, and spin-polarized electron sources.1–8 Improvements in the efficiency of frequency doubling of CO2 lasers have been pursued for over three decades.1 Nowadays, efficient generations of noncritically phase-matched second- and even third-harmonic frequencies of CO2 lasers have been realized with AgGa0.6In0.4Se2 共Refs. 2 and 3兲 and 共Ref. 4兲 crystals, respectively. AgGa0.53In0.47Se2 AgGa0.8In0.2Se2 thin films have been investigated as effective absorber materials for the top cells of tandem solar cells, and a total-area conversion efficiency of 7.3% was reported.6 AgGaxIn1−xSe2 crystals could serve as potential candidates for room-temperature x-ray and gamma radiation detectors.9 First of all, the band-gap energy of the quaternary chalcopyrite compound is tunable from 1.24 to 1.82 eV at room temperature by gradually substituting indium with gallium from x = 0 to x = 1. Second, the crystals grown by the horizontal Bridgman technique are highly insulating.10,11 Last but not least, the constituents each have a high atomic number, and the electron mobilities of the compounds are comparable to those of Cd0.9Zn0.1Te crystals. In spite of the practical importance of the AgGaxIn1−xSe2 crystals, many of their fundamental properties are still not well understood. One of the poorly understood issues is the nature of the defects. It is generally accepted that the compensation of the intrinsic defects 共vacancy, interstitial, and antisite兲 within AgGaSe2 crystals dominates the conductivity a兲
Electronic mail:
[email protected].
0021-8979/2008/103共12兲/123514/6/$23.00
of their semi-insulating materials because the effects of doping by introduction of different external dopants into those crystals have insignificant impacts on their resistivities. The identification of the defect levels in the crystals is essential for their crystal quality control, electronic device fabrication, and operation. It is well known that the carrier mobility– lifetime product is a figure of merit to evaluate a semiconductor detector. The carrier lifetimes are fully limited by both intrinsic and extrinsic defect traps. The photoluminescence 共PL兲 technique is a sensitive tool to identify the defect type, characterize the crystal quality, and measure the semiconductor band-gap energy. Few investigations concerning the PL and defect levels of AgGaSe2 and AgGaxIn1−xSe2 crystals have been reported.5,10,12–17 However, substantial progress, both theoretical and experimental, has been made on the identification of the defect levels in other chalcopyrite compounds, such as CuInSe2 and CuGaSe2.18–24 Although the shallow donor levels predicted by the theoretical calculations are deeper than those from the experimental results, the con− 2+ 2− cept that defect complexes such as 共2VCu + Gacu 兲 and 共CuGa 2+ + GaCu兲 could more easily being created than the point defects produced separately in CuGaSe2 crystals shedding lights on our defect assignments. In a previous paper, we reported the AgGaSe2 crystal growth and composition using low-temperature PL and Raman spectroscopies.16 The composition homogeneity of the crystals along the crystal growth direction has been realized. It is expected that the substitution of gallium with indium can add more defects and hence provide some insights on the Ga-related defects. In this article, we report having grown AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 crystals, collected their PL spectra at differ-
103, 123514-1
© 2008 American Institute of Physics
Downloaded 04 Sep 2008 to 134.131.125.49. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
123514-2
Cui et al.
J. Appl. Phys. 103, 123514 共2008兲
ent conditions by varying the excitation laser intensities and temperatures, measured the band-gap energies, and gave assignments to the defect levels.
II. EXPERIMENT
The AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 crystals were grown by the horizontal Bridgman technique. The starting materials were synthesized from stoichiometric amounts of 99.9999% Ag, 99.99999% Ga, 99.9999% In, and 99.9999% Se. Synthesized materials were loaded in prebaked high-purity pyrolitic boron nitride crucibles and sealed in quartz ampoules under high vacuum. Growth runs were carried out in a three-zone transparent gold-coated furnace, and the direction of solidification was realized by translating the furnace at a rate of 5 mm/day. The grown crystals were about 15 cm long, 1 cm wide, and 7 mm thick. Each single crystalline ingot was cut and polished for the PL measurements. The polishing of the samples was carried out with successive finer grit-sized polishing pads, and the final polishings were carried out with 0.05 m size alumina on felt pad. The polycrystalline parts located in the first to freeze section of the ingots were excluded from the measurements. Before the PL measurements, the crystals were characterized with a scanning electron microscopy 共SEM兲 and an energy dispersive x-ray spectroscopy 共EDS兲 using a JEOL JSM5310LV system coupled with a NORAN Voyager system. SEM/EDS results showed that the crystals for the PL measurements were slightly Ga-rich and Ag-poor. The PL measurements were carried out in two separate systems, a Jobin-Yvon LabRam-INFINITY micro-Raman system and a SPEX 1877D Triplemate spectrometer. The PL spectra collected at temperatures in the range of 70–300 K were recorded in the micro-Raman system with the backscattering geometry in confocal configuration. A 632.8 nm He–Ne laser was used as the light source. For the temperature variation of the PL measurements, the samples were mounted on a Raman microprobe compatible with a MMR microminiature refrigeration system coupled to a K-20 programmable temperature controller. The laser was focused on the sample with a 10⫻ objective 关numerical aperture 共NA兲 ⫽0.25兴, the chosen laser power was 0.1 mW with intensity of 200 W / cm2. At room temperature, the laser was focused on the sample with a 100⫻ objective 共NA= 0.9兲, the applying power was 0.01 mW and the light intensity was 250 W / cm2. For the PL measurements at temperatures from 8 to 300 K, the samples were cooled down using an APD Cryogenic, Inc. dual HC-4MK I helium compressors. The crystals were illuminated with the 488 nm line of an ILT 5500A air-cooled argon-ion laser with the power varied from 0.1 to 100 mW through neutral density filters. The laser spot diameter was about 1 mm. The PL spectra were detected using a Triplemate spectrometer in conjunction with a liquid nitrogen cooled charge coupled device detector. A 0.1 mm slit and a grating with 300 grooves/mm were employed for the spectrometer.
FIG. 1. Room-temperature PL spectra of AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 excited by a 633 nm He–Ne laser at intensity of 250 W / cm2
III. RESULTS AND DISCUSSION
The band-gap energy of a direct-band-gap semiconductor material at room temperature can be estimated by locating the exciton peak position in the room-temperature PL spectrum. Figure 1 shows typical room-temperature PL spectra of AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 crystals excited by a 633 nm He–Ne laser. It can be seen that the PL spectrum for each sample consists of a single asymmetric bell-shaped peak and that the peak slightly skews toward high energies. The peak position of the AgGaxIn1−xSe2 crystal shifts from 1.795 to 1.746 and 1.650 eV as the indium concentration x increases, respectively, from 0 to 0.1 and 0.2. If the binding energies of free excitons 共FEs兲 are assumed to be 20 meV,25 the room-temperature band gaps of the AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 crystals are estimated to be 1.82, 1.77, and 1.67 eV, respectively. The corresponding full widths at half maximum 共FWHMs兲 were measured as 29, 61, and 54 meV, respectively. The PL FWHM value of a crystal is a quality indicator of the crystal. It was expected that the main part of the FWHM, which reflected the compositional variation of AgGaxIn1−xSe2, would behave as those of other alloy semiconductors, increasing monotonically with x as x increases from 0 to 0.2.26,27 However, it can be seen that the excitonic peak of AgGa0.8In0.2Se2 is narrower than that of AgGa0.9In0.1Se2. Since the two spectra were collected at the same conditions, it may be concluded that the AgGa0.8In0.2Se2 crystal with the relative smaller value of the FWHM has better homogeneity than that of the AgGa0.9In0.1Se2 crystal. It is also worth noting that the room-temperature PL spectrum of the AgGa0.8In0.2Se2 crystal is surface treatment dependent. For a cleaved sample without surface polishing, a 1.41 eV surfacerelated defect peak always accompanies the excitonic emission and sometimes even dominates the PL spectra. In principle, it would be preferable to run a PL measurement at lower excitation intensity in order to reduce the laser
Downloaded 04 Sep 2008 to 134.131.125.49. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
123514-3
J. Appl. Phys. 103, 123514 共2008兲
Cui et al.
FIG. 2. Temperature dependence of the FE PL peaks of a AgGaSe2 crystal. The crystal was excited with 488 and 633 nm lasers at intensities of 2 and 200 W / cm2, respectively.
heating effect; however, weaker laser intensity may not only result in lower ratio of signal to noise but also will broaden the FWHM of a PL spectra of an alloy semiconductor, and finally will lead to underestimate its band-gap energy Eg.27 Figure 2 shows the typical temperature dependence of the FE PL peaks of the AgGaSe2 crystal. With two different setups, the crystal was excited with light of wavelengths of 488 and 633 nm at intensities of 2 and 200 W / cm2, respectively. It can be seen that as the temperatures varied from 70 to 300 K, the peak energy values measured from the higher intensity laser were always larger than those from the lower intensity laser. The higher the temperature, the larger the difference between the two corresponding values. However, the discrepancy at each temperature T was still less than kT / 2, where k is the Boltzmann constant. The FE peak position blueshift behavior at higher intensity was independent of the laser wavelength and the experimental setup. The behavior was not due to the heating effect since it would not be expected that a higher laser intensity would correspond to a smaller laser heating effect. If the heating effect would dominate the discrepancy, the FE exciton peak energies measured with the 2 W / cm2 laser should be higher than those with the 200 W / cm2 laser. As we have mentioned earlier in this paragraph, the discrepancy results from a difference between the FWHMs of the excitonic spectra collected from the two systems. In fact, the phenomenon that the excitonic line shifted to lower energies with decreasing excitation intensity was found and studied in Ga0.47In0.53As alloy semiconductors,27 in which the amount of peak energy shift was found to be proportional to the linewidth itself. Here, we take into account the phenomenon with the simply formula used in Ref. 14, where the FE emission peak energy distribution of an AgGaSe2 crystal was simplified to be asymmetric according to the formula
FIG. 3. AgGaSe2 PL spectra at different temperatures excited with a 488 nm argon-ion laser at intensity of 2 W / cm2
冉 冊 冋
I = C exp
册
共E − Ex兲2 Ex − E exp − . kT 22
共1兲
In which C is a constant, Ex is the FE energy, and is the standard energy deviation. The FE peak energy E p and the FE energy Ex are related as E p = Ex-2/kT.
共2兲
The energy deviation and the FWHM will be given as FWHM = 2.35 .
共3兲
The FWHM used here is oversimplified,27 but Eqs. 共2兲 and 共3兲 were adequate to describe our data. The FWHM obtained with the 488 nm laser excitation was larger than that with the 633 nm laser. For example, the FWHM at 296 K obtained with the 488 nm laser was 60 meV, twice as large as that obtained with the 633 nm laser 共29 meV兲. Therefore, the E p corresponding to a higher FWHM should have a lower value. One important reason for the PL spectrum to be preferentially collected below 10 K is to have a larger luminescence signal and a minimum thermal broadening effect. However, if one has to measure a spectrum of a quaternary compound at a higher temperature, Eg values obtained from different excitation intensities are expected to be different. It can be seen that the AgGaSe2 has anomalous temperature dependence since the temperature coefficients are positive below 80 K and negative above 100 K. The temperature coefficient is −1.3⫻ 10−4 eV/ K in the range of 115–300 K and 8.9⫻ 10−5 eV/ K in the low-temperature range. The values are in good agreement with those obtained from reflectivity measurements.25 Investigation of the temperature evolution of the PL spectra is helpful to identify defect levels because the intensities of different PL bands may decrease differently upon an increment of temperature. Typical PL spectra of the AgGaSe2 crystal at different temperatures are shown in Fig. 3. In the exciton emission region, peaks at 1.813 and 1.807 eV are resolved below 34 K. The 1.813 eV peak originates from the
Downloaded 04 Sep 2008 to 134.131.125.49. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
123514-4
J. Appl. Phys. 103, 123514 共2008兲
Cui et al.
FE peak, while the 1.807 eV peak is due to a donor bound exciton 共D0, X兲, while the energy required to free the bound exciton is 6 meV. The 共D0, X兲 peak merges with the FE peak above 59 K. The band-gap energy Eg at 8 K is 1.833 eV assuming that the FE binding energy is 20 meV.25 The donor activation energy ED can be estimated using the Haynes factor,28 which is given as 0.2 for a donor defined as the ratio between the energy required to free the bound exciton from a defect 关6 meV for 共D0, X兲兴 and the energy ED required to free the bound carrier from the same defect. Therefore, ED is estimated to be 30 meV. We can identify the donor defect as gallium-on-silver antisite GaAg. The defect was reported in Se-rich AgGaSe2 compound and identified as a SeAg antisite;5,29 however, there have been no further theoretical calculations and the experimental evidences to support the assignment. Four edge emissions peaked at 1.772, 1.756, 1.743, and 1.720 eV were observed. The 1.772 eV peak did not shift when the illumination intensity varied between 13 and 13 W / cm2. The peak seems not to be associated with a bound exciton either since the activation energy would be too high. It is reasonable to believe that the 1.772 eV peak is due to a free electron to a neutral acceptor 共e, A0兲 transition.13 The peak energy EeA at temperature T is given by EeA = Eg − EA + kT/2,
共4兲
where EA is the activation energy of an acceptor. It is clear that the 共e, A0兲 emission position is light intensity independent provided that the heating effect is negligible. EA is calculated to be 61 meV. The activation energy value of the acceptor is consistent with that reported earlier in Refs. 5 and 30. The defect was identified as one of cation vacancies, VGa or VAg.5 The donor-acceptor pair 共DAP兲 transition energy EDAP is light intensity dependent. EDAP is a function of the distance r between a charged acceptor and a charged donor, which is given as follows: EDAP = Eg − EA − ED + e2/共4r兲.
共5兲
As the excitation power decreases, the number of photoexcited DAPs will decrease and r will increase and the DAP emission peak will show a redshift. If the illumination power is kept infinitely low, the last term of Eq. 共5兲 will be negligible and EDAP will equal to Eg − EA − ED. The peaks located at 1.756, 1.743, and 1.720 eV are DAP emissions because they are all shifted toward lower energies 共to 1.754, 1.732, and 1.698 eV兲 as the illumination light intensity decreases from 2 to 13 mW/ cm2. The 1.743 eV peak is attributed to a transition of the 39 meV donor D which has been identified in the excitonic emission region and the 61 meV acceptor A; the 1.756 eV peak is due to a transition of the 18 meV D1 to 61 meV A; the 1.720 eV peak is due to a transition of D1 with the acceptor A1 located at 117 meV above the valence band. The shallow acceptors with ionization energies of 60 and 100 meV and the shallow donor with energy of 12 meV had been found in CuGaSe2 crystals,22,23 and it seems that the shallow defect levels correspond to those in AgGaSe2 crystals. Effective-mass theory predicted three candidates for
FIG. 4. Light intensity dependence of the PL of a AgGa0.9In0.1Se2 crystal. The crystal was excited with a 488 nm argon-ion laser and cooled to 8 K.
shallow donors in AgGaSe2,17 silver interstitial Agi, gallium antisite GaAg, and selenium vacancy Vse. Among the donors’ activation energies, the level Agi might be the shallowest, followed by the first 共ground state level兲 and the second donor levels of GaAg. First-principles calculations also showed that the activation energy of VCu of the CuInSe2 共CuGaSe2兲 compound has a minimum value, followed by the first activation energy levels of VIn共VGa兲 and CuIn共CuGa兲.19,20 Therefore, the 18 meV D1 donor is assigned to the Agi although the SEM/EDS result shows that the AgGaSe2 sample is slightly Ag-poor, the 39 meV D is to the first activation energy level of the GaAg antisite, and the 61 meV A and 117 meV A1 acceptors are assigned to the first level of the VGa vacancy and to the first level of the AgGa antisite, respectively. The PL spectra of AgGaxIn1−xSe2 crystals differed significantly from those of AgGaSe2, although indium and gallium belong to the same IIIA family. The amount of indium added to the alloy not only influences the band-gap energy but also alters the defect chemistry. Figure 4 shows a typical light intensity dependence of the PL spectra of an AgGa0.9In0.1Se2 crystal at 8 K. It can be seen that the peak positions of 1.673 and 1.570 eV peaks in the PL spectra with excitation intensity of 13 mW/ cm2 have a stronger light intensity dependency when compared to the peak at 1.545 eV. The three peaks are assigned to 共DAP兲1, 共DAP兲2, and 共DAP兲3, respectively. Neither free nor bound excitons were observed at this temperature. The failure to observe of excitons is probably due to exponentially tailed donor states introduced by the indium atoms.31 When temperature was increased up to 30 K, the 共DAP兲3 peak intensity decreased accordingly and finally disappeared. Above 80 K, the 共DAP兲1 peak disappeared, while the FE peak gradually appeared and dominated the PL spectra. Figure 5 shows the evolution of the PL spectra for temperatures between 100
Downloaded 04 Sep 2008 to 134.131.125.49. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
123514-5
Cui et al.
FIG. 5. Temperature evolution of the FE peaks of a AgGa0.9In0.1Se2 crystal. The crystal was excited with a 488 nm argon-ion laser at intensity of 13 W / cm2
and 188 K. It can be seen that the relative intensity of the FE to the 共DAP兲2 increased with increasing temperatures. The band-gap energy at 8 K cannot be derived directly from Fig. 4 due to the absence of a FE peak. However, because AgGa0.9In0.1Se2 and AgGaSe2 crystals have similar temperature coefficients above 125 K, it is reasonable to speculate that they have similar coefficients below 80 K, in which case, the band gap of AgGa0.9In0.1Se2 at 8 K can be estimated to be 1.77 eV. The corresponding EA + ED are estimated to be 0.10, 0.20 and 0.22 eV, respectively, for the 共DAP兲1, 共DAP兲2, and 共DAP兲3 emission bands. The 共DAP兲1 emission can be attributed to a transition between the first activation energy levels of D 共GaAg兲 and A 共VGa兲, while 共DAP兲2 is to the first level of A1 共AgGa兲 and a donor defect with activation energy around 80–100 meV. The defect is more like the second activation level of D 共GaAg兲, instead of the first level of Vse.17 The 共DAP兲3 may be the first longitudinal optical phonon replica of the 共DAP兲2 although the intensity of the 共DAP兲3 is higher than that of the 共DAP兲2 at a higher intensity laser illumination, as shown in Fig. 4. On the other hand, the 共DAP兲3 may also be attributed to a transition between the second activation energy levels of D 共GaAg兲 and A1 共AgGa兲. Excitation-intensity dependence of typical PL spectra of AgGa0.8In0.2Se2 crystal at 9 K is shown in Fig. 6. The FE emission bump located around 1.63 eV is barely resolved with illumination intensity at 2 W / cm2 and above. The PL spectra were almost featureless. Only two shallow defect peaks extracted from a curve fitting were found and identified as 共D0, h兲 and 共e, A0兲. The ED and EA were estimated to
J. Appl. Phys. 103, 123514 共2008兲
FIG. 6. Light intensity dependence of the PL peaks of a AgGa0.8In0.2Se2 crystal. The crystal was excited with a 488 nm argon-ion laser and cooled to 9 K.
be 30 and 60 meV, respectively. The two defects originate from D 共GaAg兲 and A 共VGa兲, which are defects identified in the AgGaSe2 crystals. Temperature dependences of the FE peak energies of AgGa0.9In0.1Se2 and AgGa0.8In0.2Se2 crystals are shown in Fig. 7. The temperature coefficients of the AgGa0.8In0.2Se2 crystal are −8.5⫻ 10−5 eV/ K in the range of 125–240 K and 5.9⫻ 10−4 eV/ K in the low-temperature range. The absolute value of the temperature coefficient between 125 and 240 K is smaller than that reported from photoconductivity and op-
FIG. 7. Temperature dependence of the FE emission peak energies of AgGa0.9In0.1Se2 and AgGa0.8In0.2Se2 crystals.
Downloaded 04 Sep 2008 to 134.131.125.49. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
123514-6
J. Appl. Phys. 103, 123514 共2008兲
Cui et al.
tical absorption measurements, while the value in the lowtemperature range is higher.32 For AgGa0.9In0.1Se2, the temperature coefficient is around −1.5⫻ 104eV/ K in the range of 125–40 K. The temperature dependence of the band-gap energy has contributions from two parts, as described by the equation below,33
冉 冊 冉 冊冉 冊
Eg dEg = dT T
+
V
Eg ln V
T
d ln V . dT
共6兲
The first and second terms of the equation are due to the electron-phonon interaction and the thermal expansion effect, respectively; the thermal expansion has a positive temperature coefficient, while the electron-phonon interaction has a negative one.25 By comparing the temperature coefficients of AgGaSe2 and AgGa0.8In0.2Se2, one can find that the thermal expansion effect increases significantly after substituting 20% of the gallium atoms with indium. The enhancement mechanism needs further investigation. It was reported that the thermal expansion effect in the AgGaS2 crystal could be curtailed dramatically by substituting as little as 1% Ag with Cu, and the soft-phonon mode was suggested to be responsible for the suppression effect.33 In summary, we have studied the temperature and laserexcitation-intensity dependences of the PL of AgGaSe2, AgGa0.9In0.1Se2, and AgGa0.8In0.2Se2 single crystals grown by the horizontal Bridgman technique. We identified the radiative defect levels and estimated the temperature coefficients of the band-gap energies. Since the AgGa0.8In0.2Se2 crystal has a narrower FWHM of PL at room temperature and less defect bands than that of the AgGa0.9In0.1Se2 crystal, it should be expected that the crystal will be a better candidate for room-temperature x-ray and gamma detector applications. ACKNOWLEDGMENTS
This work was supported by an AFOSR Grant No. F49620-01-1-0479. The authors at Fisk University gratefully acknowledge financial support from the NSF-supported Center of Research Excellence in Science and Technology 共CREST兲 共Cooperative Agreement No. CA-0420516兲. The authors thank Dr. Enrique Silberman for his valuable comments and suggestions.
1
R. L. Byer, M. M. Choy, R. L. Herbst, D. S. Chemla, and R. S. Feigelson, Appl. Phys. Lett. 24, 65 共1974兲. 2 G. C. Bhar, S. Das, D. V. Satyanarayan, P. K. Datta, U. Nundy, and Yu. N. Andreev, Opt. Lett. 15, 2057 共1995兲. 3 P. G. Schunemann, S. D. Setzler, and T. M. Pollak, J. Cryst. Growth 211, 257 共2000兲. 4 E. Takaoka and K. Kato, Opt. Lett. 24, 902 共1999兲. 5 K.-M. Nigge, F. P. Baumgartner, and E. Bucher, Sol. Energy Mater. Sol. Cells 43, 335 共1996兲. 6 K. Yamada, N. Hoshino, and T. Nakada, Sci. Technol. Adv. Mater. 7, 42 共2006兲. 7 I. H. Choi and P. Y. Yu, Appl. Phys. Lett. 87, 231909 共2005兲. 8 A. Janotti and S.-H. Wei, Appl. Phys. Lett. 81, 3957 共2002兲. 9 See, for example, G. F. Knoll, Radiation Detection and Measurement, 3rd ed. 共Wiley, New York, 1999兲. 10 B. Tell and H. M. Kasper, Phys. Rev. B 4, 4455 共1971兲. 11 U. N. Roy, Y. Cui, R. Hawrami, A. Burger, L. Orona, and J. T. Goldstein, Solid State Commun. 139, 527 共2006兲. 12 I.-H. Choi and P. Y. Yu, J. Phys. Chem. Solids 56, 595 共1995兲. 13 G. Masse and E. Redjai, J. Phys. Chem. Solids 47, 99 共1986兲. 14 M. C. Petcu, N. C. Giles, P. G. Schunemann, and T. M. Pollak, Phys. Status Solidi B 198, 881 共1996兲. 15 K. Yoshino, H. Komaki, K. Itani, S. F. Chichibu, Y. Akaki, and T. Ikari, J. Cryst. Growth 236, 257 共2002兲. 16 U. N. Roy, Y. Cui, R. Miles, A. Burger, J. T. Goldstein, Z. W. Bell, and D. A. Carpenter, J. Appl. Phys. 98, 093523 共2005兲. 17 R. Marquez and C. Rincon, Mater. Lett. 40, 66 共1999兲. 18 S. B. Zhang, S.-H. Wei, and A. Zunger, Phys. Rev. Lett. 78, 4059 共1997兲. 19 S.-H. Wei, S. B. Zhang, and A. Zunger, Appl. Phys. Lett. 72, 3199 共1998兲. 20 S. B. Zhang, S.-H. Wei, A. Zunger, and H. Kadayama-Yoshida, Phys. Rev. B 57, 9642 共1998兲. 21 C. Domain, S. Laribi, S. Taunier, and J. F. Guillemoles, J. Phys. Chem. Solids 64, 1657 共2003兲. 22 S. Siebentritt, I. Beckers, T. Riemann, J. Christen, A. Hoffmann, and M. Dworzak, Appl. Phys. Lett. 86, 091909 共2005兲. 23 A. Bauknecht, S. Siebentritt, J. Albert, and M. Ch. Lux-Steiner, J. Appl. Phys. 89, 4391 共2001兲. 24 G. L. Gu, B. H. Tseng, and H. L. Hwang, J. Phys. Chem. Solids 64, 1901 共2003兲. 25 L. Artus and Y. Bertrand, Solid State Commun. 61, 733 共1987兲. 26 K. Oettinger, D. M. Hofmann, Al. L. Efros, B. K. Meyer, M. Salk, and K. W. Benz, J. Appl. Phys. 71, 4523 共1992兲. 27 E. F. Schubert and W. T. Tsang, Phys. Rev. B 34, 2991 共1986兲. 28 J. R. Haynes, Phys. Rev. Lett. 4, 361 共1960兲. 29 H. L. Park, S. Y. Gee, S. K. Chang, S. Park, S. K. Lee, J. Y. Lee, C. S. Yoon, J. E. Kim, H. Y. Park, W. T. Kim, and H. Lim, Phys. Status Solidi B 162, K9 共1990兲. 30 H. Neumann, E. Nowak, B. Schumann, and G. Kuhn, Cryst. Res. Technol. 18, 477 共1983兲. 31 S. Ozaki and S. Adachi, J. Appl. Phys. 100, 113526 共2006兲. 32 S. Hahn and W. Kim, Phys. Rev. B 27, 5129 共1983兲. 33 I.-H. Choi and P. Y. Yu, Phys. Rev. B 63, 235210 共2001兲.
Downloaded 04 Sep 2008 to 134.131.125.49. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
