Compensation origins in II–VI CZT materials

Materials Science and Engineering B71 (2000) 297 – 300 www.elsevier.com/locate/mseb Compensation origins in II– VI CZT materials A. Zumbiehl a,*, S. Mergui b, M. Ayoub a, M. Hage-Ali a, A. Zerrai c, K. Cherkaoui c, G. Marrakchi c, Y. Darici d b a PHASE-CNRS, 23 Rue du Loess, BP 20, F-67037 Strasbourg Cedex, France Electrical Engineering Department, Florida International Uni6ersity, Miami, FL 33199, USA c L.P.M., INSA Lyon, Bat. 502, 20, a6 A. Einstein, F-69621 Villeurbanne, France d Physics Department, Florida International Uni6ersity, Miami, FL 33199, USA Abstract It is well known that II – VI CdTe, and CdZnTe (CZT) materials suffer from the presence of cadmium vacancies (V-Cd) and their complexes with impurities and defects, which lead to low resistivity and trapping. These defects are known generally as the A-centers, around 0.1– 0.2 eV. In order to increase the resistivity, intentional and non intentional chemical and physical compensations are conducted; for CdTe, halogens (Cl, Br, I) or In are generally used, while for CZT, the compensation origins are still unknown. In this paper, we try to study the compensation by measuring deep levels and very shallow levels at temperatures as low as helium temperature by photoluminescence (PL) and photoinduced current transient spectroscopy (PICTS), and model the resistivity in order to clarify the origins of the compensation and the high resistivity in CZT materials. © 2000 Published by Elsevier Science S.A. All rights reserved. Keywords: Resistivity; Simulation; Levels; Band; PICTS; Photoluminescence; CZT material 1. Introduction 2. Experimental Material characteristics of CdZnTe allow this material to be a very good choice for nuclear radiation detectors and photorefractive applications. Both applications involve the same need for a high resistivity material which is related to defect levels, the type of levels, activation energy and concentration. However, these parameters are not well known in CZT. Electronic spectrometry like photoinduced current transient spectroscopy (PICTS), thermostimulated current (TSC) and photoluminescence (PL) provide information about defect levels inside the gap, but do not give accurate values on the concentration of the defects. By using a compensation model, our measurements by PICTS, TSC and PL, and the existing results from the literature, we can adjust these parameters to fit the measured resistivity values, which make possible to study the behaviour and the nature of these defects levels. Samples of CdZnTe grown by the high pressure Bridgman method (HPB), from three different ingots were tested by PICTS, and PL at helium temperature. The samples investigated in this study were all the size of 10 × 10×2 mm3. The electrical resistivity is 1010 V cm. In principle, CZT grown by HPB has a high resistivity without intentional compensation. All samples were etched with a solution of 5% Br– methanol for 2 min and rinsed in methanol. Ohmic contacts were deposited by gold electroless. The PL optical excitation sources used for filling traps were an IR light-diode emitting at 0.95 mm and a He – Ne laser. The PICTS one is a laser diode at 1.51 eV. The experimental set-up for PL and PICTS are described elsewhere in the literature [1]. 3. Results * Corresponding author. Fax: +33-3-8810-6230. E-mail address: [email protected] (A. Zumbiehl) Fig. 1 shows typical photoluminescence spectrum at 8 K for a n-type material [2,3]. The main three energy levels were identified as P1 for D°X at Eg −13 meV, P2 0921-5107/00/$ - see front matter © 2000 Published by Elsevier Science S.A. All rights reserved. PII: S 0 9 2 1 - 5 1 0 7 ( 9 9 ) 0 0 3 9 4 - 3 298 A. Zumbiehl et al. / Materials Science and Engineering B71 (2000) 297–300 for A°X at Eg−24 meV (with a binding energy of Eg −13 meV) and P3 for the A center from 0.14 to 0.22 eV but generally described at 0.14 eV in the literature [2,4,5]. Fig. 2 shows PL spectra from samples from three different slices cut from top, middle and tail of the same ingot. The data show the variation of the peak energy positions, with the segregation of zinc throughout the ingot, which corresponds to the gap energy variation. Since we consider the valence band roughly constant, the transition energy remains the same for the three slices. The PICTS spectra obtained are shown on Fig. 3. Five main levels from P3 to P7 are shown. P3, generally attributed to complex cadmium vacancies VCd-D appears both in PL and PICTS data. In previous results, five levels have been found by PICTS and eight by PL in this P3 band, which exhibit an acceptor behaviour. P4 and P5 at, respectively, 0.31 and 0.49 eV, are reported as acceptors levels [6]. P7 at 1.05 eV is reported as a Fig. 1. PL spectrum for Cd0.9Zn0.1Te HPB 5/19 sample. Fig. 2. PL spectrum for CdZnTe HPB-19 samples (same ingot). A. Zumbiehl et al. / Materials Science and Engineering B71 (2000) 297–300 299 Fig. 3. PICTS spectrum for Cd0.9Zn0.1Te HPB 5/19 sample. Fig. 4. Simulated resistivity for Cd0.9Zn0.1Te HPB 5/19 sample with d =1015 and 1016 cm − 3. donor level and is found in vanadium, and other IV line elements doped materials. There is still a controversy about P6 at 0.9 eV, sometimes attributed as a donor level [7] while others [8] show the presence of deep acceptor levels at 0.6 eV by electron paramagnetic resonance (EPR) for Ge-doped materials. Therefore, we consider P6 to be an acceptor level in our compensation model. Since we have a n-type material with a shallow ND –NA of some importance, P6 as a deep acceptor level even with a low ionization, seems to be determinant in the compensation process [9]. 4. Discussion Level concentrations and energies are directly related to charge carrier concentrations, and therefore to resistivity. This is done by the neutrality condition: − p +% [N+ D ] =n +% [NA ] where n and p are electron and hole concentrations, − respectively, N+ D and NA are ionized donor/acceptor concentrations. The resistivity is described by the following expression [9]: A. Zumbiehl et al. / Materials Science and Engineering B71 (2000) 297–300 300 r= 1 q(nmn +pmp ) At first, it was assumed that compensation is due to the introduction of deep donor levels to compensate the shallow cited acceptors [10,11,7]. However, the simulation of the resistivity by using only the 0.2– 1.1 eV band, measured by PICTS and TSC methods, only down to the liquid nitrogen (LN) temperature, with the cited admitted hypothesis concerning the nature of theses levels, predicts the necessity of an additional non-existent donor band at 0.6– 0.7 eV to explain the high resistivity of the measured sample. The A-center position and the compensation processes have been subject to discussion for the EPR and optically detected magnetic resonance (ODMR) community [10,11,7,8]. Another hypothesis could be made: the high resistivity could be the result of the compensation of an intense very shallow donor levels P1 by shallow acceptor levels P2, P3 and P4 and a deep acceptor level P6. We should add that the deep donor level P7 at 1.05 eV has a negligible influence on the compensation model at this energy because of the poor ionisation. In our compensation model, we used two ways to achieve the resistivity calculation. At first, the concentration d= [P1] −([P2]+ [P3] +[P4]) was taken as 1015 cm − 3 and then 1016 cm.− 3 The results are summarised in Fig. 4, where resistivity is calculated as a function of d and the P6 concentration. The resistivity shows a drastic increase where [P6]= d. Then, there is a plateau over four decades of [P6] variations. In this region, it is possible to compensate within a large domain of [P6] concentrations, up to 1019 cm − 3. Compensation in a very narrow range could exist with another level like P4 or P5. It should be mentioned that the maximum resistivity is reached when nmn =pmp around 1017 cm − 3 for the first way and around 1018 cm − 3 for the second one. The material should be n-type just before the left side of the maximum and p-type on the right side. . 5. Conclusion We have presented another model using a deep acceptor level instead of purely deep donor level to compensate shallow levels (both acceptors and donors). Energy values were extracted from PICTS measure- ments for deep levels and from PL for shallow levels. Concentrations were extracted from our relative spectra shape, and the literature. We must keep in mind that the admitted value for the cadmium vacancies lay around 2.1016 cm − 3, especially with chlorine compensation [12]. Therefore all our values are in this range. The A-center level was taken as 1016 cm − 3 according to the results of Suzuki for his mobility measurements and modelling [5]. As mentioned before, shallow levels (both acceptors and donors) which are completely ionized are determining for the type and the amount of carrier to be compensated. For our n-type material, shallow donor level P1 gives the nature of the majority carriers at low temperature, as shown by PL [3]. Therefore, a simple deep acceptor level P6 up to 1019 cm − 3 is enough to compensate the carrier difference d. This calculation presents better correlation with the last EPR measurement and shows that we have in fact self-compensation in the material, because P6 is always in the same order of magnitude than the introduced P1 in spite of the other shallow levels. This can be an indication that P6 could be a cadmium vacancy related level. References [1] A. Zerrai, K. Cherkaoui, G. Marrakchi, G. Bremond, J. Cryst. Growth 197 (1999) 646. [2] K. Hjelt, M. Juvonen, T. Tuomi, S. Nenonen, E. Eissler, M. Bavdaz, Phys. Stat. Sol. (a) 162 (1997) 747. [3] H. Hermon, M. Schieber, R.B. James, et al., NIM A 410 (1) (1998) 100. [4] C.B. Norris, C.E. Barnes, Rev. Phys. Appl. 12 (1977) 219. [5] K. Suzuki, S. Seto, S. Dainaku et al., J. Electron. Mater. 25 (8) (1996) 12. [6] A. Castaldini, A. Cavallini, B. Fraboni, P. Fernadez, J. Piqueras, J. Appl. Phys. 83 (4) (1998) 2121. [7] M. Fierderle, C. Eiche, M. Salk, R. Schwarz, K.W. Benz, W. Stadler, D.M. Hofmann, B.K. Meyer, J. Appl. Phys. 84 (12) (1998) 6683. [8] H.J. Von Bardeleben, T. Arnoux, J.C. Launay, J. Cryst. Growth 197 (1999) 718. [9] A. Zumbiehl, P. Fougeres, M. Hage-Ali, J.M. Koebel, P. Siffert, A. Zerrai, K. Cherkaoui, G. Marrakchi, G. Bremond, J. Cryst. Growth 197 (1999) 670. [10] B.K. Meyer, D.M. Hofmann, W. Stadler, M. Salk, C. Eiche, K.W. Benz, Mater. Res. Soc. 602 (1993) 189. [11] D.M. Hofmann, W. Stadler, P. Christmann, B.K. Meyer, Nucl. Inst. Meth. Phys. Res. A 380 (1996) 117. [12] M. Ohmori, Y. Iwase, Mater. Sci. Eng. B16 (1993) 283.