Semi-Insulating Te-Saturated CdTe - IEEE Xplore

IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 52, NO. 5, OCTOBER 2005

1925

Semi-Insulating Te-Saturated CdTe R. Grill, J. Franc, P. Höschl, I. Turkevych, E. Belas, and P. Moravec

Abstract—The evolution of defect structure and self-compensation is theoretically studied within quasichemical formalism in undoped and donor-doped Te-saturated CdTe during the cooling in the temperature interval 700 C–100 C. We show, that proper thermal treatment, including low temperature (cca 200 C) dwell, allows to prepare semi-insulating CdTe with deep level doping below the limit 1013 cm 3 , which is demanded in detector industry. New high-temperature transport data are used to refine on previous native defect properties for the modeling. Variant defect models are analyzed. Diffusion rates at lowered temperature are annotated to approve the model for real-time experimental verification. Index Terms—CdTe, deep defect, detector, self-compensation.

I. INTRODUCTION

T

HE preparation of room temperature (RT) CdTe and Zn Te gamma- and x-ray detectors Cd requires to solve apparently contradictory demands of a low free carrier density, which has to be close to the intrinsic carrier cm in CdTe at RT, and the high carrier density . This goal cannot be reached mobility-lifetime product without significant reduction of trapping centers in the crystal [1], [2]. Even in the best quality materials synthesized from 7N purity elements the controllable density of shallow defects cm with the state-of-the-art does not decrease below technological procedures. The native defects density introduced into material during the crystal growth can hardly reach cm with lower value, too. The deep trap density N cm is necessary in order trapping cross section to avoid considerable loss of photo-generated carriers and the depreciation of the detector [3]. A straightforward procedure to figure out this task is based on the search of a deep defect with the ionization level near the middle of the band gap (midgap) and with cm . Such defect introduced into the material at the density N cm would compensate the shallow defects and pin the Fermi energy near the midgap without strong [4]–[7]. Recent experimental trapping and degradation of findings [8], however, do not support such a scheme and show of midgap levels in the cm range. Moreover, cm is asked [2]. The fact that reduced N even down to high quality detectors on CdTe exist [5], [9] points to some underestimated processes at the detector fabrication, which

Manuscript received November 11, 2004. This work was supported by the Grant Agency of the Czech Republic under Contract 202/02/0628. This work is a part of the research program MSM0021620834 financed by the Ministry of Education of the Czech Republic. The authors are with Charles University, Faculty of Mathematics and Physics, Institute of Physics, Prague 2, CZ-121 16, Czech Republic (e-mail: [email protected]). Digital Object Identifier 10.1109/TNS.2005.856801

result in some cases in a spontaneous formation of the detector grade material and fulfillment of the above mentioned demands. Defect statistics is mainly evaluated in thermodynamic equilibrium established by external conditions (temperature, pressure of one component) during the annealing [10]. Electric properties at RT are deduced assuming fast cooling from the annealing conditions and freeze-up of the high temperature defect structure [6]. This approach is, however, far from reality in bulk samples, where even the fastest quenching lasts at least tens second and the defect reactions, which proceed during the cooling, influence the defect structure significantly [11]. Consequently, the RT defect structure and carrier densities evaluated within the model of frozen high-temperature defect structure do not correspond to real experimentally observed RT quantities. Though the importance of defect reactions at low temperature to the compensating processes was addressed [10], [12] the particular studies of the formation of semi-insulating CdTe (SICT) with respect to minimized deep level doping were not reported yet. The evolution of defect structure during the cooling of Te-rich CdTe based on defect properties obtained by ab-initio calculations [10] is reported in [13]. The possibility to obtain SICT with cm as a consequence of strong self-compensation N and shallow donor Te anof shallow acceptor Cd vacancy V below 220 C is demonstrated there. The ab-initio tisite Te data [10] yield, however, the phase diagram of CdTe, which differs from experimental findings [14], [15], and their validity it thus questionable. In this paper we rely on experimental measurements of phase diagram of CdTe [14], [15] and base the model on defect parameters, which fit well both temperature-composition (T-x) phase diagram and high temperature transport data. We perform theoretical study of processes occurring in the crystal during the cooling at Te-saturation according common treatment at the detector fabrication. We use standard quasichemical formalism [16]–[18] and look for a system, which affords the possibility to obtain SICT also with reduced density of deep level below cm . We analyze sithe requested detector grade limit multaneously various defect models differing by ionization energies of V and Te . The formation of donor-V complex is included as well. The modeling is based on rectified native defect structure consistent with on new high-temperature transport data. This report follows the paper devoted to this topic [19], where similar findings as reported here are demonstrated on a simplified model of shallow divalent V at Te saturation and a compensating shallow extrinsic donor. II. THEORY The defect structure in CdTe is determined by the Fermi en, the Cd chemical potential , and by the chemical ergy

0018-9499/$20.00 © 2005 IEEE

1926


potential of the dopant. In equilibrium with ambient vapor is fixed by Cd pressure P . The deviation from stoichiometry can be tuned this way, where is the Te atom fraction. If Te excess reaches maximum Te solubility limited by stability region in T-x phase diagram [20], Te starts to precipitate from the Te-supersaturated CdTe matrix and stabilizes at the SLG (solid-liquid-gas) three-phase equilibrium. Such process can be expected always in Te-rich CdTe during the cooling of the solidified crystal due to the retrograde solubility of Te. We shall assume henceforth, that the effective P (Te-sat) experienced by crystal interior at Te satP uration is given by the SLG equilibrium. Using saturated vapor pressure over Te and Gibbs energy of formation of CdTe(s) according to [14], P (Te-sat) is expressed as P

Te

principal native defects. Subsequently, we shall analyze consequences of respective models to the detector fabrication. The principal defect reactions at Te saturation describe the escape of V and Te into Te precipitates Te V

Cd P

Cd Cd

Te

V

Cd

Te

Cd

Te

V

(4)

Cd Te

P

K

Te

eV

sat Te

Te

(5)

is the relevant formation energy related to corresponding deand entropy determine confect reaction, the energy tributions to the vibrational free energy. The parameter

K eV K

(3)

V

Te

K

Cd

V

eV

K

Cd

Cd

eV K

(Te-sat) in (1)

P

K

K

(2)

sat Cd

V

K

Te

The neutral defect densities are then set by P via the reactions and respective forms [17]

eV

sat atm

Te

(1) (6)

The Clausius-Clapeyron equation and thermodynamic data [21] were used to extrapolate P (Te-sat) below Te melting point (722.65 K), where no experimental data are published. Electric properties at RT are influenced by defect reactions, which proceed during the cooling, due to different electric characteristics (valence, ionization energies) of the input/output reacting defects. The final defect structure is determined by the at which the reactions have frozen. Generally temperature speaking, various defect reactions freeze at different temperatures and the final defect structure is influenced by the entire history of the crystal. The evolution of the defect structure is as a calculated here in a model assuming Cd vacancy V as a divalent donor, comdivalent acceptor, Te antisite Te pensating extrinsic monovalent donor (D) and its complex with Cd vacancy (A-center) V D as a monovalent acceptor. Both doping with shallow donors currently used for the detector fabrication (In , ClTe , Al ), and midgap-donor doping (Ge , Sn ) will be studied. Dominant native defects routinely accepted in CdTe are the Cd vacancy, the Cd interstitial Cd and the Te antisite [10], [20]. However, the density of Cd is negligible at Te saturation. The existence of Te is well-grounded in Te-rich CdTe [10], [20], but its electrical character has not been set yet. Also ionization energies of V were not fixed definitely and second ionization level of V both shallow and midgap is used in recent literature [10], [22]. Therefore, we shall extend our investigation to several defect models, which differ by ionization energies of

connects

to P

as P

(7)

cm is the Cd or Te atom density, is the is the mass of Cd atom. The label Boltzmann constant, and X is used for neutral defects overall. The densities of multiply ionized defects are calculated for acceptors and donors [17] E E

E E

(8) (9)

where E and E are the acceptor and donor one-electron ionand are the degeneracies of the deization energies. fects. The index means the ionization degree. The formation of A-centre acceptor from singly ionized V and D is described by A A

V

D V

V D

D E

(10)

(4) for A-center rewhere configuration degeneracy lated to donor in Cd (Te) sublattice and E is the A-centre formation energy related to the given defect reaction.

GRILL et al.: SEMI-INSULATING Te-SATURATED CdTe

1927

TABLE I BEST FIT DEFECT PARAMETERS. DONOR AND ACCEPTOR IONIZATION ENERGIES ARE GIVEN RELATIVELY TO RESPECTIVE BANDS E AND E . TWO ENERGIES FOR Cd CORRESPOND TO THE TWO TETRAHEDRAL INTERSTITIAL SITES. Te IS ASSUMED NEUTRAL IN THE BASIC MODEL. THE a,b,c,d,e LABELLED VALUES DEPICT DEFECT MODELS WITH VARIANT IONIZATION ENERGIES OF V OR Te . UNLABELLED VALUES ARE STEADY IN ALL MODELS

The Fermi energy trality condition V

is obtained solving the electric neu-

V

A

Cd Te

Cd

Te

D

(11)

The total density of donor D is fixed by N

D

D

A

A

(12)

In addition, we shall assume, that midgap level with low denexists, which pins after quenching to RT. The sity N electric character (donor or acceptor) of this low density level does not affect the final results. The only principal feature of the midgap level is its capability to accumulate the uncompensated charge of shallow levels, which is determined by the position of the Fermi energy relatively to the midgap level. The search for , is of its minimum density, which allows midgap pinning of the primary interest of this paper. Extrinsic acceptors are not taken into account in this paper. They oppose the formation of SICT and must be compensated by increased density of shallow donors. The minimization of the density of extrinsic acceptors is thus desirable.

Fig. 1. Experimental (open diamonds—500 C, upper triangles—600 C, squares—700 C, circles—800 C, bottom triangles—900 C and diamonds—1000 C) and theoretical 1=eR . The dash line P outlines the position of the p-n line which was obtained from the theoretical calculations based on the presented model. The dash-dotted line P shows the position of the line of congruent sublimation. The arrows show the Cd and Te saturation limit.

0

III. RESULTS AND DISCUSSION A. Defect Modeling Before we present results of the modeling outlined above, we shall report on some improvements of basic model parameters, which were done in comparison to our previous results [20]. New parameters given in Table I are based on completed set of high-temperature galvanomagnetic measurements at 1000 C. They differ only slightly from parameters published in [20]. New data could not be fitted with previously used gap energy [23]. Therefore, small correction had to be done. The E reads E

eV

(13)

where the defect-model-dependent parameter for model b and for other models a,c,d,e discussed further in the text. At the same time a better fit of new experimental data was achieved when the role of satellite L-minima on transport was weakened in comparison to our paper [24]. This implies, that the strong decrease of electron mobility at high temperature explained by inter-valley scattering should be caused at least partly by temperature dependence of dielectric constant, which was suggested as an alternative explanation of the observed effect [24]. Results of the calculations together with experimental points eR , where R is Hall coefficient, are presented in of

Fig. 2. Fit of the tellurium atom fraction in CdTe along the three-phase curve (solid line) together with relevant experiments [14] (full circle) and [15] (open square).

Fig. 1. New model describes well all experimental data obtained with Cd overpressure. Simultaneously, the T-x phase diagram has been fitted to experimental points as shown in Fig. 2 and Te was adjusted this way.

1928


Fig. 3. Native defect structure of five analyzed examples. Respective examples a,b,c,d,e are depicted by fitted values in Table I.

Fig. 5. RT hole density of CdTe at Te saturation doped by shallow donor [D] = cm and frozen at temperature . Labels correspond to examples of native defect structure depicted in Fig. 3. Dash line plots the RT electron density ). in example b where electrons dominate holes (

T

10

n >p

Fig. 4. RT hole density of undoped CdTe at Te saturation with defect structure frozen at temperature . Labels correspond to respective examples of native defect structure discussed in the text.

T

Unfortunately, the transport data at Te overpressure (left from in Fig. 1) are not of a good precision due to difficulties with obtaining experimental data and relatively large experimental error at Te-saturated conditions. Consequently, the defect model cannot be rectified in this respect and unambiguous assessment of defect structure cannot be made yet. We shall deal with such obstacle by simultaneous analysis of several examples, which characterize principal features of different native defect configurations. All examples fit well both transport data obtained at Cd in Fig. 1) and T-x diagram in Fig. 2. pressure (right from Characteristic defect structures of five analyzed examples are depicted schematically in Fig. 3. Corresponding model parameters are given in Table I and in (13). B. Cooling The RT hole density of undoped CdTe calculated for models outlined in Fig. 3 is plotted in Fig. 4. The dominant feawith lower , which is ture of all models is the decrease of caused by the reduction of native defect density due to precipitation. Significant deviations within studied defect models are observed at high . In case of midgap Te level (case c and e) on such level. SICT is formed at high due to pinning of The measurements of RT resistivity on undoped quenched CdTe could thus give principal information about native defect structure of CdTe. The effect of quenching velocity on formation of

SICT was reported in [23], unfortunately on samples with unsatisfactory purity. The result of case d differs from the calculation in [13], where similar defect structure was discussed but with ab-initio parameters [10], which are different from parameters in Table I based on experiments. Especially the formation eneV is significantly higher than 5.94 eV ergy of Te used in [13]. below Shallow donor doping has a dramatic effect to 400 C in all cases, see Fig. 5. The most striking effect is observed in the model b, where compensated second level of V pins in the midgap. In other models, the reduction of , which approaches to the semi-insulating state at low , is due E , which partly acts as deep to relatively deep E V level as shown in [19]. Fig. 6 demonstrates the effect of midgap level to the formation of SICT. Evidently, except model b, the low temperature behavior is not influenced by the respective model and SICT is C in all cases. formed at Similar result is obtained also in the case of included formation of A-center in Fig. 7. The plotted results were calculated eV, which was deduced from usual Coulomb for E model (see [16, Ch. 10]) for a donor in Cd sublattice (e.g. In ) E

(14)

The dielectric constant and D V distance in CdTe. Ionization energy E meV characterizing typical A-centre acceptor [25] was used. Small variations (few meV) of this ionization energy corresponding to the chemical nature of the involved donor do not affect reported results significantly. We can conclude that the formation of SICT is not strongly influenced by native defect structure within analyzed defect models. Semi-insulating CdTe is obtained by aging at C in all cases. Assuming donor in the Te sublattice, typically Cl , the shorter DTe V distance


Fig. 6. RT hole density of CdTe at Te saturation doped by shallow donor [D] = 10 cm and with midgap level density N = 10 cm frozen at temperature . Labels correspond to examples of native defect structure discussed in the text. Dash lines plot the RT electron density at temperatures ). where electrons dominate holes (

T

n >p

1929

Fig. 8. Carrier densities and neutral and charged defect densities calculated = 10 cm for defect model a in Fig. 3 with [D] = 10 cm and N without A-center formation. The dash line plots , which is identical with the line a in Fig. 6. The Fermi energy shown by dash-dotted line is depicted by the right axis.

p

Fig. 7. RT hole density of CdTe at Te saturation doped by shallow donor with midgap level density N = 10 cm and included [D] = 10 cm A-center formation and frozen at temperature . Labels correspond to examples of native defect structure discussed in the text.

Fig. 9. Carrier densities and neutral and charged defect densities calculated = 10 cm and for defect model e in Fig. 3 with [D] = 10 cm , N included A-center formation. The dash line plots , which is identical with the line e in Fig. 7. The Fermi energy shown by dash-dotted line is depicted by the right axis.

results in increased E eV [26] and the formation of more stable A-center, which is not fully compensated by cm is formed remaining free donors. CdTe with C in this case. after annealing at Figs. 8–10 show detailed plots of carrier densities and principal neutral and charged defect densities for typical examples of defect interplay. Fig. 8 depicts the case a plotted in Fig. 6. The self-compensation of V and D below 350 C is clearly the increase of visible. Despite of the fast decrease of V stabilizes at Fermi energy compensates such course and V the density, which strongly compensates D . If the remaining uncompensated charge can be disposed to the midgap level after quenching to RT, SICT is obtained.

Fig. 9 shows the densities for the case e plotted in Fig. 7. Due to fast disappearance of Te its midgap level cannot participate at the Fermi level pinning at the low temperature. The self-compensation is assured by the mutual compensation of D , A , and V . The formation of A-center decreases below the constantly. midgap and The formation of SICT is simplified, if midgap donor doping (Ge [27], Sn [28]) is used. Such a situation is demonstrated cm in Fig. 10, where the midgap donor doping N in the defect model a is used. The high midgap level density allows to reach SICT at increased C without strong self-compensation of shallow defects. Such a procedure cannot be, however, recommended for the detector fabrication due to

T

p

1930


below given limit. This finding approves the model for experimental verification. The diffusion data have to be, however, applied with care. Chemical diffusion has been measured in nearly stoichiometric CdTe only and might differ significantly at Te saturation. New measurements in Te rich conditions are thus advisable. IV. CONCLUSION

Fig. 10. Carrier densities and neutral and charged defect densities calculated for defect model a where midgap donor doping with density = 10 cm and A-center formation is assumed. N

Self-compensation of point defects and formation of semi-insulating material was studied theoretically in donor-doped Te saturated CdTe to simulate procedures commonly used in the detector fabrication. We have shown that apparently contradictory demands to grow semi-insulating CdTe with deep level cm can be accomplished by low-temperdoping below ature thermal treatment, which includes precipitation of excess Te. The credibility of the approach is justified by used model parameters, which describe well also high temperature transport data and the phase diagram. All analyzed defect models produced semi-insulating CdTe if doped by shallow donor occupying Cd sublattice and equilibrated at Te saturation below 220 C. Doping by donor in Te sublattice produces similar results at lower . The diffusion rate at low temperature is fast enough to reach real-time equilibrium. The fit of T-x phase diagram results in fast disappearance of Te antisites, which consequently do not influence electrical properties at low temperature even if Te as midgap or shallow donor is assumed. REFERENCES

Fig. 11. Estimated relaxation time of diffusion-limited defect reaction. Respective lines correspond to various diffusion range (1: 0.1 cm, 2: 10 cm, 3: 10 cm, 4: 10 cm, 5: 10 cm, 6: 10 cm. The dash line characterizes real-time experiments with t = 10 s.

strong trapping of photogenerated carriers on the midgap level and the deterioration of the detector performance. The practical application of the outlined theory is limited by C, the relaxation time to reach the equilibrium at where the precipitation (2) accompanied by the diffusion of V and Te to Te precipitates can take relatively long time. Estimated relaxation time of diffusion-limited processes is plotted in Fig. 11 assuming chemical diffusion coefficient eV cm s [29] and various charcm. Taking into acacteristic diffusion scales s the decount characteristic time for real time annealing fect reaction close to 200 C occurs for cm. This corresponds to precipitate density cm . The precipitate density was reported in CdTe in the cm [30], [31], which yields well interval

[1] T. E. Schlesinger, J. E. Toney, H. Yoon, E. Y. Lee, B. A. Brunett, L. Franks, and R. B. James, “Cadmium zinc telluride and its use as a nuclear radiation detector material,” Mater. Sci. Eng., vol. 32, pp. 103–189, Apr. 2001. [2] C. Szeles, “CdZnTe and CdTe materials for X-ray and gamma ray radiation detector applications,” Phys. Stat. Sol. B, vol. 241, pp. 783–790, Mar. 2004. [3] J. W. Mayer, “Search for semiconductor materials for gamma ray spectroscopy,” in Semiconductor Detectors, G. Bertolini and A. Coche, Eds. Amsterdam, The Netherlands: North Holland, 1968, pp. 445–498. [4] B. K. Meyer and W. Stadler, “Native defect identification in ii–vi materials,” J. Cryst. Growth, vol. 161, pp. 119–127, Apr. 1996. [5] M. Chu, S. Terterian, D. Ting, C. C. Wang, H. K. Gurgenian, and S. Mesropian, “Tellurium antisites in CdZnTe,” Appl. Phys. Lett., vol. 79, pp. 2728–2730, Oct. 2001. [6] R. Grill, J. Franc, P. Höschl, E. Belas, I. Turkevych, L. Turjanska, and P. Moravec, “Semiinsulating CdTe,” Nucl. Instrum. Meth. A, vol. 487, pp. 40–46, Jul. 2002. [7] C. Szeles, Y. Y. Shan, K. G. Lynn, and E. E. Eissler, “Deep electronic Zn Te,” Nucl. Instrum. Meth. levels in high-pressure Bridgman Cd A, vol. 380, pp. 148–152, Oct. 1996. [8] A. Castaldini, A. Cavallini, B. Fraboni, P. Fernandez, and J. Piqueras, “Deep energy levels in CdTe and CdZnTe,” J. Appl. Phys., vol. 83, pp. 2121–2126, Feb. 1998. [9] M. Funaki, T. Ozaki, K. Satoh, and R. Ohno, “Growth and characterization of CdTe single crystals for radiation detectors,” Nucl. Instrum. Meth. A, vol. 436, pp. 120–126, Oct. 1999. [10] M. A. Berding, “Native defects in CdTe,” Phys. Rev. B, vol. 60, pp. 8943–8950, Sep. 1999. [11] K. Zanio, “Cadmium telluride,” in Semiconductors and Semimetals, R. K. Willardson and A. C. Beer, Eds. New York: Academic, 1978, vol. 13, pp. 115–163. [12] N. V. Agrinskaya and T. V. Mashovets, “Self-compensation in semiconductors—a review dedicated to the 100th anniversary of the birthday of Frenkel, Yakov, Ilich,” Semiconductors, vol. 28, p. 843, Sep. 1994. [13] R. Grill, J. Franc, I. Turkevych, P. Höschl, E. Belas, and P. Moravec, “Semi-insulating Cdte with a minimum deep level doping,” Phys. Stat. Sol. C, vol. 2, pp. 1489–1494, Mar. 2005.


[14] R. Fang and R. F. Brebrick, “CdTe i: solidus curve and composition-temperature-tellurium partial pressure data for Te-rich CdTe(s) from optical density measurements,” J. Phys. Chem. Solids, vol. 57, pp. 443–450, Apr. 1996. [15] J. H. Greenberg, V. N. Guskov, V. B. Lazarev, and O. V. Shebershneva, “Vapor-pressure scanning of nonstoichiometry in cadmium telluride,” Mater. Res. Bull., vol. 27, pp. 847–854, Jul. 1992. [16] F. A. Kröger, The Chemistry of Imperfect Crystals. Amsterdam, The Netherlands: North Holland, 1974. [17] M. A. Berding, M. van Schilfgaarde, and A. Sher, “First-principles calculation of native defect densities in Hg Cd Te,” Phys. Rev. B, vol. 50, pp. 1519–1534, Jul. 1994. [18] Y. Marfaing, “Point defects in narrow gap ii–vi compounds,” in Electronic Materials 3, Narrow-Gap II–VI Compounds for Opto-Electronic and Electromagnetic Applications, P. Capper, Ed. London, U.K.: Chapman & Hall, 1997, pp. 238–267. [19] R. Grill, J. Franc, I. Turkevych, P. Höschl, E. Belas, and P. Moravec, “Semi-insulating CdTe with a minimized deep level doping,” J. Electron. Mater., vol. 34, pp. 939–943, Jun. 2005. [20] R. Grill, J. Franc, P. Höschl, I. Turkevych, E. Belas, P. Moravec, M. Fiederle, and K. W. Benz, “High temperature defect structure of Cdand Te-Rich CdTe,” IEEE Trans. Nucl. Sci., pt. 2, vol. 49, no. 6, pp. 1270–1274, Jun. 2002. [21] CRC Handbook of Chemistry and Physics, 79th ed., D. R. Lide, Ed., CRC, Boca Raton, FL, 1998. [22] P. Emanuelsson, P. Omling, B. K. Meyer, M. Wienecke, and M. Schenk, “Identification of the cadmium vacancy in CdTe by electron paramagnetic resonance,” Phys. Rev. B, vol. 47, pp. 15 578–15 580, Jun. 1993.

1931

[23] D. Nobel, “Phase equilibria and semiconducting properties of cadmium telluride,” Philips Res. Rep., vol. 14, pp. 361–399, Mar. 1959. [24] J. Franc, R. Grill, L. Turjanska, P. Höschl, E. Belas, and P. Moravec, “High temperature mobility of CdTe,” J. Appl. Phys., vol. 89, pp. 786–788, Jan. 2001. [25] D. M. Hofmann, W. Stadler, K. Oettinger, B. K. Meyer, P. Omling, M. Salk, K. W. Benz, E. Weigel, and G. Müller-Vogt, “Structural properties of defects in Cd Zn Te,” Mater. Sci. Eng. B, vol. 16, pp. 128–133, Jan. 1993. [26] P. Höschl, P. Moravec, J. Franc, E. Belas, and R. Grill, “Defect equilibrium in semi-insulating CdTe(Cl),” Nucl. Instrum. Meth. A, vol. 322, pp. 371–374, Nov. 1992. [27] M. Fiederle, V. Babentsov, J. Franc, A. Fauler, K. W. Benz, R. B. James, and E. Cross, “Defect structure of Ge doped CdTe,” J. Cryst. Growth, vol. 243, pp. 77–86, Aug. 2002. [28] J. Franc, M. Fiederle, V. Babentsov, A. Fauler, K. W. Benz, and R. B. James, “Defect structure of Sn-doped CdTe,” J. Electron. Mater., vol. 32, pp. 772–777, Jul. 2003. [29] R. Grill, L. Turjanska, J. Franc, E. Belas, I. Turkevych, and P. Höschl, “Chemical self-diffusion in CdTe,” Phys. Stat. Sol. B, vol. 229, pp. 161–164, Jan. 2002. [30] R. S. Rai, S. Mahajan, S. McDewitt, and C. J. Johnson, “Characterization of CdTe, (Cd,Zn)Te, and Cd(Te,Se) single crystals by transmission electron microscopy,” J. Vac. Sci. Technol. B, vol. 9, pp. 1892–1896, May/Jun. 1991. [31] P. Rudolph, “Melt growth of ii–vi compound single crystals,” in Recent Development of Bulk Crystal Growth, M. Milssiki, Ed. Trivandrum, India: Research Signpost, 1998, pp. 127–164.