Regular and Anomalous-Type Conversion of p-CdTe during Cd-Rich Annealing

Journal of Electronic Materials, Jun 2005 by Belas, E, Franc, J, Grill, R, Toth, A L, Et al

The kinetics of the p-to-n conversion and effect of the anomalous n-to-p reconversion of p-CdTe during annealing at 400-700°C under Cd-rich overpressure have been investigated. The p-to-n conversion is related to diffusion of Cd interstitials together with gettering of foreign fast diffusing acceptors to the center of the sample. The propagation of the n-type layer during annealing at 500°C was found to be significantly slower then the standard square-root dependence on annealing time. The anomalous n-to-p reconversion of the converted n-type sample was observed after sufficient long time annealing at 600°C.

Key words: Annealing, CdTe, diffusion, p-n junction formation, photoluminescence

INTRODUCTION

CdTe is an important material in the field of x-ray and gamma-ray detectors, electro-optical devices, solar cells, or substrates for narrow gap (HgCd)Te epitaxy. It is well known that structural defects and impurities in the substrates strongly affect the final quality of the epitaxial films. The standard technological operation for reducing structural defects such as inclusions or precipitates and for adjusting the concentration of native defects is annealing under overpressure of one of the material components. Electrical and optical properties after annealing at various temperatures have been frequently investigated in the past and the p-to-n conversion was observed in p-CdTe after annealing under Cd overpressure.1-5 The standard square-root dependence of the thickness of the converted n-type layer on annealing time was evaluated in CdTe from improvement of the IR transmission during time-step annealing at 600°C.3 However, the detailed investigation of the p-to-n conversion has not yet been presented.

Low concentration of fast diffusing group I impurities (Li, Na, Cu, Ag) in the substrates is also one of the key parameters for a production of high-quality (HgCd)Te epilayers. Impurities can diffuse into the epilayer during epitaxial growth and deteriorate the quality of the final (HgCd)Te films.6 Annealing in Cd overpressure is one of the possible ways for CdTe purification.7-10

In the present paper, a detailed investigation of the p-to-n conversion and the anomalous n-to-p reconversion of p-CdTe during Cd-rich annealing is reported. Electron beam induced current (EBIC) measurements are compared with the Hall effect, conductivity, and photoluminescence (PL) measurements to study the kinetics of the p-n junction formation. Slow propagation of the n-type layer during Cd diffusion, which is not possible to describe by a standard diffusion theory, is observed, and the interaction between Cd interstitials (Cd^sub i^) and Te precipitates (Te^sub ppt^) is proposed for an explanation of the results. The anomalous n-to-p reconversion, which is observed after long-time annealing at 600°C, is supposed to be related to an activation of acceptors from dissolved Te^sub ppt^.

EXPERIMENTAL PROCEDURE

Undoped CdTe single crystals were grown by the Vertical Gradient Freeze method in Prague. All samples were as-grown p-types with the hole concentration 8 × 10^sup 15^-2 × 10^sup 16^ cm^sup -3^ and hole mobility 80-100 cm^sup 2^/Vs at room temperature (RT). Cubic samples (usually 3 × 3 × 3 mm^sup 3^) were polished and etched in 3% Br-methanol solution before annealing in order to remove a damaged surface layer. The annealing under Cd overpressure was performed in a vacuum-sealed quartz ampoule, where the Cd source was held at 3°C below the temperate of the sample. The ampoule was quenched in a water bath after annealing. The depth of the p-n junction was measured by the EBIC method. Samples were polished from one side to half of their initial thickness, contacted, and the position of the p-n junction was determined from the position of the EBIC maximum, excited by the electron beam scanning the polished cross section of the sample. Electrical conductivity and the Hall coefficient were measured by the van der Pauw method. Electrical contacts were prepared by electroless deposition of Au from AuCl^sub 3^ solution and samples were chemomechanically polished in 3% bromine-ethylene glycol solution before contact preparation. Photoluminescence measurements were performed at 4.2 K using a He-Ne laser with a power of 8 mW and an average power flux of 20 Wcm^sup -2^. The PL spectra were measured by an infrared Fourier transform spectrometer BRUKER IFS66/S in a He flow cryostat.

RESULTS

The mixed images of the secondary electron and EBIC measurements after 16 and 65 h annealing at 500°C are shown in Figs. 1 and 2, where the light EBIC contour represents the position of the p-n junction. It can be seen (Fig. 2) that the square shape of the sample does not affect the contour of the p-n junction after longer time annealing and a spherical core is created. The depth of the p-n junction as a function of annealing time at 500°C and 600°C is plotted in Fig. 3, where two sets of samples from two different parts of the ingot were used for the annealing at 500°C. It was found that the propagation of the p-n junction is not a usual square-root function of the annealing time. The solid lines representing the best fits give approximately the fourth root time dependence for both sets of samples annealed at 500°C. In addition, the first set of samples shows slower p-n junction propagation than samples from the different part of the ingot probably due to higher initial deviation from the stoichiometry. The result for annealing at 600°C is closer to the standard diffusion profile. Converted n-type layer was also prepared by 16 h annealing at 400°C, where the depth of the p-n junction was 85 μm. The p-n junction was not detected after 4 h annealing at 700°C, when the compensated high-resistivity sample (p = 2 × 106 Ωcm) was obtained.