An investigation, using both capacitance and current transient spectroscopy, was made of highly Te-doped (greater than 1017/cm3) material in order to determine the presence of EL2 traps. The important parameter for the detection of EL2 was the relative position of the electron quasi-Fermi level in the depletion region. Two peaks were detected in less highly doped samples. One was attributed to EL2 (Ec - Et = 0.80eV) and the other to ECX (Ec - Et = 0.34eV). In the case of the most highly doped samples, a third peak appeared at Ec - Et = 0.59eV. The EL2 peak did not appear in the deep-level transient spectra of the most highly doped samples, but it was still present in the current transient and photo-induced current transient spectra. This indicated that EL2 did not annihilate in highly Te-doped material. Its disappearance was found to be an artefact of deep-level transient spectroscopy, and was attributed to thermal noise. The latter became comparable to the distance between EL2 and the quasi-Fermi level. This effect strongly affected the capacitance spectroscopic method, and resulted in an impaired reliability at high temperatures. Arrhenius plots of the EL2 level moved towards lower temperatures with increasing doping concentration. This could be explained by the combined effect of changes in the density of states near to the conduction band edge, of dopant-induced strain in the lattice, of a Burstein shift for n-type materials and by majority carrier screening. The net result of all of these effects was a narrowing of the band gap width, and a variation in the measured activation energies of deep traps, with increasing dopant concentration. An alternative explanation for the shift was a rearrangement, of the atomic configuration of EL2, that was induced by an increasing dopant concentration. This hypothesis was supported by the appearance of a third level in the most highly-doped samples.

A.Castaldini, A.Cavallini, B.Fraboni, J.Piqueras: Journal of Applied Physics, 1995, 78[11], 6592-5