PHYSICAL REVIEW B VOLUME 45, NUMBER 2Q 15 MAY 1992-II Photolnminescence of heavily doped n-type CdSe Miguel Levy, ' W. K. Lee, and M. P. Sarachik City College of the City University of New York, New York, New York 10031 S. Geschwind~ A T& T Bell Laboratories, Murray Hill, New Jersey 07974 (Received 4 November 1991) We report on a study of the photoluminescence of a series of indium-doped n-type CdSe samples with carrier concentrations above the critical concentration for the metal-insulator transition. The photo- luminescence was measured for excitation energies both greater and smaller than the optical gap, and ex- citation spectroscopy was also used as a probe. As in earlier results for CdS the spectra of CdSe exhibit a sharp edge at high energies corresponding to the decay of electrons from the Fermi level in the conduc- tion or donor-impurity band to states in the valence band, and a broad tail at lower energies due to tran- sitions from the rest of the Fermi sea. While in CdS the high-energy edge moved monotonically to higher energy with increasing carrier concentration, ND — N&, this monotonic behavior did not hold for two of the CdSe samples. We suggest that this unexpected behavior can be attributed to the effect of different levels of compensation among the samples used in the present studies. The full results are inter- preted in terms of band filling, electronic interactions, and band-gap renormalization in the presence of compensation. INTRODUCTION Photoluminescence studies of heavily doped semicon- ductors with dopant concentrations above the critical concentration for the metal-insulator transition yield use- ful information regarding the filling of the conduction band in these materials, the size of the fundamental gap, as well as donor and acceptor energy levels. As the dopant concentration is raised, there is an increase in the energy of the peak of the fluorescent radiation due to the filling of the conduction band, which is partly offset by a decrease associated with band tailing and with a narrow- ing of the band gap due to many-body effects and electron-impurity interactions. The photoluminescence of CdS, which is expected to be similar to that of CdSe, has been measured by Kuki- moto et ttl. , ' and by Geschwind and Devlin (see Ref. 2). Spectra obtained by Geschwind and Devlin for dopant concentrations above the critical concentration for the m. etal-insulator transition were found to exhibit a high- energy edge and a broad low-energy tail, where the sharp feature on the high-energy side corresponds to an elec- tron decaying from the Fermi level in the conduction band to the top of the valence band (shown as process 1 in Fig. 1). Girvin studied these spectra in some theoreti- cal detail and offered a model to account for the depen- dence of the position of the high-energy edge on carrier concentration. He considered various sources of electric potential which perturb the electron and hole energy lev- els, and have a bearing on the photoluminescence spectra. Thus, the Moss-Burnstein shift that arises from simple band filling of the conduction band is modified by electron-electron interactions that renormalize the funda- mental band gap towards lower energies. Moreover, the phonon cloud dressing the polaronic electrons in these Fermi Level Bond Gap: I. 84eV ' Crysta I-Field '~ Splitting: 25meV ceptor Holes Bond A Spin-orbit Splitting: Valence Bond C FIG. 1. Energy-level diagram indicating processes described in the text. partially ionic crystals is, to some extent, stripped away by the electron screening of the electron-lattice interac- tion. Furthermore, the presence of holes generated by the excitation light polarizes the Fermi sea of electrons, contributing a correlation energy term to the interband energy difference. Finally, the random impurity potential perturbs the wave functions of the electrons in the con- duction band and the holes in the valence band and modifies their energies. Conduction Bond 45 11 685 1992 The American Physical Society