Résumé
Zinc selenide (ZnSe) is now a material of established interest for the design of optoelectronic devices, following the achievement of both types of conductivity by M. Haase et al. (Appl. Phys. Lett., 59 (1991) 1272). The p-type doping of this material is now well established in molecular beam epitaxy, and work is in progress in many laboratories to transpose this to metal-organic vapour phase epitaxy (MOVPE). One of the problems to be solved in MOVPE, for the successful achievement of p-type doping, is that of obtaining a material with very low electron background concentrations, to avoid compensation effects. It has previously been demonstrated, by Jones, Wright and co-workers (Semicond. Sci. Technol., 6 (1991) A36; J. Cryst. Growth, 94 (1989) 441; J. Cryst. Growth, 104 (1991) 297) that the classical zinc precursor leads to prereactions and is invariably contaminated by trace amounts of iodine. To get rid of these problems, they employ a new precursor for zinc, that is dimethylzinc triethylamine adduct (Me(2)ZnEt(3)). Their results demonstrate that the material purity can be increased. Following Wright et al. (J. Cryst. Growth, 94 (1989) 441; J. Cryst. Growth, 104 (1991) 297), we have employed this original precursor to grow extremely high purity ZnSe. We investigated the growth conditions and determined the optimum growth parameters to be a growth temperature of 280 degrees C, a VI/II molar ratio of 5 and a reactor pressure of 40 Torr. The samples were characterised by low temperature photoluminescence and reflectivity experiments. We observed that the use of Me(2)ZnEt(3) led to a dramatic decrease of the donor bound exciton intensity in the 2 K photoluminescence spectrum, along with a correlated increase of the free exciton intensity. For the first time, free excitons and excited states up to the second (3s) were observed in reflectivity spectra. The spectra were theoretically fitted using a model similar to the one of Zheng et al. (Appl. Phys. Lett., 52 (1988) 287), in the local approximation, including a ''dead layer'' as prescribed by Hopfield and Thomas (Phys. Rev., 132 (1988) 563). Using this model, we accurately deduced the transition energy positions and oscillator strengths of the observed transitions. Moreover, we deduced that the damping parameter of the HH (1s) transition is 1.4 meV. Such a sharpness clearly indicates the high crystalline quality of the material.