It is a long-standing puzzle that ZnSe is difficult to dope p type, while ZnTe-which is very similar to ZnSe-is very easily doped p type. We report ab initio calculations which show that the solubilities of Li and Na acceptors are much greater in ZnTe than the solubilities of the same acceptors in ZnSe. We trace the origin of this difference to the bonding properties of the acceptors with the neighboring chalcogens. Our results also explain the experimentally observed dependence on dopant

more »

... tration of the dislocation density in p-type ZnSe epilayers grown on GaAs. Efforts to make optical devices from wide band-gap II-VI semiconductors like ZnSe have been hampered by the fact that they can be easily doped either n type or p type but not both.lA For example, ZnSe is easily doped n type, but only recently, after many years of extensive effort, it became possible to dope it sufficiently p type so the blue lasers could be fabricated.5 In contrast, ZnTe can only be doped p type.2'3 This difference between ZnSe and ZnTe is particularly striking because the two materials are very similar in most of their properties. Indeed, selenium derives its name (from the Greek for Moon) from its resemblance to tellurium (from the Latin for Earth). Over the years, there have been many suggestions to account for these doping difficulties.614 Among the most popular were ( 1) compensation, either by native point defects (vacancies, interstitials, or antisites) or by dopant atoms at alternative sites (e.g., interstitials) where they would act. as traps that capture free carriers; (2) strong lattice relaxations which convert the anticipated shallow dopants into nondoping deep levels; and (3) low solubility for dopants. In a recent paper," we constructed a quantitative theory which includes all of the above phenomena on an equal footing. We carried out first-principles calculations for p-type ZnSe, and found that (a) native defects have concentrations that are too small to affect doping levels; (b) nondoping impurity configurations~ play only a minor role in limiting doping; and (c) the dominant effect is the solubility of the acceptors (Li, Na, and N), limited by the formation energy of other phases, such as Li,Se, etc. These results provide a definitive account of the relative roles of the mechanisms that can potentially limit the doping of p-type TnSe. But the question remains: Why does ZnTe behave differently? In this letter we present a first-principles theoretical study of the solubilities of acceptors in ZnSe and ZnTe. We show that the solubilities of Li and Na acceptors in ZnTe are much higher than their solubilities in ZnSe. These results-which contain no adjustable parameters or experimental inputs-provide a natural explanation for the observed difference between the two semiconductors. We trace the origin of the solubility difference between the two materials to the bonding properties between the dopants and the neighboring chalcogens. Our calculated solubilities also provide a simple explanation for a set of expetiental observations regarding the dislocation structure of p-type ZnSe grown epitaxially on GaAs substrates. We first give a brief review of the theoretical methods used in the calculations. Full details are provided elsewhere.15,16 The total energies of ZnSe, ZnTe, and the acceptor defects are calculated using norm-conserving pseudopotentials and density-functional theory in the local-density approximation. l8 These calculations successfully describe the bulk properties of both ZnSe and ZnTe by using a mixed-basis set to capture the behavior of the Zn d electron states.16 Calculations for acceptor defects are performed in 32-atom supercells, and relaxations are included. The estimated error bar for the defect total energy calculations is 0.1 eV. The formation energy of a Llzn acceptor in ZnSe is determined by the relative abundance of Zn, Se, and Li atoms in the environment in which the crystal is grown. As a consequence, the formation energy of the acceptor is a function of the chemical potentials for the Zn and Li atoms, pzn and PLi. The Zn chemical potential is defined as the energy required to add a Zn atom to the system. Since forming a substitutional Liz, acceptor requires the addition of one Li atom to the crystal and the removal of one Zn atom, the formation energy of the neutral acceptor is Eform( ZnSe:Liz,) = 8 ( ZnSe:Lizn) -pLi +pzn, where 8 (ZnSe:Liz,) is a total energy derived from a supercell calculation for substitutional Li. Given the formation energy, the acceptor concentration is determined by the usual Boltzmann form. Thermodynamics places limits on the values of the Zn and Li chemical potentials. The