Defect physics of the CuInSe2 chalcopyrite semiconductor

We studied the defect physics in CuInSe2, a prototype chalcopyrite semiconductor. We showed that (i) it takes much less energy to form a Cu vacancy in CuInSe2 than to form cation vacancies in II-VI: compounds (ii) defect formation energies vary considerably both with the Fermi energy and with the chemical potential of the atomic species, and (iii) the defect pairs such as (2V(Cu)(-)+In-Cu(2+)) and (2Cu(In)(2-)+In-Cu(2+)) have particularly low formation energies (under certain conditions, even exothermic). Using (i)-(iii), we (a) explain the existence of unusual ordered compounds CuIn5Se8, CuIn3Se5, Cu2In4Se7, and Cu3In5Se9 as a repeat of a single unit of (2V(Cu)(-)+In-Cu(2+)) pairs for each n=4, 5, 7, and 9 units, respectively, of CuInSe2; (b) attribute the very efficient p-type self-doping ability of CuInSe2 to the exceptionally low formation energy of the shallow defect Cu vacancies; (c) explained in terms of an electronic passivation of the In-Cu(2+) by 2V(Cu)(-) the electrically benign character of the large defect population in CuInSe2. Our calculation leads to a set of new assignment of the observed defect transition energy levels in the band gap. The calculated level positions agree rather well with available experimental data. [S0163-1829(98)01516-1].