Copper sulfides with different copper concentration exist in mineral form ranging from CuS to Cu2S. Among these, chalcosite Cu2S, and digenite Cu1.8S were the subject of extensive research for decades mainly because of their use as the absorber in photovoltaic cells. Yet, their electronic structure is poorly understood because their crystal structure is complex. Most of the results published so far report the semiconducting nature of these compounds with the energy band gap being in the range of 0.84 to 1.9 eV. The crystal structure consists of a close-packed lattice of S with mobile Cu occupying various types of interstitial sites with a statistical distribution depending on temperature. In this thesis we present the first computational study of their electronic band structure.
Initially, we investigated the simpler antifluorite structure. Both local density approximation (LDA) and self-consistent quasiparticle GW calculations with the full-potential linearized muffin-tin orbital method give a semimetallic band structure. Inspection of the nature of the bands shows that the lowest conduction band is mainly Cu-s-like except right near the center of the Brillouin zone where a Cu-s-like state lies about 1 eV below the valence band maximum. Significantly, in GW calculations, this state shifts up by several 0.1 eV but not sufficiently to open a gap. A random distortion of the Cu atoms from the perfect antifluorite positions is found to break the degeneracy of the d state at the Gamma-point and thus opens up a small gap of about 0.1 eV in LDA. As our next step we constructed supercell models for the cubic and hexagonal phases with the Cu positions determined by a weighted random number generator. The low temperature monoclinic phase was also studied. The computed total energies of these structures follow the same order as the reported phases with increasing temperatures. All these models gave similar small band gaps of order 0.1 – 0.2 eV. However, their conduction band is now mainly s-like and addition of an expected Cu-s level shift opens the gap to about 0.5 eV. Some simpler hexagonal model structures gave slightly larger band gap but were found to be unrealistic. The optical absorption data all show a strong intraband absorption with a minimum in absorption at about 1 eV. Our calculations suggest a significantly lower gap of order 0.5 eV with low absorption cross section, the true nature of which is masked by the free carrier absorption. As part of our study of the related Cu-compounds, we analyzed the quasiparticle effects beyond LDA obtained from a GW calculation on the effective masses and Kohn-Luttinger hamiltonian parameters for CuBr.