Defect Control of Kesterite Cu2ZnSn(S,Se)4 for Thin-film Solar Cell Absorbers

Su-Huai Wei
National Renewable Energy Laboratory

The kesterite-structured semiconductors Cu2ZnSnS4 and Cu2ZnSnSe4 are drawing considerable attention recently as the active layers in earth-abundant low-cost thin-film solar cells. The addi-tional number of elements in these quaternary compounds, relative to binary and ternary semi-conductors, results in increased flexibility in the material properties. Conversely, a large variety of intrinsic lattice defects can also be formed, which have important influence on their optical and electrical properties, and hence their photovoltaic performance. Using first-principles density functional theory (DFT) we have systematically calculated defect formation and ionization energies in these kesterite materials and compared them with the better studied chalcopyrite materials CuGaSe2 and CuInSe2. We find that the increased number of elements in the quaternary kesterite materials makes the properties of the intrinsic lattice defects more complicated than for the previously studied ternary chalcopyrites. They are also more important in influencing the photovoltaic performance. Specifically, we find that: (1) The narrow chemical potential range is limited by the various competing secondary compounds, such as CuS, Cu2S, Cu2SnS3, ZnS, SnS and SnS2. ZnS and ZnSe coexistence is highly possible in the kesterites with low Cu/(Zn+Sn) and high Zn/Sn ratios. (2) The CuZn antisite is the dominant point defect in the stoichiometric Cu2ZnSnS4 and Cu2ZnSnSe4 samples. Its ionization level is deeper than that of VCu, but its high population can still produce a significant hole concentration, determining the intrinsic p-type conductivity and making n-type doping difficult. (3) The formation energy and ionization level of VCu in kesterites are similar to those in the chalcopyrites, but the population is much lower than CuZn in the stoichiometric samples. Under Cu poor and Zn rich conditions (Cu/(Zn+Sn) ˜ 0.8), VCu becomes more dominant and contrib-utes to p-type conductivity, which reflects the situation in the real solar cells with low Cu/(Zn+Sn) ratio and high efficiency. (4) The frequently observed non-stoichiometry in the quaternary kesterites results from the facile formation of self-compensated defect clusters, such as [VCu+ZnCu], [ZnSn+2ZnCu] and [2CuZn+SnZn]. (5) [2CuZn+SnZn] clusters induce electron-trapping states in the absorber materials, and are thus detrimental to the solar cell per-formance. Their facile formation and high population even in the near-stoichiometric Cu2ZnSnS4 and Cu2ZnSnSe4 samples degrades the efficiency of the solar cells with Cu/(Zn+Sn) and Zn/Sn ratios near unity, so a rather Zn rich and Cu, Sn poor condition is required to prevent its formation and improve the solar cell performance. This explains the empirical observation that Cu poor and Zn rich condition is crucial for the high solar cell effi-ciency. (6) The electron-trapping caused by [2CuZn+SnZn] in Cu2ZnSnSe4 is much weaker than in Cu2ZnSnS4, which could be the main reason for the observation that the Cu2ZnSn(S,Se)4 solar cells achieve the highest-efficiency when S composition is low. The lower population of isolated deep donor defects such as SnZn and VSe in Cu2ZnSnSe4 is another reason. This work is done in collaboration with Shiyou Chen, Aron Walsh and Xingao Gong. The work at NREL is funded by the US Department of Energy, under Contract No. DE-AC36-08GO28308.

Back to Workshop I: Solar Cells