Mixed chalcopyrite semiconductors like Cu(In,Ga)S2 and Cu(In,Ga)Se2 are characterized by the coexistence of different local atomic arrangements around the S or Se anion. The resulting anion displacement strongly influences the material bandgap. We studied the atomic-scale structure of Cu(In,Ga)S2 as a function of composition using x-ray absorption spectroscopy and valence force field simulations. Applying a specially developed model for not fully random cation distributions, we find that structural relaxation of the anion with respect to In and Ga contributes significantly more to the bandgap bowing observed for Cu(In,Ga)S2 and Cu(In,Ga)Se2 than relaxation with respect to Cu and group-III atoms.

No abstract

There is particular interest to investigate compositional inhomogeneity of Cu(In,Ga)Se2 solar cell absorbers. We introduce an approach in which focused ion beam prepared thin lamellas of complete solar cell devices are scanned with a highly focused synchrotron X-ray beam. Analyzing the resulting fluorescence radiation ensures high resolution compositional analysis combined with high spatial resolution. Thus, we are able to detect subtle variations of the Ga/(Ga + In) ratio down to 0.01 on a submicrometer scale. We observed that for sequentially processed solar cells a higher selenization temperature leads to absorbers with almost homogenous Ga/(Ga + In) ratio, which significantly improved the conversion efficiency.

The atomic-scale structure and vibrational properties of semiconductor alloys are determined by the energy required for stretching and bending the individual bonds. Using temperature-dependent extended x-ray absorption fine-structure spectroscopy, we have determined the element-specific In-As and Ga-As effective bond-stretching force constants in (In,Ga)As as a function of the alloy composition. The results reveal a striking inversion of the bond strength where the originally stiffer bond in the parent materials becomes the softer bond in the alloy and vice versa. Our findings clearly demonstrate that changes of both the individual bond length and the surrounding matrix affect the bond-stretching force constants. We thus show that the previously used common assumptions about the element-specific force constants in semiconductor alloys do not reproduce the composition dependence determined experimentally for (In,Ga)As.

No abstract