Publication date: 17th February 2025
Formamidinium tin iodide (FASnI₃) is a promising lead-free perovskite for photovoltaics, but its limited thermal stability and lower efficiency compared to lead-based counterparts remain significant challenges. Accurate theoretical modelling of FASnI₃ is essential for understanding its structural and electronic properties. While hybrid functionals and spin–orbit coupling (SOC) are commonly employed to match experimental band gaps, they are often applied only to minimal unit cells or small supercells that lack structural realism [1, 2, 3].
Here, we investigate how supercell size, FA orientation, and the level of theory influence the predicted band gap and structure of α-FASnI₃. We find that when sufficiently large supercells (at least 6×6×6) are relaxed along all degrees of freedom, the system naturally converges toward a cubic average structure—while preserving local octahedral distortions, minimizing residual dipoles, and recovering threefold symmetry. Band gap fluctuations at 300 K also decrease with increasing system size, indicating improved structural stability. These findings emphasize that structural convergence must precede theoretical refinement—accurate functionals cannot compensate for unphysical geometries.
Laboratory of Computational Chemistry and Biochemistry (LCBC) Financial Support
Scientific IT and Application Support (SCITAS, EPFL) and Swiss National Supercomputing Centre (CSCS) High Performance Computing Resources