Halide perovskites are promising materials for optoelectronic applications in solar cell devices due to their excellent optoelectronic performance. However, they suffer from several dynamical degradation problems, which are difficult to characterize. Atomistic simulations can provide valuable insights, although the high computational cost of first-principles methods such as DFT makes it challenging to model dynamical processes in large perovskite systems. To overcome these limitations, researchers from Eindhoven University of Technology (TU/e) have refined GFN1-xTB parameters to accurately describe the structural, energetic, and dynamical properties of inorganic halide perovskites.
In a recent study, the semi-empirical density functional tight binding method, GFN1-xTB, has been refined using ParAMS to improve the performance of computing properties of perovskites containing Cs, Pb, I, and Br atoms. A training set based on DFT calculations has been generated to train a set of parameters of the GFN1-xTB Hamiltonian. The performance of the refined parameters has been benchmarked against experiments and DFT calculations, showing an accurate description of the phase transition of these halide perovskites.
The study shows that the phase stability is strongly correlated to the displacement of ions in the perovskites. In the orthorhombic phase, the directional movement of the Cs cations increases their distance to the surrounding halides, which can trigger decomposition to the nonperovskite phase. However, once enough thermal energy is available, the increased Cs−halide distance can be compensated by increased halide fluctuations, resulting in a transition to the phase-stable tetragonal or cubic phases. Furthermore, it is shown that the mixing of halides increases halide displacement, thus decreasing the phase transition temperatures and therefore improving the phase stability of the perovskites.
S. Raaijmakers, M. Pols, J. M. Vicent-Luna, S. Tao, Refined GFN1-xTB Parameters for Engineering Phase-Stable CsPbX3 Perovskites, J. Phys. Chem. C, 126, 9587-9596 (2022)
The supporting information includes the reference data and optimized DFTB parameters.