Effect of bromide incorporation on the electronic & photovoltaic properties of Sn-based perovskite devices: A multiscale investigation utilizing first principles approach and numerical simulation, aided by machine learning models

In this study, the effects of Bromide doping into iodide sites in cesium tin triiodide (CsSnI3), methylammonium tin triiodide (MASnI3) and formamidinium tin triiodide (FASnI3), were investigated using a novel multiscale computational approach. Density functional theory (DFT) was used to probe the impact of Br addition on the electronic structure and the bandgap, while Solar Cell Capacitance Simulator (SCAPS) was used to numerically simulate & optimize solar devices containing these perovskite absorbers. A set of supervised machine learning algorithms were used to model the relationship between SCAPS input and output parameters to determine which input parameters have significant contributions to solar cell efficiency. This information was used to couple DFT & SCAPS; allowing the use of accurate band gap values predicted by DFT-1/2 approach, as inputs in device scale simulations. The novel framework was then successfully applied to predict the combined effect of bromide concentration and solar cell geometry on the power conversion efficiency (PCE). Furthermore, the predicted trends were explained both in the context of the underlying electronic structures and device performance parameters. For a set of common device configurations our work predicted maximum efficiencies of 4.08 %, 9.61 % and 12.1 % for pristine CsSnI3, MASn(I0.75Br0.25)3 and FASn(I0.75Br0.25)3 with absorber layer thicknesses of 400 nm, 300 nm and 600 nm respectively. These results highlight the potential of our computational approach in predicting the impact of perovskite stoichiometry on the device performance as a function of modifications to the electronic structure.