Charting the thermodynamic stability of hybrid perovskite alloys with machine learning

Abstract

Alloy-based perovskite semiconductor materials offer tunable properties and improved stability, but their complexity has impeded accurate modeling, hindering development of solar cell devices. We present a machine-learning (ML) accelerated atomistic modeling approach for the phase stability of (Cs/FA)Pb(Br/I)3 and (Cs/FA)Sn(Br/I)3 perovskites (where FA is formamidinium), that explicitly accounts for finite-temperature entropy effects. To make such quaternary alloy calculations tractable, we adopt a two-level strategy, combining 1) graph neural network interatomic potentials trained on density functional theory data for efficient structure relaxations with 2) direct energy predictions from unrelaxed structures to further accelerate the process. Our strategy enables computations of free energy landscapes across compositions and phases, capturing alloy disorder and molecular orientations of FA. Our results reveal narrower regions of stable compositions for the Sn-based system compared to its Pb-based counterpart, limiting compositional engineering strategies for realizing devices that are robust in long term. High I contents result in maximum stability and no stabilization is observed near the center of the composition space. Our results guide the design of stable perovskite-based devices in experiments and demonstrate a workflow that is generalizable to other complex alloy modeling tasks.