Atomistic Simulations of Thermal and Chemical Expansions of PrNi_xCo_1-xO_3-δ Accelerated by Machine Learning Potentials

The electrodes and solid-state electrolytes in protonic ceramic electrochemical cells (PCECs) experience significant lattice expansions when exposed to high steam concentrations at elevated temperatures. In this paper, phonon calculations based on a new machine learning potential (MLP) are employed to elucidate the volume expansions of the proton-conducting PrNi_xCo_1-xO_3-δ (PNC) lattices, manifested under a combined influence of oxygen vacancies ( $V_O^{··}$ ) and proton uptake ( ${\mathrm{OH}}_O^{·}$ ) in the bulk at varying Ni/Co occupancies. It is revealed that the Ni/Co occupancy contributes to thermal and chemical expansions differently, where thermal expansions are related to Co occupancy. In contrast, chemical expansions are more closely associated with the Ni occupancy. Both $V_O^{··}$ and ${\mathrm{OH}}_O^{·}$ lead to higher thermal expansions when compared to the pristine PNC. The temperature increase will negatively impact the hydration-induced chemical expansions. For combined thermal and chemical expansions, it is predicted that the strategies that boost the PCEC's electrochemical performance may harm the electrode-electrolyte interfacial stability, when the Ni occupancy is high, due to severe chemical expansions. Mitigating chemical expansions of the Ni-abundant PNC will benefit the interfacial stability. The presented computational methods for phonon calculations, based on emerging machine learning interatomic potential techniques are anticipated to have a lasting impact on future PCEC development.