Diffusion-Limited Solid-State Synthesis Selectivity in the Presence of Oxidation/Reduction Processes

Abstract Solid-state synthesis outcomes arise from an interplay between thermodynamics and kinetics, with their relative importance governed by external control parameters such as temperature profiles and precursor selection. A recently introduced framework, combining first-principles thermodynamics with machine-learning-enabled simulations of correlated ion transport, successfully captured this interplay in the Ba–Ti–O system by linking phase selectivity to ion mobility within “liquid-like”, amorphous interphases. Notably, in systems where valence states can change during synthesis, ion dynamics are strongly coupled to oxidation state. Here, we extend the framework to such mixed-valence systems and demonstrate its predictive power for the synthesis of ternary metal ferrites in the Ba–Fe–O and Y–Fe–O systems, benchmarked against in situ XRD experiments. Similar to the Ba–Ti–O system, we hypothesize that the rate-limiting transport of ionic species governing polymorph growth occurs through a “liquid-like,” noncrystalline interphase. To capture this behavior, we employ a charge-informed machine-learning potential (CHGNet) that enables modeling of correlated, redox-dependent transport. We find that the framework accurately predicts the sequence of phase formation and the time evolution of phase distributions in powder synthesis reactions, in agreement with experimental reports. These results show that diffusion-limited phase selectivity can be treated on a unified footing across mixed-valence oxides, further clearing a pathway toward rational, in silico synthesis design that simultaneously accounts for thermodynamic and kinetic factors.