A graph-based network for predicting chemical reaction pathways in solid-state materials synthesis
Abstract Accelerated synthesis of inorganic materials remains a significant challenge in the search for novel, functional materials. Many of the chemical principles which enable "synthesis by design" in synthetic organic chemistry do not exist in solid-state chemistry, despite extensive computed/experimental thermochemistry data. We present a chemical reaction network model constructed from thermochemistry databases that captures features of the thermodynamic phase space which synthesis reactions traverse. Directed edges in the network are assigned weights via a transformation that maps reaction parameters to costs. We devise a computationally tractable approach for suggesting likely reaction pathways via application of pathfinding algorithms and linear combination of lowest-cost paths in the network. We demonstrate initial success of the reaction network in predicting a complex metathesis reaction pathway toward yttrium manganese oxide YMnO3. The reaction network presents new opportunities for enabling reaction pathway prediction, rapid iteration between experimental/theoretical results, and ultimately, control of synthesis of solid-state materials.