Exploring Inorganic Materials Synthesis Prediction With High Dimensional Phase Diagrams.

다국어 초록 (Multilingual Abstract)

Why are certain compounds synthesizable while others are not? This question represents a fundamental inquiry in the field of solid-state chemistry. It also serves as a central focus within the realm of computational materials prediction, aligning closely with the objectives of the Materials Genome Initiative. This dissertation primarily focuses on the prediction of material synthesis with a dual-pronged approach. Firstly, it extensively explores material stability by constructing high-dimensional phase diagrams driven by fundamental thermodynamics. This process aims to generate phase diagrams for complex experimental synthesis conditions, and offers a comprehensive visual thermodynamic representation of material stability. Additionally, this dissertation delves into targeted material synthesis, utilizing high dimensional phase diagram to enhance target material stability or design efficient synthesis recipes. These approaches accelerate the realization of theoretically predicted materials and guides the process of high-throughput robotic experimental synthesis, ultimately advancing our understanding and capabilities in materials synthesis. Phase diagrams are crucial tools for materials scientists, indicating the equilibrium phases under specific thermodynamic conditions. While most phase diagrams are two-dimensional, with axes typically representing temperature-pressure or temperature-composition, the complexity of modern materials demands consideration of additional thermodynamic factors, such as elastic, surface, electromagnetic, or electrochemical work. This expansion necessitates phase diagrams in higher dimensions (≥3). In our pursuit of constructing high-dimensional phase diagrams with any thermodynamic variable on its axes, we explore the duality between extensive and intensive conjugate variables in equilibrium and non-equilibrium thermodynamics. This duality takes various forms, including distinctions between closed and open boundary conditions, relationships between Internal Energy and its Legendre transformations, and the point-line duality in convex hulls versus half-space intersections. Specifically, we derive the duality relationships for chemical work involving extensive composition variables (N) and intensive chemical potentials (μ). Designing thermodynamic conditions to enhance or diminish the stability of a target material is a crucial task in materials engineering. For instance, during materials synthesis, the objective is often to increase the stability of a target phase relative to its precursors or competing byproduct phases. To facilitate this, we introduced a generalized Clausius-Clapeyron relation, guiding the identification of optimal directions on a high-dimensional phase diagram for stabilizing or destabilizing a target phase. Using this approach, we analyze the acid stability of manganese oxide catalysts through a 4-dimensional Pourbaix diagram with axes representing pH, redox potential, nanoparticle size, and aqueous potassium ion concentration. Additional discussions on Pourbaix diagrams with varying natural variables and in different solvents like aqueous solutions, supercritical ammonia and ethanol contribute to the broadening of Pourbaix diagram applications. Efficient synthesis is essential for realizing predicted materials and producing complex ones. However, solid-state synthesis of multicomponent oxides often encounters challenges from undesired by-product phases, which can stall reactions kinetics. Here, we present a thermodynamic strategy to navigate high-dimensional phase diagrams, seeking precursors that avoid low-energy competing by-products and maximize reaction energy for rapid kinetics. Validating this strategy using a robotic inorganic materials synthesis laboratory, we find our predicted precursors often yield purer target materials than traditional ones. Robotic labs offer a data-driven platform for experimental synthesis science, guiding both human and robotic chemists.