How to accelerate the inorganic materials synthesis: from computational guidelines to data-driven method?

The development of novel functional materials has attracted widespread attention to meet the constantly growing demand for addressing the major global challenges facing humanity, among which experimental synthesis emerges as one of the crucial challenges. Understanding the synthesis processes and predicting the outcomes of synthesis experiments are essential for increasing the success rate of experiments. With the advancements in computational power and the emergence of machine learning (ML) techniques, computational guidelines and data-driven methods have significantly contributed to accelerating and optimizing material synthesis. Herein, a review of the latest progress on the computation-guided and ML-assisted inorganic material synthesis is presented. First, common synthesis methods for inorganic materials are introduced, followed by a discussion of physical models based on thermodynamics and kinetics, which are relevant to the synthesis feasibility of inorganic materials. Second, data acquisition, commonly utilized material descriptors, and ML techniques in ML-assisted inorganic material synthesis are discussed. Third, applications of ML techniques in inorganic material synthesis are presented, which are classified according to different material data sources. Finally, we highlight the crucial challenges and promising opportunities for ML-assisted inorganic materials synthesis. This review aims to provide critical scientific guidance for future advancements in ML-assisted inorganic materials synthesis.