A paper from the Inorganic Crystal Chemistry Laboratory (headed by Associate Professor WAGATA Hajime) and the Laboratory of Data and Chemical Engineering (headed by Associate Professor KANEKO Hiromasa) graced the cover of Volume 64, Issue 12 (2025) of the international academic journal Industrial & Engineering Chemistry Research.
In this study, the experimental conditions for synthesizing LaFeO3 crystals with desired properties were optimized using Bayesian optimization. Initially, candidate experimental conditions were determined based on the design of experiments. LaFeO3 crystals were synthesized under these conditions, and their properties were analyzed. A Gaussian process regression model was constructed using the experimental conditions as explanatory variables and the crystal properties as objective variables. This model predicted the properties of new candidate conditions, and an acquisition function was calculated to select the most promising candidates. The process of model construction, candidate selection, and synthesis was repeated to achieve the desired crystal properties. Three case studies demonstrated the successful synthesis of LaFeO3 crystals with small crystallite sizes in a limited number of experiments. Additionally, an index to evaluate impurities was introduced, allowing for multiobjective optimization that simultaneously reduced impurity formation and crystallite size.
≪Japanese version≫
