Continuous Flow Synthesis of Prussian Blue and Analogues Assisted by AI

Abstract Prussian blue and its analogs are promising materials for numerous applications. Interest in this class of materials arises from their broad pore distribution, redox properties, high biocompatibility, low‐cost components, straightforward manufacturability, and adaptability through analog development. A key challenge is the synthesis of well‐defined, small‐dimensioned materials using machine learning approaches. This study presents a strategy to address this limitation via machine learning‐driven microfluidic synthesis. Employing unsupervised Bayesian optimization with Gaussian processes effectively reduces optimization time and minimizes the need for prior knowledge. As a proof of concept, Prussian blue, and cobalt‐based analogs are synthesized, with UV–vis spectroscopy providing feedback for the machine learning algorithm. The optimized protocols are subsequently applied to larger‐scale preparations, demonstrating that the standardized methods have the potential for the commercial production of high‐quality materials. Comprehensive characterization of the materials confirms their cubic morphology, small dimensionality, and mixed‐valency of the metal elements.