Toward autonomous materials synthesis via reaction–diffusion coupling

Autonomous experimentation presents enormous opportunities for materials synthesis under far-from-equilibrium conditions due to the vast parameter space that can be controlled, the complexity of the emerging products, and the potential for machine learning to drive synthesis beyond traditional methods. This work presents important steps toward such autonomous materials synthesis via reaction–diffusion coupling. As a model system, we used the periodic precipitation of copper(II) hydroxide and targeted the synthesis of “Liesegang” precipitation bands with a well-defined spacing. Our workflow involved conducting high-throughput synthesis experiments, mapping the spatial distribution of the products as a function of initial reactant concentrations, and scalarizing the emerging patterns using empirical scaling laws. Subsequently, machine learning models were deployed to process the scalarized descriptors and inform the next round of experimental conditions in order to converge toward a target precipitation pattern, without human input. We anticipate that these results open pathways for autonomous materials synthesis via reaction–diffusion coupling to create complex products with user-defined chemistry, morphology, and spatial distribution within a reaction medium.