Closed-loop optimization of nanoparticle synthesis enabled by robotics and machine learning

Abstract

Colloidal nanoparticles are attractive materials for various energy and chemical applications. Due to their strictly tunable struc-ture-function relationships, reproducibly synthesizing structurally homogeneous nanoparticles is a critical step toward making the nanoparticle technology commercially viable. However, due to a lack of general theoretical foundations for complex nanoparticle formation phenomena, the current synthesis optimizations of nano -particles are mostly conducted based on the intuitions and trial-and -error-driven manual processes that are slow to explore a large synthesis parameter space. To accelerate these time-consuming and resource-demanding conventional synthesis approaches, we here describe a closed-loop pipeline that consists of robotic synthe-sis, automated materials characterization, machine-learning optimi-zation, and computational prediction of desired structure-property relationships. We discuss the need and the current levels of automa-tion in different parts of nanoparticle synthesis experiments with future directions and outlook.