Application of machine learning to nanomaterial design with silver nanoprisms

Abstract Achieving reproducible synthesis of nanomaterials with tunable optical and morphological properties remains a central challenge in materials design. Conventional trial-and-error strategies struggle with the nonlinear transitions governing nanoparticle growth, often limiting control over plasmonic responses. Here, we introduce a convolutional neural network (CNN) framework that couples in situ time-resolved UV–Vis spectroscopy with the synthesis of silver nanoprisms, extracting predictive rules for morphology and optical behavior. By leveraging transient spectral dynamics rather than endpoint data alone, the model captures hidden growth pathways and accurately predicts final size distributions and plasmonic signatures from a modest experimental dataset. This machine-learning-assisted methodology integrates directly into synthesis workflows, reducing experimental burden while enhancing reproducibility. Beyond silver nanoprisms, the strategy provides a transferable route for rational design of nanomaterials with tailored optical functionalities, advancing the broader goal of data-driven materials design.