Design Automation Of Microfluidic Single And Double Emulsion Droplets With Machine Learning

This article presents a comprehensive experimental dataset of both SE and DE droplets comprised of many different fluids to train models that accurately predict droplet diameter and generation rate across a diverse range of fluid properties, geometries, flow rates, and device surface properties. Additionally, the article demonstrates that the models generalize to additional device geometries, fluids, and materials by experimentally validating “blind” predictions. Finally, the article introduces DAFD 3.0, an automated search algorithm to create a design automation tool for SE and DE droplets that can return the necessary design and flow rates to achieve the user-specified diameter and rate for different fluids.