Human-in-the-Loop Enhances Autonomous Polymer Phase Mapping

Autonomous experimentation is a promising framework for conducting experimental campaigns, as it can improve the pace of research and increase resource utilization efficiency. However, beyond the challenge of creating fully autonomous systems without human intervention, their effectiveness can be limited by slow experimental steps. For polymer blends that require long annealing times, lasting hours or even days, the time saved by automating small tasks like drop casting or image analysis is only a small part of the overall time needed for sample preparation, processing, and characterization. To address these challenges, we propose a "human-in-the-loop" (HITL) framework for studying the miscibility of polymer blends and nanocomposites. Utilizing previously computed parameters for Gaussian Process-informed Bayesian Optimization, this HITL framework can be layered on top of experimental techniques already available in common laboratories, such as drop casting, oven annealing, and optical microscopy. We use a previously studied blend of poly(methyl methacrylate) and poly(styrene-ran-acrylonitrile) to validate the performance of this HITL approach and demonstrate its potential applicability to unknown or unstudied polymer blend and nanocomposite systems.