Abstract

Design of materials in a high-throughput setting requires modeling approaches that can provide rapid assessments to support real-time decisions. Machine learning (ML) emerges as an excellent tool to support such models. While ML models may lack the accuracy and insights delivered by full physics-based computational solutions, these approaches are often perfectly sufficient to identify material chemistries and microstructures that merit further exploration. In particular, these data-driven ML models are a very sensible approach in the context of a high-throughput experimentation paradigm that provides lower quality data in statistically significant quantities. By providing rapid assessments to support screening of new materials, ML models support real-time decision-making for control and optimization of high-throughput processes on the path to materials design. A current facility in this regard is under development at Johns Hopkins University – the AI for Materials Design (AIMD) facility, which highlights some of the challenges, pitfalls and opportunities inherent in an integrated high-throughput and automated materials design framework. The role of ML models in guiding this automated materials design is highlighted and discussed in the context of a few example applications.