An Autonomous Lab Approach for Stabilizing Perovskite Thin Films Across Scales

Halide perovskite thin films hold great potential for energy conversion and quantum technologies, but their instability under real-world conditions remains a significant challenge. This talk introduces an autonomous lab approach that accelerates the discovery and stabilization of perovskite thin films by integrating robotic synthesis, machine learning (ML), and computer vision. The approach is designed to optimize synthesis experiments, enabling precise control over the deposition process and the heterogeneity of the films. A key feature of this approach is accelerated characterization using computer vision, which provides rapid insights into thin film morphology, phase transitions, and degradation behavior. Additionally, explainable ML models are employed to predict structure-property relationships, enabling a deeper understanding of the underlying factors that influence material stability. These models guide the search for stable compositions by linking film structures to their performance characteristics. The design of stable 2D/3D heterostructures in thin films is also a critical focus, offering new pathways to enhance the long-term performance of perovskites. By combining automated experimentation with explainable ML, this approach allows for the systematic exploration of the chemical space, paving the way for breakthroughs in perovskite thin film technologies for energy conversion and quantum devices.