AutoBot—Automated Thin Film Semiconductor Synthesis and Characterization Towards Closed Loop Platforms

The development of AutoBot, a robot-assisted liquid handling platform for synthesizing and characterizing thin film materials, will be presented. This platform automates thin film synthesis from chemical precursor solutions through spin coating and annealing, followed by in-line and off-line characterizations. Heterogeneous data representing spectroscopic, structural, and morphological properties are collected through transmission measurements, photoluminescence (PL) spectra/image collection, X-ray diffraction, and scanning electron microscopy. In pursuing machine learning (ML)-driven and closed-loop experiments, our team has encountered several questions and challenges. I will begin by discussing some initial successes and lessons learned when designing closed-loop experiments. Initial benchmarking experiments of the AutoBot platform are used to assess consistency and reproducibility, comparing robot performance with that of a human operator. These experiments also demonstrate the benefit of systematic parameter screening in identifying relevant thin film fabrication variables to enhance reproducibility. The model system for workflow development is metal halide perovskite semiconductors, chosen for their rich compositional flexibility and process tunability.