Autonomous synthesis of thin films with pulsed laser deposition using in situ Raman and reflectance spectroscopy

Synthesis of thin films is traditionally a slow, sequential processes carried out with significant human effort and intuition with the task of optimizing some material property. Recent advances in autonomous systems that combine artificial intelligence with synthesis and characterization are enabling the exploration of large parameter spaces at rates beyond what is possible by traditional, human-driven experiments which greatly accelerates the discovery or optimization of materials. Here, we demonstrate autonomous synthesis of thin film materials with pulsed laser deposition (PLD) enabled by combining automation, in situ Raman and reflectance spectroscopy, and machine learning. We grow monolayer WSe₂ thin films with PLD and assess their quality with in situ Raman spectroscopy. Bayesian optimization with Gaussian process regression is used to determine the sequence of experiments to maximize the quality of the WSe₂ films. The autonomous routine showed a 10x increase in throughput over traditional PLD workflows and discovered two distinct growth windows after sampling only 0.25% of a 4D parameter space. The autonomous workflow presented and the PLD platform we developed is applicable to any material that can be grown by PLD.