Bayesian Optimization\nfor Controlled Chemical Vapor\nDeposition Growth of WS<sub>2</sub>

We applied Bayesian optimization (BO), a machine learning\n(ML)\ntechnique, to optimize the growth conditions of monolayer WS<sub>2</sub> using photoluminescence (PL) intensity as the objective function.\nThrough iterative experiments guided by BO, an improvement of 86.6%\nin PL intensity is achieved within 13 optimization rounds. Statistical\nanalysis revealed the relationships between growth conditions and\nPL intensity, highlighting the importance of critical conditions,\nincluding the tungsten source concentration and Ar flow rate. Furthermore,\nthe effectiveness of BO is demonstrated by comparison with random\nsearch, showing its ability to converge to optimal conditions with\nfewer iterations. This research highlights the potential of ML-driven\napproaches in accelerating material synthesis and optimization processes,\npaving the way for advances in two-dimensional (2D) material-based\ntechnologies.