Toward Controlled Synthesis of 2D Crystals by CVD: Learning from the Real-Time Crystal Morphology Evolutions.

Abstract

The rich morphology of 2D materials grown through chemical vapor deposition (CVD), is a distinctive feature. However, understanding the complex growth of 2D crystals under practical CVD conditions remains a challenge due to various intertwined factors. Real-time monitoring is crucial to providing essential data and enabling the use of advanced tools like machine learning for unraveling these complexities. In this study, we present a custom-built miniaturized CVD system capable of observing and recording 2D MoS2 crystal growth in real time. Image processing converts the real-time footage into digital data, and machine learning algorithms (ML) unveil the significant factors influencing growth. The machine learning model successfully predicts CVD growth parameters for synthesizing ultralarge monolayer MoS2 crystals. It also demonstrates the potential to reverse engineer CVD growth parameters by analyzing the as-grown 2D crystal morphology. This interdisciplinary approach can be integrated to enhance our understanding of controlled 2D crystal synthesis through CVD.