Machine Learning-Assisted Large-Area Preparation of MoS₂ Materials.

Area of Science:

• Materials Science
• Semiconductor Physics
• Computational Materials Science

Background:

• Molybdenum disulfide (MoS2) is a promising semiconductor for optoelectronics and integrated circuits due to its favorable properties and low cost.
• Chemical vapor deposition (CVD) enables high-quality, large-size MoS2 synthesis, but its complex process and difficult area control pose challenges.
• Machine learning (ML) offers powerful tools for materials science, including exploring complex synthesis mechanisms.

Purpose of the Study:

• To develop and apply a machine learning Gaussian regression model to understand the CVD growth mechanism of MoS2.
• To identify key growth parameters influencing MoS2 synthesis and optimize the CVD process for large-size material production.
• To provide an efficient, cost-effective solution for MoS2 material preparation by minimizing trial-and-error.

Main Methods:

• Construction of a machine learning Gaussian regression model to analyze MoS2 growth parameters.
• Evaluation of model performance using goodness of fit (r2), mean squared error (MSE), and Pearson correlation coefficient (p-value).
• Feature importance analysis to determine the impact of parameters like carrier gas flow rate (Fr), molybdenum sulfur ratio (R), and reaction temperature (T) on MoS2 growth.

Main Results:

• The ML model achieved optimal performance with 15 iterations.
• Carrier gas flow rate (Fr), molybdenum sulfur ratio (R), and reaction temperature (T) were identified as critical factors for CVD growth of MoS2.
• The model accurately predicted MoS2 synthesis size across 185,900 conditions, identifying optimal ranges for large-size material production.
• Model predictions showed minimal relative error when validated against literature and experimental data.

Conclusions:

• Machine learning provides an effective approach to optimize the CVD synthesis of MoS2.
• The developed ML model successfully predicts MoS2 growth and identifies key parameters, enabling efficient production of large-size materials.
• This study offers a significant reduction in time and cost for MoS2 material preparation, accelerating its application in electronics.