Quantifying the CVD-grown two-dimensional materials via image clustering.

Area of Science:

• Materials Science
• Nanotechnology
• Artificial Intelligence

Background:

• Chemical vapor deposition (CVD) is a key method for synthesizing high-quality two-dimensional (2D) materials.
• Manual quality assessment of CVD-grown 2D materials from optical images is labor-intensive and time-consuming.
• Machine learning (ML) offers potential for automating material quality evaluation.

Purpose of the Study:

• To develop an unsupervised ML strategy for automated quality assessment of CVD-grown 2D materials using optical images.
• To integrate Self-Organizing Map (SOM) and k-means clustering for image analysis.
• To provide an efficient tool for materials scientists to evaluate material quality.

Main Methods:

• Utilized an unsupervised machine learning approach for image clustering.
• Integrated Self-Organizing Map (SOM) and k-means algorithms for optical image analysis.
• Applied the methodology to optical images of CVD-grown 2D materials.

Main Results:

• The proposed unsupervised clustering algorithm demonstrated high accuracy, with results closely matching expert labels.
• The data-driven strategy effectively categorized the quality of CVD-grown 2D materials.
• The method proved to be an efficient tool for quality assessment.

Conclusions:

• The developed unsupervised ML methodology offers an effective toolkit for evaluating CVD-grown 2D materials.
• This approach significantly reduces the time and labor associated with manual quality assessment.
• The methodology has broad applicability across various material systems and synthesis techniques.