Determining the orderliness of carbon materials with nanoparticle imaging and explainable machine learning

Area of Science:

• Materials Science
• Nanotechnology
• Catalysis

Background:

• Carbon materials are crucial in diverse applications, including electronics and catalysis.
• Material properties are significantly influenced by surface defects, which are irregularities in electron density.
• Palladium nanoparticle deposition followed by scanning electron microscopy (SEM) imaging visualizes these defects.

Purpose of the Study:

• To develop a machine learning (ML) approach for analyzing SEM images of carbon materials.
• To distinguish between ordered and disordered defect arrangements based on nanoparticle positioning.
• To evaluate the impact of defect ordering on catalytic performance.

Main Methods:

• Utilized a machine learning model for image analysis of SEM data.
• Characterized defect distribution by analyzing the ordering of deposited palladium nanoparticles.
• Experimentally assessed the influence of defect ordering on carbon-carbon bond formation catalysis.

Main Results:

• Developed a highly interpretable ML approach for classifying defect ordering in carbon materials.
• Demonstrated a correlation between the degree of defect ordering and catalytic efficiency.
• Successfully distinguished between materials with ordered and disordered defect arrangements.

Conclusions:

• The developed ML method enables automated analysis of SEM images for defect characterization.
• Understanding defect ordering is key to optimizing carbon material properties for applications like catalysis.
• This work advances the automated analysis of microstructural features in materials science.