Deep Learning Analysis of Polaritonic Wave Images

Area of Science:

• Materials Science
• Physics
• Data Science

Background:

• Deep learning (DL) shows promise for analyzing large scientific datasets.
• Nanoscale imaging generates complex data requiring advanced analysis techniques.

Purpose of the Study:

• To apply DL to analyze nanoscale images of polaritonic waves.
• To develop a rapid method for quantifying polaritonic wave characteristics and material parameters.

Main Methods:

• Utilized a convolutional neural network (CNN) trained on simulated near-field images.
• Developed a protocol for rapid image regression to extract quantitative data.
• Validated the CNN-based analysis on experimental images of graphene/α-RuCl3 interfaces.

Main Results:

• The CNN accurately quantifies polaritonic wave wavelength and quality factor.
• The method achieves analysis speeds at least 3 orders of magnitude faster than conventional procedures.
• Successfully extracted polaritonic characteristics and material parameters from experimental data.

Conclusions:

• DL offers a powerful and efficient framework for analyzing nanoscale imaging data.
• The developed CNN protocol provides a generalizable method for scanning probe microscopy.
• This approach accelerates the quantitative extraction of physical information from complex material images.