STM Image Analysis using Autoencoders

TL;DR

This paper investigates the use of convolutional autoencoders to analyze and reconstruct STM images of various crystalline structures, demonstrating deep learning's potential in atomic-scale image analysis.

Contribution

The study introduces two CAE architectures trained on simulated STM images, evaluating their performance and exploring latent space representations for materials science applications.

Findings

CAE models achieve low MSE and high SSIM scores

Latent space captures meaningful features of lattice structures

Challenges remain in full image reconstruction and interpretability

Abstract

This study explores the application of Convolutional Autoencoders (CAEs) for analyzing and reconstructing Scanning Tunneling Microscopy (STM) images of various crystalline lattice structures. We developed two distinct CAE architectures to process simulated STM images of simple cubic, body-centered cubic (BCC), face-centered cubic (FCC), and hexagonal lattices. Our models were trained on $17\times17$ pixel patches extracted from $256\times256$ simulated STM images, incorporating realistic noise characteristics. We evaluated the models' performance using Mean Squared Error (MSE) and Structural Similarity (SSIM) index, and analyzed the learned latent space representations. The results demonstrate the potential of deep learning techniques in STM image analysis, while also highlighting challenges in latent space interpretability and full image reconstruction. This work lays the foundation for future advancements in automated analysis of atomic-scale imaging data, with potential applications in materials science and nanotechnology.