Predicting Lattice Parameters from Atomic-Scale Images of Two Dimensional Materials Using Deep Learning

Published

December 15, 2025

Author(s)

Sayak Chakrabarty, Kamal Choudhary, Youjia Li, Daniel Wines, Vishu Gupta, Muhammed Nur Talha Kilic, Alok Choudhary, Ankit Agrawal

Abstract

Determining lattice parameters in two-dimensional (2D) materials is essential for materials characterization and discovery. In this work, we propose a deep-learning-driven pipeline that addresses the regression task of estimating the lattice constants (a) and (b) and the angle gamma directly from 2D images using computer vision. We evaluate our approach on three different 2D material datasets: JARVIS-2D (JV2D) and Computational 2D Materials Database (C2DB), and a newly created dataset derived from the Alexandria database. Multiple architectures are compared, including DenseNet121, Vision Transformers (ViT-L/14) paired with Multi Layer Perceptron (MLP) heads, and GoogleNet. DenseNet121 achieves accurate performance, with mean absolute errors as low as 0.18 Angstrom for the Alexandria-based dataset and 0.17 Angstrom for C2DB and 0.59 Angstrom for JV2D, as well as up to 96 % accuracy in classifying Bravais lattice types for the Alexandria-based dataset.

Citation

Journal of Physical Chemistry C

Pub Type

Journals

Download Paper

https://doi.org/10.1021/acs.jpcc.5c04792

Local Download

Citation

Chakrabarty, S. , Choudhary, K. , Li, Y. , Wines, D. , Gupta, V. , KILIC, M. , Choudhary, A. and Agrawal, A. (2025), Predicting Lattice Parameters from Atomic-Scale Images of Two Dimensional Materials Using Deep Learning, Journal of Physical Chemistry C, [online], https://doi.org/10.1021/acs.jpcc.5c04792, https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=960024 (Accessed April 18, 2026)