Download all files (15.4 MB)

Collection

Citation

Description:

Spatially-resolved ARPES is a powerful tool for probing local electronic structures in low-dimensional materials, but its analysis becomes challenging for spatially inhomogeneous samples due to spectral variations, feature overlap, and minor shifts. Here, we present a practical framework based on non-negative matrix factorization (NMF), which decomposes ARPES spectra into physically interpretable components without relying on prior assumptions. Visualizing the activation matrix as spatial heatmaps reveals latent spectral structures and provides an intuitive map of how individual components are distributed, enabling identification of the domains and local electronic variations. We validate this framework using epitaxial graphene on SiC, demonstrating its ability to quantitatively disentangle spectral features associated with layer thickness, step structures, and growth conditions. This study establishes the NMF-based framework as a scalable and robust tool for managing large-scale datasets and assessing electronic inhomogeneity in low-dimensional materials.

Rights:

• Creative Commons BY Attribution 4.0 International
  

Keyword: Photoemission spectroscopy, non-negative matrix factorization, graphene, machine learning, data-driven analysis

Date published: 2026-06-19

Publisher: Taylor & Francis

Journal:

• Science and Technology of Advanced Materials (ISSN: 27660400) vol. 6 2688747

Funding:

Manuscript type: Author's version (Accepted manuscript)

MDR DOI: https://doi.org/10.48505/nims.6365

First published URL: https://doi.org/10.1080/27660400.2026.2688747

Related item:

Other identifier(s):

Contact agent:

Updated at: 2026-06-24 15:38:41 +0900

Published on MDR: 2026-06-24 18:27:28 +0900