Deciphering alloy composition in superconducting single-layer FeSe1-xSx on SrTiO3(001) substrates by machine learning of STM/S data

TL;DR

This paper demonstrates how machine learning applied to STM data can accurately determine local alloy composition and reveal nanoscale chemical inhomogeneity's impact on superconductivity in FeSe1-xSx films.

Contribution

It introduces a machine learning approach combining K-means clustering and singular value decomposition to analyze STM data for alloy composition determination.

Findings

Accurate local Se/S composition identification in FeSe1-xSx films.

Revealed correlations between chemical inhomogeneity and superconductivity.

Provided a reliable method for analyzing multi-component alloys with STM data.

Abstract

Scanning tunneling microscopy (STM) is a powerful technique for imaging atomic structure and inferring information on local elemental composition, chemical bonding, and electronic excitations. However, traditional methods of visual inspection can be challenging for such determination in multi-component alloys, particularly beyond the dilute limit due to chemical disorder and electronic inhomogeneity. One viable solution is to use machine learning to analyze STM data and identify patterns and correlations that may not be immediately apparent through visual inspection alone. Here, we apply this approach to determine the Se/S concentration in superconducting single-layer FeSe1-xSx alloy epitaxially grown on SrTiO3(100) substrate by molecular beam epitaxy. First, defect-related dI/dV tunneling spectra are identified by the K-means clustering method, followed by singular value decomposition to distinguish between those from S and Se. Such analysis provides an efficient and reliable determination of local elemental composition, and further reveals correlations of nanoscale chemical inhomogeneity to superconductivity in single-layer iron chalcogenide films.