Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopy

Abstract Atomic arrangements and local sub-structures fundamentally influence emergent material functionalities. These structures are conventionally probed using spatially resolved studies and the property correlations are deciphered by a researcher based on sequential explorations, thereby limiting...

Full description

Saved in:
Bibliographic Details
Main Authors: Ganesh Narasimha, Dejia Kong, Paras Regmi, Rongying Jin, Zheng Gai, Rama Vasudevan, Maxim Ziatdinov
Format: Article
Language:English
Published: Nature Portfolio 2025-06-01
Series:npj Computational Materials
Online Access:https://doi.org/10.1038/s41524-025-01642-1
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850207488028180480
author Ganesh Narasimha
Dejia Kong
Paras Regmi
Rongying Jin
Zheng Gai
Rama Vasudevan
Maxim Ziatdinov
author_facet Ganesh Narasimha
Dejia Kong
Paras Regmi
Rongying Jin
Zheng Gai
Rama Vasudevan
Maxim Ziatdinov
author_sort Ganesh Narasimha
collection DOAJ
description Abstract Atomic arrangements and local sub-structures fundamentally influence emergent material functionalities. These structures are conventionally probed using spatially resolved studies and the property correlations are deciphered by a researcher based on sequential explorations, thereby limiting the efficiency and scope. Here we demonstrate a multi-scale Bayesian deep-learning based framework that automatically correlates material structure with its electronic properties using scanning tunneling microscopy (STM) measurements in real-time. Its predictions are used to autonomously direct exploration toward regions of the sample that optimize a given material property. This method is deployed on a low-temperature ultra-high vacuum STM to understand the structure-property relationship in a europium-based semimetal, EuZn2As2, a promising candidate relevant to magnetism-driven topological phenomena. The framework employs a sparse-sampling approach to efficiently construct the scalar-property space using minimal measurements, about 1–10% of the data required in standard hyperspectral methods. Moreover, we formulate the problem hierarchically across length scales, implementing autonomous workflow to locate mesoscopic and atomic structures that correspond to a target material property. This framework offers the choice to design scalar-property from the spectroscopic data to steer sample exploration. Our findings reveal correlations of the electronic properties unique to surface terminations, local defect density, and point defects.
format Article
id doaj-art-04a72b0d13d94287b1028d95d70e3c1e
institution OA Journals
issn 2057-3960
language English
publishDate 2025-06-01
publisher Nature Portfolio
record_format Article
series npj Computational Materials
spelling doaj-art-04a72b0d13d94287b1028d95d70e3c1e2025-08-20T02:10:31ZengNature Portfolionpj Computational Materials2057-39602025-06-0111111010.1038/s41524-025-01642-1Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopyGanesh Narasimha0Dejia Kong1Paras Regmi2Rongying Jin3Zheng Gai4Rama Vasudevan5Maxim Ziatdinov6Center for Nanophase Material Sciences (CNMS), Oak Ridge National Laboratory (ORNL)Center for Nanophase Material Sciences (CNMS), Oak Ridge National Laboratory (ORNL)SmartState Center for Experimental Nanoscale Physics, Department of Physics and Astronomy, University of South CarolinaSmartState Center for Experimental Nanoscale Physics, Department of Physics and Astronomy, University of South CarolinaCenter for Nanophase Material Sciences (CNMS), Oak Ridge National Laboratory (ORNL)Center for Nanophase Material Sciences (CNMS), Oak Ridge National Laboratory (ORNL)Center for Nanophase Material Sciences (CNMS), Oak Ridge National Laboratory (ORNL)Abstract Atomic arrangements and local sub-structures fundamentally influence emergent material functionalities. These structures are conventionally probed using spatially resolved studies and the property correlations are deciphered by a researcher based on sequential explorations, thereby limiting the efficiency and scope. Here we demonstrate a multi-scale Bayesian deep-learning based framework that automatically correlates material structure with its electronic properties using scanning tunneling microscopy (STM) measurements in real-time. Its predictions are used to autonomously direct exploration toward regions of the sample that optimize a given material property. This method is deployed on a low-temperature ultra-high vacuum STM to understand the structure-property relationship in a europium-based semimetal, EuZn2As2, a promising candidate relevant to magnetism-driven topological phenomena. The framework employs a sparse-sampling approach to efficiently construct the scalar-property space using minimal measurements, about 1–10% of the data required in standard hyperspectral methods. Moreover, we formulate the problem hierarchically across length scales, implementing autonomous workflow to locate mesoscopic and atomic structures that correspond to a target material property. This framework offers the choice to design scalar-property from the spectroscopic data to steer sample exploration. Our findings reveal correlations of the electronic properties unique to surface terminations, local defect density, and point defects.https://doi.org/10.1038/s41524-025-01642-1
spellingShingle Ganesh Narasimha
Dejia Kong
Paras Regmi
Rongying Jin
Zheng Gai
Rama Vasudevan
Maxim Ziatdinov
Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopy
npj Computational Materials
title Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopy
title_full Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopy
title_fullStr Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopy
title_full_unstemmed Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopy
title_short Uncovering multiscale structure-property correlations via active learning in scanning tunneling microscopy
title_sort uncovering multiscale structure property correlations via active learning in scanning tunneling microscopy
url https://doi.org/10.1038/s41524-025-01642-1
work_keys_str_mv AT ganeshnarasimha uncoveringmultiscalestructurepropertycorrelationsviaactivelearninginscanningtunnelingmicroscopy
AT dejiakong uncoveringmultiscalestructurepropertycorrelationsviaactivelearninginscanningtunnelingmicroscopy
AT parasregmi uncoveringmultiscalestructurepropertycorrelationsviaactivelearninginscanningtunnelingmicroscopy
AT rongyingjin uncoveringmultiscalestructurepropertycorrelationsviaactivelearninginscanningtunnelingmicroscopy
AT zhenggai uncoveringmultiscalestructurepropertycorrelationsviaactivelearninginscanningtunnelingmicroscopy
AT ramavasudevan uncoveringmultiscalestructurepropertycorrelationsviaactivelearninginscanningtunnelingmicroscopy
AT maximziatdinov uncoveringmultiscalestructurepropertycorrelationsviaactivelearninginscanningtunnelingmicroscopy