Atom Identification in Bilayer Moiré Materials with Gomb-Net

Austin C. Houston; Sumner B. Harris; Hao Wang; Yu Chuan Lin; David Geohegan; Kai Xiao; Gerd Duscher

doi:10.1021/acs.nanolett.5c01460

Atom Identification in Bilayer Moiré Materials with Gomb-Net

Austin C. Houston, Sumner B. Harris, Hao Wang, Yu Chuan Lin, David Geohegan, Kai Xiao, Gerd Duscher

Research output: Contribution to journal › Article › peer-review

Abstract

Moiré patterns in van der Waals bilayer materials complicate the analysis of atomic-resolution images, hindering the atomic-scale insight typically attainable with scanning transmission electron microscopy. Here, we report a method to detect the positions and identities of atoms in each of the individual layers that compose twisted bilayer heterostructures. We developed a deep learning model, Gomb-Net, which identifies the coordinates and atomic species in each layer, deconvoluting the moiré pattern. This enables layer-specific mapping of atomic positions and dopant distributions, unlike other commonly used segmentation models which struggle with moiré-induced complexity. Using this approach, we explored the Se atom substitutional site distribution in a twisted fractional Janus WS₂-WS_2(1-x)Se_2x heterostructure and found that layer-specific implantation sites are unaffected by the moiré pattern’s local energetic or electronic modulation. This advancement enables atom identification within material regimes where it was not possible before, opening new insights into previously inaccessible material physics.

Original language	English
Pages (from-to)	9277-9284
Number of pages	8
Journal	Nano Letters
Volume	25
Issue number	23
DOIs	https://doi.org/10.1021/acs.nanolett.5c01460
State	Published - Jun 11 2025

Funding

This work was supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division. Development of machine learning methods was supported by the Center for Nanophase Materials Sciences (CNMS), which is a U.S. Department of Energy, Office of Science User Facility at Oak Ridge National Laboratory. The microscopy and machine learning in this work was partially supported by the AI Tennessee Initiative and the Electron Microscopy Center at the University of Tennessee, Knoxville. For the MOCVD growth of 2D WS crystals Y.-C.L. acknowledges funding from NEWLIMITS, a center in nCORE as part of the Semiconductor Research Corporation (SRC) program sponsored by NIST through award number 70NANB17H041. 2

Keywords

machine learning
moiré materials
scanning transmission electron microscopy
twisted bilayers

Access to Document

10.1021/acs.nanolett.5c01460

Cite this

@article{c7904fc5d7f1476dad047e6036aac5cf,

title = "Atom Identification in Bilayer Moir{\'e} Materials with Gomb-Net",

abstract = "Moir{\'e} patterns in van der Waals bilayer materials complicate the analysis of atomic-resolution images, hindering the atomic-scale insight typically attainable with scanning transmission electron microscopy. Here, we report a method to detect the positions and identities of atoms in each of the individual layers that compose twisted bilayer heterostructures. We developed a deep learning model, Gomb-Net, which identifies the coordinates and atomic species in each layer, deconvoluting the moir{\'e} pattern. This enables layer-specific mapping of atomic positions and dopant distributions, unlike other commonly used segmentation models which struggle with moir{\'e}-induced complexity. Using this approach, we explored the Se atom substitutional site distribution in a twisted fractional Janus WS2-WS2(1-x)Se2x heterostructure and found that layer-specific implantation sites are unaffected by the moir{\'e} pattern{\textquoteright}s local energetic or electronic modulation. This advancement enables atom identification within material regimes where it was not possible before, opening new insights into previously inaccessible material physics.",

keywords = "machine learning, moir{\'e} materials, scanning transmission electron microscopy, twisted bilayers",

author = "Houston, \{Austin C.\} and Harris, \{Sumner B.\} and Hao Wang and Lin, \{Yu Chuan\} and David Geohegan and Kai Xiao and Gerd Duscher",

note = "Publisher Copyright: {\textcopyright} 2025 American Chemical Society.",

year = "2025",

month = jun,

day = "11",

doi = "10.1021/acs.nanolett.5c01460",

language = "English",

volume = "25",

pages = "9277--9284",

journal = "Nano Letters",

issn = "1530-6984",

publisher = "American Chemical Society",

number = "23",

}

TY - JOUR

T1 - Atom Identification in Bilayer Moiré Materials with Gomb-Net

AU - Houston, Austin C.

AU - Harris, Sumner B.

AU - Wang, Hao

AU - Lin, Yu Chuan

AU - Geohegan, David

AU - Xiao, Kai

AU - Duscher, Gerd

PY - 2025/6/11

Y1 - 2025/6/11

N2 - Moiré patterns in van der Waals bilayer materials complicate the analysis of atomic-resolution images, hindering the atomic-scale insight typically attainable with scanning transmission electron microscopy. Here, we report a method to detect the positions and identities of atoms in each of the individual layers that compose twisted bilayer heterostructures. We developed a deep learning model, Gomb-Net, which identifies the coordinates and atomic species in each layer, deconvoluting the moiré pattern. This enables layer-specific mapping of atomic positions and dopant distributions, unlike other commonly used segmentation models which struggle with moiré-induced complexity. Using this approach, we explored the Se atom substitutional site distribution in a twisted fractional Janus WS2-WS2(1-x)Se2x heterostructure and found that layer-specific implantation sites are unaffected by the moiré pattern’s local energetic or electronic modulation. This advancement enables atom identification within material regimes where it was not possible before, opening new insights into previously inaccessible material physics.

AB - Moiré patterns in van der Waals bilayer materials complicate the analysis of atomic-resolution images, hindering the atomic-scale insight typically attainable with scanning transmission electron microscopy. Here, we report a method to detect the positions and identities of atoms in each of the individual layers that compose twisted bilayer heterostructures. We developed a deep learning model, Gomb-Net, which identifies the coordinates and atomic species in each layer, deconvoluting the moiré pattern. This enables layer-specific mapping of atomic positions and dopant distributions, unlike other commonly used segmentation models which struggle with moiré-induced complexity. Using this approach, we explored the Se atom substitutional site distribution in a twisted fractional Janus WS2-WS2(1-x)Se2x heterostructure and found that layer-specific implantation sites are unaffected by the moiré pattern’s local energetic or electronic modulation. This advancement enables atom identification within material regimes where it was not possible before, opening new insights into previously inaccessible material physics.

KW - machine learning

KW - moiré materials

KW - scanning transmission electron microscopy

KW - twisted bilayers

UR - https://www.scopus.com/pages/publications/105007500039

U2 - 10.1021/acs.nanolett.5c01460

DO - 10.1021/acs.nanolett.5c01460

M3 - Article

C2 - 40454431

AN - SCOPUS:105007500039

SN - 1530-6984

VL - 25

SP - 9277

EP - 9284

JO - Nano Letters

JF - Nano Letters

IS - 23

ER -

Atom Identification in Bilayer Moiré Materials with Gomb-Net

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this