KIT | KIT-Bibliothek | Impressum | Datenschutz

Mapping local atomic structure of metallic glasses using machine learning aided 4D-STEM

Kang, Sangjun ORCID iD icon 1; Wollersen, Vanessa 1; Minnert, Christian 2; Durst, Karsten 2; Kim, Hyoung Seop; Kuebel, Christian ORCID iD icon 1; Mu, Xiaoke 1
1 Institut für Nanotechnologie (INT), Karlsruher Institut für Technologie (KIT)
2 Technische Universität Darmstadt (TU Darmstadt)

Abstract:

Amorphous materials, e.g., polymers, metallic and oxidic glasses, consist of heterogeneous atomic/molecular packing at the nanoscale. Spatial variation of the local structure plays an important role in determining material properties. Experimentally probing the local atomic structure within the amorphous phase has been one of the main challenges for material research. Here, we present a new approach to characterize the local atomic structure and map structural variants in the amorphous phase using machine learning (ML) aided four dimensional-scanning transmission electron microscopy (4D-STEM). We utilized nonnegative matrix factorization (NMF) to identify the local structural types of metallic glasses from the 4D-STEM dataset. Using Fe-based metallic glasses as a model system, we demonstrate that two basic structural types, one with a more liquid-like and another with a more solid-like structure, are distributed throughout the glass with a characteristic length scale of a few nanometers. Thermal annealing induces a change in their distribution and relative population but without the appearance of any additional phase. This provides new insight into the relaxation phenomena of metallic glass and solid experimental evidence for the theoretical hypothesis on atomic packing in glassy structures.


Zugehörige Institution(en) am KIT Institut für Nanotechnologie (INT)
Publikationstyp Forschungsdaten
Publikationsdatum 10.11.2023
Erstellungsdatum 10.06.2020 - 20.10.2022
Identifikator DOI: 10.35097/1802
KITopen-ID: 1000164027
HGF-Programm 43.35.03 (POF IV, LK 01) Structural and Functional Behavior of Solid State Systems
Lizenz Creative Commons Namensnennung 4.0 International
Vorab online veröffentlicht am 08.11.2023
Schlagwörter Four dimensional-scanning transmission electron microscopy (4D-STEM), Pair distribution function (PDF), Nonnegative matrix factorization (NMF), Metallic glasses
Liesmich

Raw data for the publication "Mapping local atomic structure of metallic glasses using machine learning aided 4D-STEM"

Art der Forschungsdaten Dataset
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page