AUTHOR=Wu Mingyue , Ruiz Pestana Luis 

TITLE=Identifying a machine-learning structural descriptor linked to the creep behavior of Kob-Andersen glasses

JOURNAL=Frontiers in Materials

VOLUME=10

YEAR=2023

URL=https://www.frontiersin.org/journals/materials/articles/10.3389/fmats.2023.1272355

DOI=10.3389/fmats.2023.1272355

ISSN=2296-8016

ABSTRACT=<p>A wide variety of materials, ranging from metals to concrete, experience, typically at high-temperatures or over long time scales, permanent deformations when subjected to sustained loads below their yield stress—a phenomenon known as creep. While theories grounded on defects such as vacancies, dislocations, or grain boundaries can explain creep in crystalline materials, our understanding of creep in disordered solids remains incomplete due to the lack of analogous structural descriptors. In this study, we use molecular dynamics to simulate the creep response of a Kob-Andersen glass model system under constant, uniaxial, compressive stress at finite temperature. We leverage that data to derive, using a machine-learning classification model, a structural descriptor termed <italic>looseness</italic>, <inline-formula id="inf1"><mml:math id="m1" xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:mrow><mml:mi mathvariant="double-struck">L</mml:mi></mml:mrow></mml:math></inline-formula>, which is based on simple and interpretable local structural features and can predict imminent plastic rearrangements within the glass. We show that the average <italic>looseness</italic> of the system evolves logarithmically with time, mirroring the time dependence of the creep strain and demonstrating the ability of our model to bridge local, short-term particle dynamics with the long-term macroscopic creep response. A detailed feature importance analysis reveals the particular significance of short-range structural heterogeneity in the prediction. We also scrutinize the spatial and temporal correlations of <italic>looseness</italic>, which mirror the lack of long-range order in glasses and their dynamic heterogeneity. Our research underscores the substantial predictive potential of machine-learning-derived structural indicators in systems experiencing concurrent stress and thermal excitations, paving the way for future work to elucidate the interplay between thermal and mechanical activation of structural defects in disordered solids.</p>