Dec 5, 2024
8:00pm - 10:00pm
Hynes, Level 1, Hall A
Marc Descoteaux1,Faith Chen1,Malia Wenny2,Daniel Laorenza1,Craig Brown2,Jarad Mason1,Boris Kozinsky1,3
Harvard University1,National Institute of Standards and Technology2,Robert Bosch Research and Technology Center3
Marc Descoteaux1,Faith Chen1,Malia Wenny2,Daniel Laorenza1,Craig Brown2,Jarad Mason1,Boris Kozinsky1,3
Harvard University1,National Institute of Standards and Technology2,Robert Bosch Research and Technology Center3
Layered materials leveraging the order-disorder phase transition of hydrocarbon chains have been shown to exhibit significant barocaloric effects with promise for applications in thermal energy storage and conversion [1, 2]. While the material design space of barocaloric layered materials is large, a complete understanding of how a given layer structure, composition, and hydrocarbon chain length influence the nature of the barocaloric effect is currently lacking due to limited data quantifying the molecular motions responsible for the resultant entropic changes.<br/><br/>In this work, we perform a computational study with molecular dynamics (MD) simulations of a hybrid organic-inorganic perovskite and a di-<i>n</i>-alkylammonium halide salt which have experimentally presented significantly different barocaloric properties and phase-change behavior despite containing hydrocarbon chains of the same length and exhibiting the same transition temperature. To understand the source of these differences with a resolution on the atomic-scale, we use the E(3)-equivariant machine learned force field Allegro [3], trained with an on-the-fly active learning framework based on the Bayesian force field Flare [4]. This approach enables MD simulations of barocaloric layered materials with the accuracy of density functional theory to be scaled to large systems and long timescales.<br/><br/>The computational results are benchmarked against known experimental properties of the phase change and supplemented with computed and measured quasi-elastic neutron scattering data. Overall, the work realizes a fully atomistic description of these materials and quantifies their structural and conformational dynamics, developing understanding which improves our capability to rationally design barocaloric materials with desired properties for thermal energy management.<br/><br/>[1] J. Seo <i>et al.</i> Colossal barocaloric effects with ultralow hysteresis in two-dimensional metal–halide perovskites. <i>Nat. Commun.</i> <b>13</b> 2536 (2022).<br/>[2] J. Seo <i>et al</i>. Barocaloric effects in dialkylammonium halide salts. <i>J. Am. Chem. Soc.</i> <b>146 </b>2736 (2024).<br/>[3] A. Musaelian <i>et al.</i> Learning local equivariant representations for large-scale atomistic dynamics. <i>Nat. Commun.</i> <b>14</b> 579 (2023).<br/>[4] J. Vandermause <i>et al.</i> Active learning of reactive Bayesian force fields applied to heterogeneous catalysis dynamics of H/Pt. <i>Nat. Commun.</i> <b>13</b> 5183 (2022).