Bayesian force fields from first principles for nanoparticle heterogeneous catalysis Cameron J. Owen 1 , Yu Xie 2 , JinSoo Lim 1 , Lixin Sun 2 , Jonathan Vandermause 2 and BoriKozinsky 2,3 1 Department of Chemistry & Chemical Biology, Harvard University, USA, 2 John A. Paulson School of Engineering and Applied Sciences, Harvard University, USA, 3 Robert Bosch LLC, Research and Technology Center, USA Quantitative understanding and control of interfacial reactions between the gas-phase and solid surfaces are crucial for improving numerous catalysis and energy conversion systems. Examples of these interfacial phenomena include H 2 exchange and CO adsorption on nanoparticles, both of which are important industrial processes and lead to markedly different particle behaviors. This work demonstrates a collection of robust Bayesian machine learned force fields (MLFFs) trained with an on- the-fly active learning framework implemented using Gaussian Process regression in the FLARE code. Molecular dynamics (MD) simulations are used to sample atomic configurations and density functional theory is only called upon when the Bayesian uncertainty exceeds a threshold. This workflow yields both acceleration in time-to- solution and an increase in computational efficiency relative to ab initio MD. The resulting MLFFs retain first principles accuracy, are fast, and are uncertainty-aware. Following a rigorous validation scheme, through comparison of the dynamic evolution of these particles and bulk systems to available x-ray data (i.e., extended x-ray adsorption fine spectra), long timescale MD simulations for freestanding metal nanoparticle systems (e.g., Pt, Au,PdAu, andCuPt) are performed. In addition to the bare particles, reaction mechanisms under gaseous exposure (e.g., H 2 and CO) are also investigated. These MLFFs allow for the simultaneous study of atomistic mechanisms occurring on these nanoparticles under exposure to reactive atmospheres and the evolution of their structural morphologies. References 1. Marcella, N., Lim, J.S., Płonka, A.M., Yan, G., Owen, C.J., van der Hoeven, J.E.S., Foucher, A.C., Ngan, H.T., Torrisi, S.B., Marinkovic, N.S., Stach, E.A., Weaver, J.F., Aizenberg, J., Sautet, P., Kozinsky, B., Frenkel, A.I. Decoding reactive structures in dilute alloy catalysts. Nat. Commun. 13 , 832 (2022). 2. Vandermause, J., Xie, Y., Lim, J.S., Owen, C.J., Kozinsky, B. Active learning of reactive Bayesian force fields: Application to heterogeneous hydrogen-platinum catalysis dynamics. Accepted, Nat. Commun. (2022). Available at: arXiv:2106.01949. 3. Musaelian, A., Batzner, S.L., Johansson, A., Sun, L., Owen, C.J., Kornbluth, M., Kozinsky, B. Learning local equivariant representations for large-Scale atomistic dynamics. Submitted, (2022). Available at: arXiv:2204.05249. 4. Johansson, A., Xie, Y., Owen, C.J., Lim, J.S., Sun, L., Vandermause, J., Kozinsky, B. Micron-scale heterogeneous catalysis with Bayesian force fields from first principles and active learning. Submitted, (2022). Available at:arXiv:2204.12573. 5. Foucher, A.C., Owen, C.J., Shirman, T., Aizenberg, J., Kozinsky, B., Stach, E.A. Atomic-scale STEM analysis shows structural changes of Au-Pd nanoparticles in various gaseous environments. Submitted, (2022).
P35
© The Author(s), 2022
Made with FlippingBook Learn more on our blog