471751 Developing a Machine-Learning Molecular Dynamics Approach for Nonadiabatic Surface Chemistry

Tuesday, November 15, 2016: 2:30 PM
Franciscan D (Hilton San Francisco Union Square)
Jiamin Wang and Hongliang Xin, Department of Chemical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA

Transferring energy from solid surfaces to chemical bonds of adsorbed species is a fundamental process in heterogeneous catalysis. To a great extent, the dynamics of molecule-surface interactions driven by thermal energy, i.e., heat, can be well described within the framework of the Born-Oppenheimer approximation where the electronic and nuclear motions can be treated separately. New reaction channels [1] often open up in response to electronic excitations via surface-mediated energy transfer. In this process, the energetic electrons or holes can scatter into surface species and heat up vibrational modes of adsorbates. This process involves not only the electronic ground state but also the excited state, and it is nonadiabatic with respect to the nuclear motions [2].

In the weak coupling limit, the nonadiabatic energy transfer on metal surfaces can be modeled using only the ground-state potential energy surfaces and the nonadiabacity is taken into account with electronic friction contributions in Langevin dynamics by solving Newtonian equations with a thermal fluctuation term [2,3]. In this situation, the electronic friction via partial populating and depopulating of the adsorbate density of states right above the Fermi level is governing the energy transfer. The crucial parameter in this formalism is the electronic friction of an excited charge with nuclear motions of surface adsorbates. However, it is extremely time consuming to compute electron-phonon coupling strength for large systems.

We are developing a machine-learning molecular dynamics approach that uses predicted forces and electronic friction coefficients on each of atoms by ‘learning from data’. Machine learning algorithms, such as the artificial neural networks [4], can use past trajectories as training datasets for fast and accurate prediction of forces and electronic friction coefficients, thus allows us to perform statistical analysis of many trajectories. We test the approach for three ultrafast laser induced surface reactions, 1) activation of oxygen from hollow to bridge, 2) CO desorption, and 3) CO2 formation. We chose those systems because of strong interests in understanding nonadiabatic chemistry of CO oxidation on metal surfaces and its simplicity for fundamental studies. We will also discuss electronic factors governing nonadiabatic energy transfer processes. We found that the electronic friction coefficients and the positions of adsorbate resonance orbital energies are two most important factors governing the efficiency and branching ratio of electron-assisted surface reactions. We believe that unraveling the underlying factors governing those properties is the first step for developing predictive models and structure-function relationships in harnessing nonadiabatic surface chemistry for efficient chemical and energy transformations.

[1] M. Bonn, S. Funk, C. Hess, D. N. Denzler, C. Stampfl, M. Scheffler, M. Wolf, and G. Ertl, Science 285, 1042 (1999).

[2] M. Head‐Gordon and J. C. Tully, J. Chem. Phys. 103, 10137 (1995).

[3] M. Brandbyge, P. Hedegård, T. F. Heinz, J. A. Misewich, and D. M. Newns, Phys. Rev. B 52, 6042 (1995).

[4] J. Behler, Phys. Chem. Chem. Phys. 13, 17930 (2011).

Extended Abstract: File Not Uploaded
See more of this Session: New Developments in Computational Catalysis II
See more of this Group/Topical: Catalysis and Reaction Engineering Division