Surface segregation in bimetallic systems plays a crucial role in heterogeneous catalysis. However, mechanistic studies of such restructuring phenomena remain scarce and challenging due to the large configurational space that must be explored. To this end, we develop an automated method to discover elementary surface restructuring processes in an unbiased fashion. We utilize molecular dynamics (MD) to detect restructuring events, isolate them, and optimize with transition state modeling techniques. To enable large-scale MD simulations spanning microseconds at the first-principles level of accuracy, we train a machine learning force field using Bayesian active learning and Gaussian process regression, where minimal amount of density functional theory data is requested automatically by the model based on its own internal estimate of uncertainty.<br/><br/>We investigate Pd/Ag heteroepitaxy in collaboration with surface science experiments. Pd/Ag is a selective hydrogenation catalyst, where Pd initiates the catalytic cycle and selective reaction ensues on inert Ag host. However, the corresponding surface composition and morphology are highly dynamic and hence not well understood. In vacuum, Pd prefers to reside subsurface because of higher surface energy. Our MD simulations show that layer-by-layer Pd dissolution is preceded by Ag encapsulation of the Pd cluster. Consequently, significant Ag mass transport occurs out of the surface layer, resulting in extensive vacancy pit growth as observed in scanning tunneling microscopy measurements. This metastable structure persists at room temperature.<br/><br/>In addition, our automated workflow has revealed three new classes of concerted restructuring events dominant at step edges, in addition to the well-established exchange descent mechanism: (1) Vacancy insertion; (2) Direct exchange; (3) Interlayer exchange. Our results highlight the importance of vacancies at steps and kinks in facilitating Pd-Ag place exchange. The discovered events enable us to go beyond thermodynamics and construct a complete set of mechanisms underlying the metastable dissolution mode in Pd/Ag heteroepitaxy. Investigation of more complex restructuring phenomena, such as adsorbate-induced reverse segregation, is underway. The atomistic insights presented in our work provide a step toward systematic understanding and engineering of surface segregation dynamics in bimetallic catalysts.

Automated characterization of surface restructuring events in Pd/Ag heteroepitaxy from Gaussian process molecular dynamics

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.