Multiscale Vacancy and Dislocation-Mediated Surface Segregation in CuNi Alloy Up to Microsecond Timescales with Accelerated Dynamics
Richard Garza,Jiyoung Lee,Mai Nguyen,Andrew Garmon,Meng Li,Danny Perez,Graeme Henkelman,Judith Yang,Wissam Saidi
DOI: https://doi.org/10.1017/s143192762100862x
IF: 4.0991
2021-01-01
Microscopy and Microanalysis
Abstract:Phase separation of CuNi in vacuum is well-documented in both theory and experiment: annealing enriches the exposed surfaces with Cu and the subsurface layers with Ni [1, 2]. Even before exposure to air or corrosive environments, the bimetal restructures, affecting outcomes in both oxidation and reduction experiments. Transformations on the atomic and nanoscale could influence the study of Cu-Ni catalysts e.g. for the reduction of CO2, as well as in corrosion-resistant, marine alloys, making it difficult to replicate behavior between samples observed using microscopy techniques. Fig. 1 is an image taken during HRTEM observation and in situ oxidation of CuNi(100): the metals are impossible to distinguish within the alloy, since they display low lattice mismatch, and their oxides also exhibit little difference in dspacing. Further, the atomic mechanisms, dynamics, and rate of this restructuring are difficult to observe, since the two metallic nuclei are difficult to distinguish even with Z-contrast, and elemental mapping is only available over long times (minutes to hours). The migration and coalescence of individual vacancies is also poorly resolved in experiment, yet the flow of point defects towards an exposed surface is directly related to the well-known counterflow of Ni into the bulk lattice. Computationally, the atomic transformations underlying nanoscale alloy restructuring are much more effectively resolved using molecular dynamics (MD). However, these only reveal short time scales up to nanoseconds, not the dynamics occurring over microseconds and beyond that are most significant in alloy surface segregation processes approaching experimental times. To address this challenge, we utilize two accelerated methods, ParSplice and Adaptive Kinetic MC (AKMC), to evolve to longer time frames and narrow this gap, comparing our results with standard MD to distinguish surface segregation mechanisms by their different timescales.