(174r) A Computationally Efficient Procedure for Studying Segregation in Alloy Materials with Monte Carlo Simulations

Agrahari, G., Indian Institute of Technology Bombay
Chatterjee, A., Indian Institute of Technology Bombay
We present an advanced sampling technique based on Metropolis Monte Carlo (MC) to rapidly study phase behavior of several thousand - several million atoms alloy material systems at equilibrium. When the alloy material contains surfaces and preferential enrichment of one of constituent species at the top few surface layers is observed, the MC technique can be additionally used to study the surface segregation. In MC, starting with an initially random arrangement of the constituent atoms, equilibrium structures are attained by attempting trail moves, such as random displacements and swapping of randomly chosen pairs of atoms. However, the computational cost of MC simulations can become prohibitively high beyond a few thousand atoms and it may require several days to weeks of time to ultimately reach equilibrium. A simple idea is introduced to drastically reduce the computational cost by preparing the initial arrangement with the help of a nano thermodynamic model, which we have developed recently. The nano thermodynamic model is used to first estimate the composition of different surface and bulk regions of the material. The initial structure to be used in the MC calculation is prepared using the compositions that have estimated from the nano thermodynamic model so that the prepared structure is close to equilibrium. Using Au-Pt, Ni-Pt and Ag-Au as prototype systems, we demonstrate that a significant speed-up is achieved using our approach.