Quasi-combinatorial energy landscapes for nanoalloy structure optimisation

Area of Science:

• Computational materials science
• Materials informatics
• Chemical physics

Background:

• Predicting the lowest energy structure of nanoalloys is crucial for designing materials with desired properties.
• Traditional methods struggle with the complex energy landscapes of nanoalloys.
• Nanoalloy structure prediction is often treated as a mixed-variable optimization problem.

Purpose of the Study:

• To formulate nanoalloy structure prediction as a mixed-variable optimization problem.
• To analyze the effective energy landscape of nanoalloys.
• To develop efficient global optimization strategies for nanoalloy structures.

Main Methods:

• Formulation of nanoalloy structure prediction as a mixed-variable optimization problem.
• Surveying the effective energy landscape using the Gupta potential for binary systems.
• Application of local optimization techniques and generalized basin-hopping with a Metropolis acceptance criterion.

Main Results:

• The energy landscape of nanoalloys with small lattice mismatch exhibits few local optima, suitable for local optimization.
• Local optimization techniques scale quadratically with system size.
• Generalized basin-hopping effectively optimizes binary and ternary nanoalloys by searching the space of multiminima.

Conclusions:

• Nanoalloy structure prediction can be efficiently solved using a combination of local optimization and basin-hopping.
• The proposed method is effective for global optimization of binary and ternary nanoalloys.
• Understanding the energy landscape is key to developing efficient computational materials design strategies.