Rapid Atomic Structure Prediction of Multimetallic Nanoparticles with Physics-Based Machine Learning.

Area of Science:

• Materials Science
• Computational Chemistry
• Nanotechnology

Background:

• Metal nanoparticles (NPs) are vital in catalysis, biomedicine, and electronics due to tunable properties.
• Chemical ordering in multimetallic NPs significantly impacts their stability and performance.
• Traditional methods like Density Functional Theory (DFT) are computationally intensive and limited for complex NPs.

Purpose of the Study:

• To develop and validate a rapid, accurate computational method for predicting the chemical ordering and stability of multimetallic NPs.
• To establish guidelines for applying a physics-based model (Bond-Centric Model) coupled with a genetic algorithm.
• To facilitate the design of stable multimetallic NPs with predictable atomic distributions.

Main Methods:

• Utilized a physics-based Bond-Centric Model integrated with a genetic algorithm to optimize NP chemical ordering.
• Calculated weighting factors scaling monometallic bond strengths for bimetallic interactions.
• Applied the method to 2869-atom cuboctahedron NPs across 15 bimetallic combinations and 6 trimetallic systems.

Main Results:

• The method accurately predicts chemical ordering and stability for diverse bimetallic NP compositions.
• Using small metal dimers for weighting factor calculation ensures computational efficiency and accuracy.
• Successfully extended the model to predict ordering in complex trimetallic NP systems.

Conclusions:

• The developed methodology offers a computationally efficient and accurate approach for multimetallic NP design.
• This facilitates the creation of thermodynamically stable NPs with controlled core-to-surface metal atom distribution.
• Enables accelerated discovery and optimization of NPs for advanced nanotechnological applications.