Predicting metal-metal interactions. II. Accelerating generalized schemes through physical insights

Area of Science:

• Materials Science
• Computational Chemistry
• Catalysis

Background:

• Operando-computational frameworks are crucial for predicting catalyst performance under reaction conditions.
• Density functional theory (DFT) models can predict catalyst stability descriptors.
• Existing alloy stability models offer site-by-site resolution for metal atoms in nanoparticles.

Purpose of the Study:

• To present accelerated approaches for parameterizing an alloy-stability model.
• To reduce the computational cost of catalyst screening.
• To enable rapid exploration of transition metal alloys for catalysis.

Main Methods:

• Developed accelerated parameterization methods by combining quadratic energy functions with linear correlations to bulk cohesive energies.
• Utilized interpolation across coordination number and chemical space.
• Reduced training set size from 204 to 24 DFT calculated total energies for 12 fcc p- and d-block metals.

Main Results:

• Achieved accurate predictions of alloy stability with significantly smaller training datasets.
• Validated accelerated approaches on extended surfaces and nanoparticles with low mean absolute errors (0.10 eV and 0.24 eV, respectively).
• Demonstrated efficiency boost for exploring transition metal alloy materials.

Conclusions:

• Accelerated parameterization methods enhance the efficiency of alloy-stability modeling.
• The developed approaches enable rapid and exhaustive exploration of catalytic material space.
• This work facilitates faster discovery of novel transition metal alloy catalysts.