General Model for d-Center Prediction in Multi-Principal-Element Alloys

Area of Science:

• Materials Science
• Computational Chemistry
• Catalysis

Background:

• The d-band center of transition metals is a critical descriptor for electrocatalysis.
• Predicting d-band centers is challenging, especially for disordered multi-principal-element alloys (MPEAs).

Purpose of the Study:

• To develop a general, physically interpretable model for predicting d-center values.
• To enable accurate d-center estimation across diverse surfaces and compositions.

Main Methods:

• Utilized cluster expansion theory principles to capture local coordination environments.
• Developed a predictive model using 10,680 density functional theory (DFT)-relaxed slabs and over 1.2 million d-center values.
• Investigated the impact of surface orientation, featurization schemes, and regression methods.

Main Results:

• Achieved a mean absolute error (MAE) of approximately 0.09 eV for d-center prediction.
• The model accurately estimates d-centers considering only first nearest-neighbor interactions.
• Demonstrated the influence of various factors on model accuracy and generalizability.

Conclusions:

• The developed model provides rapid and reliable d-center estimation for high-throughput screening and mechanistic interpretation in catalysis.
• Model coefficients offer insights into MPEA surface electronic behavior for experimentalists and theorists.