Machine Learning-Guided Design of L1₂-Type Pt-Based High-Entropy Intermetallic Compound for Electrocatalytic Hydrogen...

Area of Science:

• Materials Science
• Catalysis
• Computational Chemistry

Background:

• Rational design of high-entropy intermetallic compounds (HEICs) is hindered by complex structure-property relationships.
• Predictive tools for evaluating hydrogen evolution reaction (HER) activity in HEICs are lacking.

Purpose of the Study:

• To develop a data-driven framework for evaluating the HER activity of L12-type quinary Pt3M(4) HEICs.
• To identify promising HEIC catalysts for efficient hydrogen production.

Main Methods:

• Designed 15 distinct L12-type quinary HEIC compositions using 3d transition metals.
• Employed a deep neural network trained on computed datasets to predict hydrogen adsorption energy (ΔEH*) across numerous microstructures.
• Introduced a novel statistical evaluation approach to quantify microstates within optimal ΔEH* ranges.

Main Results:

• The framework successfully predicted site-specific HER performance across thousands of microstructures per composition.
• Pt3(CrMnFeCo) was identified as the most promising HER catalyst candidate.
• Experimental validation confirmed the superior HER activity of Pt3(CrMnFeCo) across a wide pH range.

Conclusions:

• The data-driven framework provides a new paradigm for HEIC catalyst design.
• Surface Co, Cr, and Fe, along with subsurface Ni and Co, were identified as key elements optimizing HER activity.
• This study deepens the mechanistic understanding of catalytic activity in complex multimetal systems.