# Finding the Pareto front for high-entropy-alloy catalysts

## TL;DR

This paper introduces a new genetic algorithm to find the best balance between activity and stability in high-entropy-alloy catalysts.

## Contribution

A novel multi-objective genetic algorithm combining machine learning and DFT calculations for optimizing catalysts.

## Key findings

- The Pareto front for catalysts includes alloys with diverse elements.
- Increasing stability reduces catalytic activity and vice versa.
- Findings align with experimental data from 545 experiments.

## Abstract

Finding catalysts that have both high activity and high stability presents a long-standing challenge. Since optimizing activity and stability are conflicting objectives, the best one can do is find the Pareto front that yields optimal tradeoffs between these features. On the Pareto front, there is a trade-off where a portion of catalytic activity must be sacrificed to gain further stability and vice versa. Here, we provide a method to optimize the front by designing a multi-objective genetic algorithm that combines machine learning, graph neural network calculations, and density functional calculations. The application considered is the oxygen evolution reaction catalyzed by high-entropy alloys. We find that the Pareto front generally contains alloys with diverse elements, but that enhancing stability inevitably inflicts a toll on activity. We compare the general conclusions of our work to a survey of 545 experiments.

We provide a new genetic algorithm approach to bi-objective optimization of catalysts composed of high-entropy alloys, in which activity and stability are conflicting targets.