# An Introduction to Electrocatalyst Design using Machine Learning for Renewable Energy Storage

**Authors:** C. Lawrence Zitnick, Lowik Chanussot, Abhishek Das, Siddharth Goyal, Javier Heras-Domingo, Caleb Ho, Weihua Hu, Thibaut Lavril, Aini Palizhati, Morgane Riviere, Muhammed Shuaibi, Anuroop Sriram, Kevin Tran, Brandon Wood, Junwoong Yoon, Devi Parikh, Zachary Ulissi  
**arXiv:** [2010.09435](https://arxiv.org/abs/2010.09435)  
**Date:** 2020-10-14  
**Summary page:** https://www.summarizepaper.com/en/paper/2010.09435v1/

## Key points

- The world needs scalable and cost-effective solutions for renewable energy storage to meet increasing energy demands and reduce climate change.
- Energy storage is necessary to transfer power from times of peak generation to peak demand, which may require storing power for hours, days, or even months.
- One potential solution is the conversion of renewable energy into other fuels such as hydrogen or methane, but this requires cost-effective solutions for running electrochemical reactions.
- Finding low-cost electrocatalysts that can drive these reactions at high rates is one of the biggest challenges in this area.
- Quantum mechanical simulations using density functional theory can be used to test and evaluate new catalyst structures, but the high computational cost limits the number of structures that can be tested.
- Machine learning offers a promising approach to efficiently approximate these calculations and find effective electrocatalysts.
- The Open Catalyst Project OC20 dataset is used for model training in this study.
- Machine learning models have been developed using this dataset and have demonstrated good performance in predicting reaction energies and identifying promising catalysts.
- Developing low-cost electrocatalysts for renewable energy storage is important, and machine learning can accelerate progress towards achieving this goal.

## Summary

The world's increasing energy needs and the urgent need to reduce climate change require scalable and cost-effective solutions for renewable energy storage. As we rely more on intermittent power sources like wind and solar, there is a growing need for energy storage to transfer power from times of peak generation to peak demand, which may require storing power for hours, days, or even months. One potential solution that can scale up to nation-sized grids is the conversion of renewable energy into other fuels such as hydrogen or methane. However, this process requires cost-effective solutions for running electrochemical reactions. One of the biggest challenges in this area is finding low-cost electrocatalysts that can drive these reactions at high rates. Quantum mechanical simulations using density functional theory can be used to test and evaluate new catalyst structures. Unfortunately, the high computational cost of these simulations limits the number of structures that can be tested. To address this challenge, machine learning offers a promising approach to efficiently approximate these calculations and find effective electrocatalysts. This paper provides an introduction to the challenges involved in finding suitable electrocatalysts and how machine learning can be applied to solve this problem. The Open Catalyst Project OC20 dataset is used for model training in this study. The authors discuss how machine learning models have been developed using this dataset and describe their performance in predicting reaction energies and identifying promising catalysts. Overall, this paper highlights the importance of developing low-cost electrocatalysts for renewable energy storage and demonstrates how machine learning can accelerate progress towards achieving this goal.