Related Experiment Video
Updated: Jul 23, 2025
11:37
Cellular Redox Profiling Using High-content Microscopy
Published on: May 14, 2017
11.0K
High-Throughput Screening of Promising Redox-Active Molecules with MolGAT
Mesfin Diro Chaka1,2, Chernet Amente Geffe1, Alex Rodriguez3
1Department of Physics, College of Natural and Computational Sciences, Addis Ababa University, P.O. Box 1176, Addis Ababa 1176, Ethiopia.
ACS Omega
|July 17, 2023
Summary
Researchers developed MolGAT, a graph neural network model, to accelerate the discovery of novel organic molecules for redox flow batteries (RFBs). This AI approach efficiently screens vast chemical libraries, identifying thousands of promising candidates for energy storage applications.
Area of Science:
- Materials Science
- Computational Chemistry
- Energy Storage
Background:
- Redox flow batteries (RFBs) are promising for large-scale energy storage due to high energy density, low cost, and environmental benefits.
- Identifying suitable organic molecules with high redox activity, solubility, stability, and fast kinetics is critical but challenging for RFB development.
- Traditional computational methods like density functional theory are accurate but time-consuming and computationally expensive.
Purpose of the Study:
- To develop and validate a machine learning model for accelerating the discovery of novel redox-active organic compounds for RFBs.
- To leverage graph neural networks (GNNs) for predicting molecular redox potentials based on structural and atomic properties.
- To screen large chemical databases for potential catholyte and anolyte materials for RFB applications.
Main Methods:
- Developed MolGAT, a GNN-based model, to predict redox potentials of organic molecules using molecular structure, atomic, and bond attributes.
- Trained MolGAT on a dataset of over 15,000 compounds, demonstrating superior performance compared to other GNN variants.
- Screened a large chemical dataset (581,014 molecules) using the trained MolGAT model.
Main Results:
- Identified 23,467 potential redox-active organic compounds from the screened chemical dataset.
- Discovered 20,716 potential catholytes with predicted redox potentials up to 2.87 V.
- Identified 2,751 potential anolytes with predicted redox potentials as low as -2.88 V.
Conclusions:
- Graph neural networks, specifically MolGAT, are effective tools for accelerating the discovery of redox-active materials for RFBs.
- The study successfully screened a vast chemical space, identifying numerous promising candidates for further investigation.
- This AI-driven approach significantly reduces the time and computational cost associated with discovering new materials for energy storage.
