Molecular optimization framework for identifying promising organic radicals for aqueous redox flow batteries

A computational strategy for finding new optimized structures for organic redox flow batteries. Credit: Sundarya SV et al.

Recent advances in the development of machine learning and optimization techniques have opened exciting new possibilities for identifying suitable molecular designs and candidate compounds and chemicals for various applications. Optimization techniques, some of which are based on machine learning algorithms, are powerful tools that can be used to select the optimal solutions to a given problem from a wide range of possibilities.

Researchers at Colorado State University and the National Renewable Energy Laboratory have applied state-of-the-art molecular optimization models to various real-world problems that necessitate the identification of promising new molecular designs. In their most recent studies, they appeared in The intelligence of nature’s machineThey specifically implemented a newly developed open source optimization framework for the task of identifying viable organic radicals for hydrophilic materials. oxidation and reduction Flow batteries and power converting devices chemical energy in electricity.

“Our project was funded by the ARPA-E program that was looking to shorten the time it takes to develop new energy materials using machine learning techniques,” Peter C. St. John, one of the researchers who conducted the study, told TechXplore. . “Finding new candidates for redox flow batteries has been an interesting extension of some of our previous work, including research published in Nature Connections and another in Scientific dataBoth research the organic roots.

John and colleagues’ new framework was inspired by their previous work on molecular optimization. The frame is mainly composed of Artificial intelligence (AI) The AlphaZero tool, developed by DeepMind, combined with a fast model derived from machine learning, made up of two graphic neural networks trained in nearly 100,000 quantum chemistry simulations.

first graph neural networks It was trained to predict redox potentials, two important factors for determining how much energy can be stored in aqueous redox flow batteries. The second predicts the electron density and the local 3D environment, both of which have been found to correlate with the lifespan of these batteries.

“We position molecule optimization as tree research, where we build molecules by adding repetitive components to a growing structure,” St. John explained. “The advantage of this approach is that we can prune large branches of the search space as molecules begin to show unrealistic substructures. We can thus limit our search space to molecules that meet a predefined set of simple criteria.”

The researchers used their molecular optimization framework to perform a series of tests aimed at identifying potential organic radicals of aqueous redox flow batteries that could be particularly stable and promising. The framework succeeded in identifying several candidate molecules that met a specific set of criteria identified by St. John and colleagues.

“We have shown that the pool of potential candidates for a particular type of charge carrier in organic redox flow batteries may be larger than previously considered,” St. John said. “We’ve also shown that molecules can be found that can lead to simpler, high-performance batteries without the use of transition metals.”

So far, the initialization The framework developed by this team of researchers has proven to be a very promising tool for tackling complex real-world problems related to engineering and chemistry. It can thus be used in the future to identify desirable new compounds and candidate molecules for many different technologies, including aqueous redox flow batteries.

“We would now like to explore adding additional parameters such as solubility and redox pairs between charged states,” St John added. “This will require additional training data, but may lead to more promising candidate structures.”


Heterogeneous acid manure can enhance the performance of aqueous redox flow batteries at low temperatures


more information:
Shree Sowndarya SV et al, A multi-targeted optimization of de novo stable organic radicals for aqueous redox flow batteries, The intelligence of nature’s machine (2022). DOI: 10.1038 / s42256-022-00506-3

Peter C. St. John et al, Predicting the enthalpy of dissociation of an organic monolithic bond with close chemical precision at a computational cost of less than a second, Nature Connections (2020). DOI: 10.1038 / s41467-020-16201-z

Peter C. St. John et al, Quantitative chemical calculations of more than 200,000 organic radicals and 40,000 closed-cell molecules, Scientific data (2020). DOI: 10.1038 / s41597-020-00588-x

© 2022 Science X Network

the quote: Molecular Optimization Framework for Identification of Promising Organic Radicals for Aqueous Redox Flow Batteries (2022, September 15) Retrieved September 15, 2022 from

This document is subject to copyright. Notwithstanding any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.