How to store renewable electricity efficiently for later use is one of the major questions of our time. The recently awarded Growth Fund Circular Batteries project REDOX BLEND aims to develop a new class of batteries for day- to week-long storage. Süleyman Er, head of DIFFER’s Chemical Energy department, leads a work package that develops a computational workflow to search for promising molecules for redox flow batteries, including machine-learning-based prediction of their key properties.
Long-duration energy storage in batteries can help relieve the current burden on the electricity grid caused by a mismatch between the generation of and demand for energy. Current batteries can store energy cost-effectively for only several hours. Situations that require longer-term storage still need new, affordable technologies.
In the REDOX BLEND (REDOX flow Batteries with Large ENergy Density) project, a consortium of Netherlands-based academic researchers, industry partners and TNO, led by Edwin Otten from Groningen University, is developing new batteries that can store renewable energy reliably and affordably for several days.
The project focuses on so-called redox flow batteries: systems that generate electricity by pumping two separate liquid electrolyte solutions through a membrane-containing cell stack. Via reversible reduction-oxidation reactions, electricity is either stored or released.
Stringent requirements
The new batteries should offer a low cost per stored kWh and be able to be charged and discharged quickly, explains Er. "This means that we are looking for materials that show fast kinetics and have a high energy density. Since the energy density is proportional to the amount of electroactive molecules that can fit into the electrolyte, this requirement is directly related to the solubility of the compounds we are looking for."
Er leads the work package that aims to identify the most promising candidate molecules for the anode and cathode electrolytes. "We are looking for materials that are stable, can be easily synthesized, are not too expensive, have a high solubility and show fast kinetics." Since water is used as the electrolyte carrier, the team will focus on materials with a redox potential within the electrochemical stability window of water – yet as close as possible to its upper limit of 1.23 V. "At higher voltages, water starts to decompose, which is something we want to avoid."
In addition, the researchers also aim for materials that are made from abundantly available elements and therefore scalable in production. "That is why part of the research focuses on organic compounds", Er explains.
Expertise in computational materials discovery
The choice for DIFFER as the leading institute for this work package is a logical one, says Er. "We have a strong track record in computational materials discovery for electrochemical energy storage, for example in lithium and sodium battery research. Over the past years, we have developed a specialized computational workflow called RedCat, which integrates machine learning and physics-based prediction tools for redox potential, solubility, and cost estimation."
The workflow will be used to screen large databases of published compounds to identify promising candidates based on their performance, and subsequently select molecules that can be synthesized and produced in a cost-effective way. After this initial selection, a shortlist of the most promising candidates will be further investigated by using machine learning techniques to predict relevant properties that have not been measured yet, such as solubility and redox potential.
"To us, the REDOX BLEND project is a good opportunity to further extend our workflow", Er says. For example, the aim is to include new databases that have not been screened and to add new predictive capacities. In their search for the optimal candidate materials, the researchers do not want to limit themselves to known classes of electrochemically active materials. "We also want to explore domains that have never before been considered for energy storage purposes."
Distant relatives
The researchers plan to identify commonalities in promising compounds to establish a kind of family tree of materials with interesting physical and chemical properties. Instead of merely looking for molecules with similar structures, they also want to map molecules that show similar performance. By pinpointing these distant relatives, which may not look alike but display similar behavior, they aim to establish a set of design rules for newly designed molecules that do not exist in literature, but meet all the requirements that together define optimal battery performance.
Although the project has not yet started, Er is already thinking about the next steps. "In a next project, we want to move towards using generative AI to design compounds on the spot. This means estimating synthetic components, predicting their properties and assessing whether they can be synthesized easily or not. This approach will significantly speed up the discovery of new materials."
As this is a Growth Fund project, Er hopes the discoveries will be picked up by the chemical and battery industry in the Netherlands. "This particular project is very exploratory in nature. But it would be exciting if we could identify new chemistries that open up new possibilities for long-term energy storage and help solve one of the most pressing challenges of the energy transition."
Text: TU/e - EIRES
Go to the News page.