Doctoral network for high-throughput screening, synthesis and characterization of active materials for flow batteries
Doctoral network for high-throughput screening, synthesis and characterization of active materials for flow batteries
The high-throughput methods to be developed in PREDICTOR will revolutionize the screening and development of materials for electrochemical energy storage. High-throughput screening methods have been used for new drug development since the 1980s and high-throughput synthesis and characterization has been used in the development of materials for catalysis, sensors, polymers, batteries and electrochemistry, with computational screening methods playing a particularly important role. However, despite the vast market potential of electrochemical storage, high-throughput methods, and accompanying self-optimization and data management systems, have not yet been developed in a fully integrated fashion in this field.
PREDICTOR will focus on the high-throughput development of materials for redox flow batteries (RFBs), as one of the most promising technologies for medium- to long-term energy storage. In RFBs, a reversible chemical change occurring within liquid electrolytes enables the rapid storage and release of energy: the accelerated development of organic electrolytes thus offers significant potential to improve these systems and tailor them for specific applications. However, the new methods will be transferable to all electrochemical energy storage techniques.
The project work is divided into three development areas, whereby the entire process consists of several sub-areas and is iterated several times (see below):
1) Modelling, simulation and computational high-throughput screening, 2) Experimental high-throughput methods, 3) Data management and validation, experimental demonstration.
In the areas of modelling, simulation and computational high-throughput screening, the foundations will be laid for the synthesis of electrolytes and active materials. Work will comprise the development of battery twins, i.e. detailed computer-based functional models that allow simulations of the properties of batteries. An existing computational high-throughput screening method, based on a digital battery twin (developed in the EU-funded project SONAR🔗) will be used as a starting point for the filter based computational screening of both conventional and shuttle-based RFBs. The screening will be based on optimized scale-specific models and machine learning methods to find the best candidates of organic active materials for the planned application. The candidates with the best electrochemical performance will be synthesized and characterized, and the obtained experimental data on the material and battery system properties will be used to validate and optimize the battery twins.
Experimental high-throughput methods will use the results of computational high-throughput screening and enable demonstration of new battery types based on syntheses and automated electrolyte characterization and fabrication, as well as automated battery testing.
PREDICTOR aims to expedite material synthesis through high-throughput methods, focusing on organic active materials for RFBs. This involves developing efficient procedures for late-stage functional group modification and scaffold building. Prior knowledge exploration through deep literature mining will inform high-throughput screening hypotheses. Experimental procedures, carried out robotically or in droplet flow format, will include characterization via UPLC and inline NMR. Another focus is the automated synthesis of metal complexes. Promising compounds identified through computational screening will be synthesized for testing, encompassing variations in conducting salts, acids, and additives.
Next, electrolytes with different types and concentrations of conducting salts, acids and additives will be produced and analysed fully automatically. To this end, a commercial liquid dispensing system will be converted so that it can measure conductivities, spectra (UV/VIS or RAMAN) and electrochemical properties of half cells (linear sweep voltammetry, cyclic voltammetry, electrochemical impedance spectroscopy) independently. Based on AI-based pattern recognition, spectra and voltammograms will be automatically quantified and contextualized.
Similarly, the high-throughput battery tests will be an automated system that can perform tests with targeted parameters on its own. Based on various actuators and sensors, a small laboratory battery system will be set up allowing data acquisition, evaluation and control of the tests using microprocessor-based control systems like PLCs.
The high-throughput synthesis, automated electrolyte analysis and production, and cell tests will employ self-optimization via various AI methods. Targets like high conductivity or kinetics will be set, and the system will autonomously conduct experiments needed for multicomponent mixture production. Analysing all acquired experimental data will automatically determine future strategies, drastically reducing the number of experimental steps needed. This enhances productivity and output while lowering costs.
The above developments will be supported by the cross-topics of data management and validation, and experimental demonstration. Data management covers the handling of the large amount of data acquired, which will be generated by all procedures, but used differently. In particular, experimental data are needed to optimise and validate models and simulations to improve future results. For this purpose, the data-generating devices will be coupled to a newly developed PREDICTOR database, allowing automatic, real-time storage of the data. This in turn will allow optimisation of the experiments and reduce steps. The database will also make the data available according to the FAIR principles (findability, accessibility, interoperability, reusability), including a corresponding ontology.
Finally, the entire process will be demonstrated, showing that the development of novel batteries can be accelerated on the basis of serial/iterative linkage using high-speed methods. Three novel RFBs will be demonstrated at TRL 3-4, based on previously uninvestigated organic redox pairs prioritised by the computational screening.
Funded by the European Union (Grant Agreement no. 101168943).
Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union.
Neither the European Union nor the granting authority can be held responsible for them.