PNNL

Abstract

The solubility of redox-active organic molecules (ROMs) in non-aqueous redox flow batteries (NRFBs) is a critical factor determining the energy density of the system. However, the scarcity of comprehensive solubility data has hindered electrolyte development. In this study, we systematically investigate the solubility behavior of ROMs in the presence of supporting salts to propose practical electrolyte formulations for NRFBs. Using automated high-throughput experimentation, we screen the solubility of 2,1,3-benzothiadiazole (BTZ) and lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) in various organic solvents. Leveraging a Random Forest inference model, we identify a binary solvent mixture of m-xylene and acetonitrile, which dissolves 3 M of both BTZ and LiTFSI—exceeding the previously reported 2 M limit in pure acetonitrile. This enhanced solubility is attributed to the combination of a LiTFSI-phobic yet BTZ-philic solvent, which mitigates solubility competition between BTZ and LiTFSI, as the latter favors highly polar solvents. This work introduces a promising electrolyte design strategy for NRFBs and demonstrates the effectiveness of high-throughput screening combined with advanced data analysis in optimizing complex multi-component systems. Furthermore, it emphasizes the urgent need for more comprehensive solubility data to facilitate the development of practical NRFB electrolytes.