SOMAS: a platform for data-driven material discovery in redox flow battery development

Abstract

Aqueous organic redox flow batteries offer an environmentally benign, tunable, and safe route to large-scale energy storage. The energy density is one of the key performance parameters of organic redox flow batteries, which critically depends on the solubility of the redox-active molecule in water. Prediction of aqueous solubility remains a challenge in chemistry. Recently, machine learning models have been developed for molecular properties prediction in chemistry and material science. The fidelity of a machine learning model critically depends on the diversity, accuracy, and abundancy of the training datasets. We build a comprehensive open access organic molecular database “Solubility of Organic Molecules in Aqueous Solution” (SOMAS) containing about 12,000 molecules that covers wider chemical and solubility regimes suitable for aqueous organic redox flow battery development efforts. In addition to experimental solubility, we also provide eight distinctive quantum descriptors including optimized geometry derived from high-throughput density functional theory calculations along with six molecular descriptors for each molecule. SOMAS builds a critical foundation for future efforts in artificial intelligence-based solubility prediction models.

High-Throughput Electrochemical Characterization of Aqueous Organic Redox Flow Battery Active Material

“…63,66,68,72,75,77 Accurate measurements of RAOM stability are critical, not only to lifetime cost analysis of potential candidates for decadal RFB operation, 10 but also to the community-wide development of rich data sets of RAOM lifetimes. Such knowledge can complement high-throughput theoretical screening, [78][79][80][81][82][83] enhance machine learning capabilities, 84,85 and motivate RAOM stability prediction -which still remains an open problem.…”

Outlier-Based Domain of Applicability Identification for Materials Property Prediction Models

“…Our dataset is composed of three prior datasets consisting of experimental aqueous solubility measurements: the SOMAS dataset (10,162 molecules) Gao et al [2022], , AqSolDB (4425 molecules) Sorkun et al [2019] and the Cui dataset (7283 molecules) Cui et al [2020]. We first merge the three datasets and resolve duplicate molecules using a method described in the Supporting Information.…”

“…Prior efforts have approached DoA identification in a similar context. For example, Sutton et al [2020] presented a method based on subgroup discovery Boley et al [2012], van Leeuwen and Knobbe [2012], Nguyen and Vreeken [2015] to find such domains with high predictive accuracies based on an impact metric. In our work, we use a different strategy to identify the subgroups and rank order them in terms of their impact on the prediction accuracy.…”

Towards a comprehensive data infrastructure for redox-active organic molecules targeting non-aqueous redox flow batteries

“…For example, big data analyses elucidated a "stability cliff" in quinones, a prevalent molecular class in aqueous RFB, encouraging researchers to explore other chemical spaces. 16 Efforts are already underway to develop data-driven pipelines and apply big-data analyses for vanadium RFB 17,18 and aqueous organic RFB [19][20][21] materials. Some big data approaches have been applied to the search for NARFB materials; for example, data-enabled high-throughput screening of redox-active molecules for NARFB has been demonstrated in a small-scale proof-of-concept study where several theoretically viable molecules for NARFB anolytes were selected from ∼1400 quinoxaline-based systems a with funnelbased screening approach focusing on reduction potential, solvation energy, and structural changes with oxidation.…”