Bio-Inspired Electroactive Organic Molecules for Aqueous Redox Flow Batteries. 1. Thiophenoquinones
详细信息 查看全文
- 作者：Sergio D. Pineda Flores ; Geoffrey C. Martin-Noble ; Richard L. Phillips ; Joshua Schrier
- 刊名：Journal of Physical Chemistry C
- 出版年：2015
- 出版时间：September 24, 2015
- 年：2015
- 卷：119
- 期：38
- 页码：21800-21809
- 全文大小：610K
- ISSN：1932-7455
文摘
Redox flow batteries (RFB) utilizing water-soluble organic redox couples are a new strategy for low-cost, eco-friendly, and durable stationary electrical energy storage. Previous studies have focused on benzoquinones, napthoquinones, and anthraquinones as the electroactive species. Here, we explore a new class of molecules鈥攖hiophenoquinones鈥攕pecifically focusing on the caldariellaquinone-, sulfolobusquinone-, and benzodithiophenoquinone-like frameworks that are used for metabolic processes in thermophilic aerobic Sulfolobus archaebacteria. We demonstrated that B3LYP/6-311+G(d,p) thermochemical calculations (using the SMD solvation model) reproduce experimental reduction potentials to within 卤0.04 V. We then studied the effect of amine, hydroxyl, methyl, fluoro, phosphonic acid, sulfonic acid, carboxylic acid, and nitro functional group modifications on the reduction potential and Gibbs energy of solvation in water (using density functional theory) and aqueous solubility (using cheminformatics). Next we enumerated all of the 10鈥?11 possible combinations of functional group substitutions on these frameworks and identified 1056 potential molecules with solubilities exceeding 2 mol/L; of these, 36 molecules have reduction potentials below 0.25 V and 15 molecules above 0.95 V (versus the standard hydrogen electrode (SHE)). The combination of high solubility and wide voltage range makes these molecules promising candidates for high performance aqueous RFB applications. Finally, using our data set of ab initio reduction potentials, we developed a cheminformatics model that predicts ab initio reduction potentials to within 卤0.09 V based solely on molecular connectivity. We found that a model trained with as few as 200 examples generates rank-ordered predictions allowed us to identify the highest performance candidates with half the number of ab initio calculations. This offers a strategy for improving the tractability of future computational searches for high performance RFB molecules.