The importance of the choice of solvent in a non-aqueous redox flow battery (NARFB) cannot be overemphasized. Several studies demonstrated the influence of the solvent on electrolyte performance in terms of reaction rates, energy/power densities, and efficiencies. In this work, we investigate capacity decay as a direct consequence of varying reactant crossover rates through membranes in different solvent environments. Specifically, we demonstrate the superiority of an 84/16 vol% acetonitrile/1,3 dioxolane solvent mixture over pure acetonitrile in terms of energy efficiency (up to 89%) and capacity retention for vanadium NARFBs - while incorporating a Nafion 115 membrane. The permeability of Nafion to the vanadium acetylacetonate active species is an order of magnitude lower when pure acetonitrile is replaced by the solvent mixture. A method to estimate relative membrane permeability is formulated from numerical analysis of self-discharge experimental data. Furthermore, tests on a modified Nafion/S10(2) membrane, which generally offered low species permeability, also show that different solvents alter membrane permeability. Elemental and morphological analyses of cycled Nafion and NafionSi membranes in different solvent environments indicate that different crossover rates induced by the choice of solvent during cycling are due to changes in the membrane microstructure, intrinsic permeability, swelling rates, and chemical stability. (C) 2016 Elsevier B.V. All rights reserved.
