Electrochimica Acta 52 (2007) 2189–2195 Evaluation of electrolytes for redox flow battery applications M.H. Chakrabarti a, R.A.W. Dryfe b, E.P.L. Roberts a,∗ a School of Chemical Engineering and Analytical Science, The University of Manchester, P.O. Box 88, Manchester M60 1QD, UK b School of Chemistry, The University of Manchester, P.O. Box 88, Manchester M60 1QD, UK Received 17 May 2006; received in revised form 1 August 2006; accepted 17 August 2006 Available online 26 September 2006 Abstract A number of redox systems have been investigated in this work with the aim of identifying electrolytes suitable for testing redox flow battery cell designs. The criteria for the selection of suitable systems were fast electrochemical kinetics and minimal cross-contamination of active electrolytes. Possible electrolyte systems were initially selected based on cyclic voltammetry data. Selected systems were then compared by charge/discharge experiments using a simple H-type cell. The all-vanadium electrolyte system has been developed as a commercial system and was used as the starting point in this study. The performance of the all-vanadium system was significantly better than an all-chromium system which has recently been reported. Some metal–organic and organic redox systems have been reported as possible systems for redox flow batteries, with cyclic voltammetry data suggesting that they could offer near reversible kinetics. However, Ru(acac)3 in acetonitrile could only be charged efficiently to 9.5% of theoretical charge, after which irreversible side reactions occurred and [Fe(bpy)3](ClO4)2 in acetonitrile was found to exhibit poor charge/discharge performance. © 2006 Elsevier Ltd. All rights reserved. Keywords: Redox flow battery; Vanadium; Chromium; Ru(acac)3; [Fe(bpy)3](ClO4)2 1. Introduction electrolytes by transport through the membrane. For example, there has been little recent interest in the development of the Redox flow batteries are electrochemical energy storage iron/chromium redox flow cell due to this problem [4]. Toredress devices that utilise the oxidation and reduction of two soluble this issue, an all-chromium redox electrolyte was investigated at redox couples for charging and discharging. They differ from the University of Manchester and the charge/discharge charac- conventional batteries in that the energy-bearing chemicals are teristics of a laboratory scale battery were reported [5,6]. Prior not stored within at the electrode surface, but in separate liquid to this, other workers have performed extensive investigations reservoirs and pumped to the power converting device for either on the all-vanadium redox system [3,7–10] and patented the charging or discharging [1,2]. Due to the use of two soluble redox technology [11]. In addition, an all-neptunium system has been couples, solid-state reactions with their accompanying morpho- evaluated [12], although the hazards of working with radioac- logical changes at the electrodes are absent [3]. Thus, there are tive electrolytes are likely to limit the practical application of this no fundamental cycle life limitations associated with these pro- system. Several prototype vanadium systems have been investi- cesses such as shedding or shape changes, which usually occur gated successfully [13–16] and some systems are well on their in conventional storage batteries. way to commercial success [17]. Despite these advantages, the redox flow battery has not been Despite such achievements, batteries employing aqueous widely exploited to date. One disadvantage of the systems devel- electrolytes have a low energy content. The energy output from oped to date is the use of two separate redox species in the half- the battery is proportional to the potential window of opera- cells, leading to the potential for cross-contamination of active tion available from the background electrolyte. The operating potential window of aqueous electrolytes is limited due to water electrolysis [5]. Organic electrolytes, which offer a wider poten- ∗ Corresponding author. Tel.: +44 161 306 8849; fax: +44 161 306 4399. tial window, have been investigated in this study. In addition, E-mail address: [email protected] (E.P.L. Roberts). species have been selected which minimize the effect of elec- 0013-4686/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2006.08.052 2190 M.H. Chakrabarti et al. / Electrochimica Acta 52 (2007) 2189–2195 trolyte cross-contamination. One approach is to use a single give an indication of the reversibility of redox couples, further system which offers three oxidation states, so that the discharged experiments are needed to demonstrate that selected systems can species is the same on each side of the cell. Such a system would be used for energy storage. For example, a redox couple may be have the advantage that any cross-contamination would only lead reversible, but the charged species may be unstable over long to some self-discharge, and little or no ‘cell balancing’ or elec- timescales, which would not be detected by cyclic voltammetry. trolyte processing would be required. An approach whereby the In this study systems which were found to exhibit fast kinetics same cation is used but with different ligands on each side of the were tested for their charge/discharge performance in a simple cell has been suggested [18], but this has not been considered in H-type cell. These experiments aimed to determine whether the this study. selected systems could be used for energy storage and to pro- Electrolyte systems can be selected on the basis of the fol- vide a preliminary indication of the relative performance of each lowing properties, which are generally desirable for redox flow system. batteries [18,19]: 2. Experimental • fast kinetics at the electrode–electrolyte interface; • a relatively large open circuit potential; 2.1. Electrolytes • reasonable cost; • high solubility in the process electrolyte. Vanadium electrolytes were prepared from vanadium (IV) sulphate (>99.99% purity, Aldrich), with the V(II)/V(III) couple In this study, the following series of redox systems in ace- generated by electro-reduction. Sulphuric acid was used as the tonitrile electrolyte were selected which apparently offered fast background electrolyte. electrode kinetics (based on literature data, e.g. [19] and [20]) Reagent grade tris(2,2-bipyridine) ruthenium (II) chloride and the potential to operate with a single electrolyte using a is available from Aldrich. Since oxidation of the chloride salt species with three oxidation states. was known to be irreversible [24], the tetrafluouroborate salt [Ru(bpy)3(BF4)2] was prepared by addition of NaBF4 in ace- (i) Ruthenium organic complexes tonitrile and precipitation of NaCl. Ruthenium acetylacetonate A number of ruthenium organic complexes which can [Ru(acac)3, 97% purity, Aldrich], tris(2,2 -bipyridine) iron(II) be both oxidized and reduced electrochemically have been perchlorate (reagent grade, GFS) and rubrene (reagent grade, reported in the literature, and some of these have been Aldrich) were used for the preparation of the respective elec- suggested as suitable candidates for a redox flow battery trolytes (Caution: perchlorate salts are potentially explosive and [21]. Tris(2,2-bipyridine) ruthenium (II) tetrafluoroborate should be handled with appropriate care). Tetraethyl-ammonium [Ru(bpy)3(BF4)2] has exhibited fast kinetics [19]. In addi- tetrafluoroborate and tetraethyl-ammonium perchlorate were tion, this system offers the possibility of cell voltages of up used as the background electrolyte. to 2.6 V, much higher than is possible in aqueous battery To remove dissolved oxygen, electrolytes were sparged for at systems [19]. Ruthenium acetylacetonate [Ru(acac)3] has least 10 min with oxygen-free dry argon (aqueous electrolytes) also been reported as offering fast oxidation and reduction or nitrogen (organic electrolytes). Water was removed from the kinetics [22] and a possible cell voltage of around 1.75 V. organic electrolytes using zeolite 4A (Merck) to a moisture level (ii) Tris(2,2-bipyridine) iron(II) perchlorate of below 0.005 wt%. This species can be oxidized and reduced [19] and offers a possible cell potential of 2.4 V. This compound is avail- 2.2. Cyclic voltammetry able commercially and is significantly cheaper than the ruthenium complexes. A graphite rod (Goodfellow) of surface area 0.06 cm2 was (iii) Rubrene used as the working electrode for cyclic voltammogram exper- Rubrene, a neutral organic species, can be oxidized and iments with the vanadium battery electrolytes. A glassy-carbon reduced electrochemically [23]. The redox potentials of electrode (I.J. Cambria Scientific) of surface area 0.07 cm2 was these reactions offer a possible cell potential of around used for cyclic voltammetry in organic media. The electrode 2.3 V. Again this compound is available commercially, was polished with alumina washed with de-ionised water and although it is significantly more expensive than the other acetone following the procedure described in literature [21]. redox species. The reference electrode used in aqueous solutions was the saturated calomel electrode along with a salt bridge. Organic These systems are compared to the all vanadium redox flow media required the use of a silver wire quasi-reference elec- battery system, which has previously been investigated in detail trode (AgQRE). A platinum counter electrode was used in each ([3,7–11]) and has been commercialized in recent years [17]. case. In this system, vanadium in four different oxidation states is Cyclic voltammetry was conducted using a standard three- used: V(II)/V(III) at the negative electrode and V(IV)/V(V) at electrode cell, with a Autolab/PGSTAT30potentiostat for poten- the positive electrode. tial control. All solutions were de-aerated prior to experiments. Each system was first tested by cyclic voltammetry in order The solution headspace was purged with inert gas for the dura- to evaluate the electrode kinetics. While cyclic voltammetry can tion of experiments.