Vanadium Electrolyte for All-Vanadium Redox-Flow Batteries: The Effect of the Counter Ion

Abstract

In this study, 1.6 M vanadium electrolytes in the oxidation forms V(III) and V(V) were prepared from V(IV) in sulfuric (4.7 M total sulphate), V(IV) in hydrochloric (6.1 M total chloride) acids, as well as from 1:1 mol mixture of V(III) and V(IV) (denoted as V^3.5+) in hydrochloric (7.6 M total chloride) acid. These electrolyte solutions were investigated in terms of performance in vanadium redox flow battery (VRFB). The half-wave potentials of the V(III)/V(II) and V(V)/V(IV) couples, determined by cyclic voltammetry, and the electronic spectra of V(III) and V(IV) electrolyte samples, are discussed to reveal the effect of electrolyte matrix on charge-discharge behavior of a 40 cm² cell operated with 1.6 M V^3.5+ electrolytes in sulfuric and hydrochloric acids. Provided that the total vanadium concentration and the conductivity of electrolytes are comparable for both acids, respective energy efficiencies of 77% and 72–75% were attained at a current density of 50 mA∙cm⁻². All electrolytes in the oxidation state V(V) were examined for chemical stability at room temperature and +45 °C by titrimetric determination of the molar ratio V(V):V(IV) and total vanadium concentration.

1. Introduction

The electrolyte, as a component of all-vanadium redox flow batteries (VRFBs), contains salts of vanadium dissolved in acids to provide ionic conductivity and enable electrochemical reactions. The charge-discharge process of VRFBs is commonly represented by a combination of the following half-cell reactions:

(cathode) VO₂⁺ + e⁻ + 2H⁺ → VO²⁺ + 2H₂O     E⁰ = 0.99 V

(1)

(anode) V²⁺ − e⁻ → V³⁺     E⁰ = −0.26 V

(2)

These reactions depict the charge and mass balance, but the counter ions are usually omitted and not considered, even though the vanadium species are ion-paired with sulfate counter ions at battery-relevant vanadium concentrations, i.e., over the one-molar range in the case of common sulfuric acid VRFB electrolytes [1,2,3,4]. Therefore, the electrochemical kinetics of vanadium reactions, and ultimately the activation losses during battery operation, are likely also dependent on the nature of the counter ion or battery chemistry. The chemistries of VRFB electrolytes, including new systems and trends in VRFB electrolyte development, have been reported in several reviews [5,6,7], but the role of counter ions has not been considered systematically. Information about the suitability of various acids as a matrix for VRFB electrolyte is contradictory. The electrolytes based on oxalic acid [8], neat hydrochloric acid [9], or mixtures of sulfuric and hydrochloric acids [10] are reported in the literature. At the same time, oxalic acid is unsuitable as an additive for the V(V) electrolyte because of its reducing ability. For example, ammonium oxalate is instead used as a rebalancing agent producing CO₂ on reaction with V(V) to form V(IV) [11]. Previously, the V(V)/V(IV) couple was investigated for use in the positive half-cell of a VRFB [12], and was considered unsuitable in combination with hydrochloric acid due to chlorine gas evolution and the formation of V(IV).

Another reason to explore new VRFB chemistries is the need to improve the thermal stability of the V(V) electrolyte, which is prone to precipitation at elevated temperatures [5,11]. There are several methods to increase the thermal stability of VRFB catholyte: application of additives [11], increase of the proton concentration by increasing the total concentration of sulfuric acid [13], or using methane sulfonic acid in higher concentrations [14], and complexation of V(V) species by chloride ions in mixed acid electrolytes [10]. It is difficult to differentiate between the proton and counter ion effects experimentally especially because the concentration of free acid remains unknown or varies with the state-of-charge (SoC) of the electrolyte.

This study is an attempt to investigate the electrochemical characteristics of V(V)/V(IV) and V(III)/V(II) couples for comparison in sulfuric and hydrochloric acids, which are the most commonly used acids for VRFB electrolytes. Since the distribution of counter ions and protons between the catholyte and anolyte during the charging of vanadium electrolyte is unknown, the conductivity of electrolyte samples obtained by electrolysis of V(IV) or 1:1 mol mixture of V(III) and V(IV) solutions (denoted as V^3.5+) are considered in this study. The concentrations of the acids were chosen to target the same conductivity of the initial V(IV) electrolyte solutions and the total vanadium concentration was set to be about 1.6 M for both acids.

2. Results and Discussion

2.1. Electrolyte Preparation

The electrolyte samples were prepared in H₂SO₄ and HCl acids using either V(IV) or V^3.5+ solution containing equimolar mixtures of V(IV) and V(III) as the initial electrolyte (Section 3.2). The abbreviations of the vanadium electrolyte samples used in the study and their analysis data are given in

Table 1. Denotations and analysis data for vanadium electrolyte samples used in the study.

Sample	Total V Conc. (M)	Molar Content (%)			Total Counter Ion Conc. (M)	Conductivity (mS∙cm−1)	Comment
		V(III)	V(IV)	V(V)
I-V4-H2SO4 **	1.66		100		4.7	330	3 M free H2SO4
I-V3-H2SO4	1.63	100				230
I-V5-H2SO4	1.61		1.5	98.5		450
I-V3.5-H2SO4	1.65	45.9	54.1			270
I-V4-HCl **	1.63		95.6	4.4	6.1	330	targeted 3 M free HCl
I-V3-HCl	1.67	98.8	1.2			160
I-V5-HCl	1.59		1.7	98.3		480	freshly prepared
I-V3.5-HCl	1.66	47.2	52.8			240
II-V4-HCl	1.57		100			365
II-V3-HCl	1.68	93.1	6.9			240
II-V5-HCl	1.54		1.7	98.3		490	freshly prepared
II-V3.5-HCl **	1.60	47.5	52.4		7.6	320
3 M H2SO4					3	810
3 M HCl					3	710

** Prepared by dissolution of vanadium salts and used as a source for other oxidation forms (Section 3.2).

and are defined in Section 3.2.

To obtain V(IV) and V^3.5+ samples in a HCl matrix, stoichiometric amounts of VCl₃, V₂O₅, and HCl were calculated based on the reaction:

2VCl₃ + V₂O₅ + 2HCl → 4VOCl₂ + H₂O

(3)

Despite the stoichiometric ratios of the reagents, the I-V4-HCl sample contained 4.4% of V(V) possibly because when VCl₃ is used as a source of V(III), it is prone to oxidation and its dissolution in HCl (even without V₂O₅) resulted in the presence of 3 to 10% mol of a V(IV) fraction. To prepare the II-V3.5-HCl sample, an excess of VCl₃ was taken to first produce a 2.2 M V^3.5+ stock solution, which was then diluted by 3 M HCl to a solution containing 1.6 M of vanadium species (II-V3.5-HCl) (Section 3.2). Such variation in the preparation procedure led to a difference between the total chloride concentration and finally the conductivity of the samples in series I and II (Table 1). The final total chloride concentrations in the electrolytes I-V4-HCl and II-V3.5-HCl were 6.1 M and 7.6 M based on the gravimetric analysis data, respectively. The conductivity of I-V4-HCl solution was 330 mS∙cm⁻¹. This value is almost two times lower than the conductivity of vanadium-free 3 M HCl solution (Table 1). This difference can be caused by ion-pairing and incomplete dissociation of vanadium species in electrolyte solution, similar to the interactions assumed to occur in sulfuric acid VRFB electrolyte [15], or because the factual concentration of the free HCl in the vanadium electrolyte samples was lower than 3 M.

To obtain the V(IV) solution with the same conductivity from vanadyl sulfate, 3 M H₂SO₄ was used as a matrix (sample I-V4-H₂SO₄). Vanadium-free 3 M H₂SO₄ solution showed higher conductivity than 3 M HCl. However, these values still correlate with the fact that dissociation of 3 M H₂SO₄ proceeds mostly during the first step in solutions with high sulfate concentrations, i.e., containing 1.6 M vanadium species [15,16]. Total sulfate concentration in I-V4-H₂SO₄ solution was 4.7 M and corresponded to the common total sulfate concentration of some commercial electrolytes.

Though the exact vanadium speciation in electrolyte solution is unknown and can involve dimeric V(V) species [4], the process of V(IV) electrolysis can be commonly represented by the following conversions:

(cathode) VO²⁺ + H₂O − e⁻ → VO₂⁺ + 2H⁺

(4)

(anode) VO²⁺ + 2H⁺ + e⁻ → V³⁺ + H₂O

(5)

where the protons are consumed in the anolyte and are formed at the catholyte. According to this, the conductivity of I-V3-H₂SO₄ is lower and that of I-V5-H₂SO₄ is higher compared with the initial I-V4-H₂SO₄ solution. In the case of HCl samples (series I), the depletion of protons in the anolyte, i.e., in I-V3-HCl, is more pronounced than in H₂SO₄. The conductivity of I-V5-HCl is 30 mS∙cm⁻¹ higher than that of I-V5-H₂SO₄ solution.

Finally, the I-V3.5 sample obtained by mixing the I-V3 and I-V4 forms was characterised by the conductivity of 270 and 240 mS∙cm⁻¹ for H₂SO₄ and HCl, respectively. In contrast, the chemically prepared II-V3.5-HCl solution exhibited higher conductivity (320 mS∙cm⁻¹); therefore, II-V4-HCl and II-V3-HCl were more conductive than I-V4-HCl and I-V3-HCl. This shows that the factual free HCl concentration in the series II is higher compared to series I, since the total vanadium concentration is the same. Both series I and II of electrolyte samples in HCl were compared with series I in H₂SO₄ regarding the electrochemical features.

2.2. UV-Vis Spectroscopy

We completed an optical spectroscopy study to evaluate whether the species of V(IV) or V(III) complexed with counter ion were present in electrolyte samples.

V(III) normally exists as an octahedral hexa-aqua cation [V(H₂O)₆]³⁺ in diluted acidic aqueous solutions [17]. The electronic absorption spectrum of V(III) species in non-complexing media such as perchloric acid solutions shows two absorption bands with maximum absorption at 400 and 590 nm [18]. The peaks can be attributed to the ³T_1g(F) → ³T_1g(P) transition at around 400 nm and the ³T_1g(F) → ³T_2g transitions at around 600 nm. At high acid concentrations, the water molecules from the vanadium solvation sphere can be substituted by chloride or sulfate anion in acidic media [18,19]. Both the V(III) spectra in H₂SO₄ (I-V3-H₂SO₄) or in HCl (I-V3-HCl and II-V3-HCl) (Figure 1a) exhibited a bathochromic shift, indicating the presence of complexed vanadium species in all the samples. Another indication of a coordination of chloride and sulfate was found in the peak shape of the ³T_1g(F) → ³T_1g(P) transition (Figure 1a). The peak features a non-Gaussian shape. One possible explanation of this irregular peak shape could be a slight trigonal distortion of the inner coordination sphere of the V(III) ion. The wavelengths of maximum absorption for V(III) and V(IV) electrolyte samples are summarized in

Table 2. Spectrophotometric data for vanadium(III) and vanadium(IV) species in various electrolyte matrixes.

Sample	V(III)		V(IV)
	3T1g(F) → 3T1g(P)	3T1g(F) → 3T2g	2B2 → 2B1	2B2 → 2E(I)
I-V3-H2SO4, I-V4-H2SO4	403 nm	610 nm	633 nm	760 nm
I-V3-HCl, I-V4-HCl	413 nm	605 nm	633 nm	760 nm
II-V3-HCl, II-V4-HCl	414 nm	620 nm	633 nm	760 nm

. Similar values of absorption maxima were reported previously by Vijayakumar et al. [3] for 1.5 M V(III) solutions in pure chloride and sulfate acidic media. The spectra for I-V3-HCl and II-V3-HCl solutions show the same features, even though the difference in conductivities between these electrolytes is about 80 mS∙cm⁻¹ and II-V3-HCl contained 7% of V(IV) (Table 1). This may indicate the difference in the amount of free HCl rather than the difference in the V(III) speciation.

The dissolution of VOSO₄ in an acidic aqueous solution yields the hydrated vanadyl cation [VO(H₂O)₅]²⁺, commonly abbreviated as VO²⁺ [17]. For the V(IV) electrolytes, the transitions of the vanadyl ion are usually attributed according to the model of Ballhausen and Gray [20]. The broad peak around 760 nm (Figure 1b) can be attributed to the ²B₂ → ²E(I) transition as the most intense transition. The shoulder at 630 nm can be assigned to the vibronically allowed ²B₂ → ²B₁ transition. The exact values for each electrolyte are summarized in Table 2. Since I-V4-HCl solution contained 4.4% of V(V) (Table 1), an equivalent amount of I-V3-HCl solution was added to the I-V4-HCl to remove the traces of V(V) before the spectrum was recorded. The HCl electrolytes (series I and II) as well as the H₂SO₄ exhibited the same wavelength for both transitions. This is in accordance with the assumptions that the coordination sphere of V(IV) remained unchanged in both of the electrolyte media and counter ions were only weekly bound to the VO²⁺ core [21].

To conclude, the chemical state of the V(III) species, i.e., the anolyte of VRFB, is more sensitive to the nature of the counter ion or acid in the electrolyte composition than the V(IV) form or catholyte of VRFB. The presence of chloride or sulfate anions in the inner coordination sphere of V(III) in I-V3-HCl, II-V3-HCl, and I-V3-H₂SO₄ samples was confirmed by electronic optical spectroscopy.

2.3. Cyclic Voltammetry

The cyclic voltammograms for V(III) and V(IV) electrolyte samples are usually characterized by quasi-reversible V(V)/V(IV) and V(III)/V(II) redox transitions [22,23]. Typical cyclic voltammograms for V(III) and V(IV) electrolyte samples at an oxidatively pre-treated glassy carbon electrode are presented in Figure 2. This pre-treatment was necessary because the voltammetric response in the solution containing V(III) and V(IV) species in sulfuric acid is sensitive to the pre-conditioning of the electrode [24,25].

All the voltammograms were considered further in terms of peak separation (∆E) and half-wave potentials (E_1/2) (

Table 3. Electrochemical characteristics of redox couples of anolyte and catholyte: half-wave potential (E_1/2) and peak-to peak separation (∆E).

Sample	Redox Couple	E1/2 (V) vs. Ag/AgCl	E1/2 (V) vs. Hg/Hg2SO4	∆E (V)	Ucell, theor. (V)
I-V3-H2SO4	V(III)/V(II)		−0.99	0.23	1.43
I-V4-H2SO4	V(V)/V(IV)		0.44	0.13
I-V3-HCl	V(III)/V(II)	−0.42		0.25	1.39
I-V4-HCl	V(V)/V(IV)	0.97		0.16
II-V3-HCl	V(III)/V(II)	−0.41		0.14	1.41
II-V4-HCl	V(V)/V(IV)	1.00		0.13

). The E_1/2 values are calculated as a mid-point potential of anodic and cathodic peaks ((E_pa + E_pc)/2) and can be taken as the formal potential (E^0’_f), if the diffusion coefficients of the species in reduced and oxidized states are similar (D_red = D_ox) [25]. Whether this requirement can be fulfilled depends on the exact vanadium speciation, which remains unknown. However, if following common practice, the E_1/2 values are taken as the formal potentials for the vanadium redox couples, then the theoretical cell voltage under open-circuit conditions is determined by the difference in E_1/2 for cathodic and anodic couples. As such, we obtained the open-circuit cell voltage (U_cell,theor) value of 1.43 V for the electrolyte in H₂SO₄ (I-V3.5-H₂SO₄). The values of U_cell,theor for other electrolyte samples are shown in Table 3. In the case of HCl electrolyte series I, the cell voltage was calculated to be lower (1.39 V). The redox potential of the V(V)/V(IV) couple in II-V4-HCl electrolyte solution is 20 mV higher than the value for I-V4-HCl sample. This may be correlated with the fact that the II-V4-HCl has a small excess of free acid compared to I-V4-HCl, as mentioned in Section 2.1. Since the redox potential of the V(V)/V(IV) couple is proton-dependent, the increase in HCl concentration would cause the E^0’_f to shift to higher values. The difference between the E_1/2 values for the V(III)/V(II) couple in I-V3-HCl and II-V3-HCl electrolytes is insignificant considering the usual error of the redox potential estimation by cyclic voltammetry (about 10 mV). Though the E^0’_f of V(III)/V(II) redox transition is commonly expected to be influenced by the complexation of V(III) species with chloride ions, the theoretical cell voltage under open-circuit conditions for series II in HCl matrix is 1.41 V and very close to that in H₂SO₄.

The V(V)/V(IV) couple was characterized by the peak-to-peak separation (∆E) value of 0.13–0.16 V in all the electrolytes: I-V4-HCl, II-V4-HCl, and I-V4-H₂SO₄ (Table 3). These solutions are comparable regarding the conductivity (Table 1). Generally, the ∆E is affected by both the uncompensated ohmic drop between working and reference electrodes (which is proportional to the electrolyte conductivity) and by the kinetics of the redox reaction. Since the conductivity of I-V3-H₂SO₄ solution was higher than that of I-V3-HCl (Table 1), it is difficult to differentiate between these two factors. However, the V(III)/V(II)redox couple in solutions with the same conductivity, I-V3-H₂SO₄ and II-V3-HCl, was characterized by a lower ∆E value in the HCl matrix in contrast with H₂SO₄. This can be attributed to the difference in the coordination sphere due to complexation (Section 2.2).

Even if the total concentration of vanadium species in HCl and H₂SO₄ samples is similar (Table 1), peak currents of voltammograms in HCl series I and II are higher than those in H₂SO₄ (Figure 2). Since the H₂SO₄ and HCl have different viscosities, the difference in peak currents is hardly possible to be attributed to the difference in diffusion coefficients of vanadium species. The lower peak currents may also be associated with the presence of more than one complex species in solution that have slightly different redox potentials, also leading to the broadening of the peaks.

To investigate the characteristics of the electrolyte under current flow conditions, i.e., battery operation conditions, charge-discharge tests were to be performed using HCl and H₂SO₄ electrolytes.

2.4. Cell Test

The typical charge-discharge cycles for the cells operated with HCl and H₂SO₄ vanadium electrolytes in V^3,5+ oxidation form are presented in Figure 3. The ohmic resistance of these cells was verified by electrochemical impedance measurements before charging, which was about 70 mOhm for all cells despite the difference in electrolyte conductivities of 30–80 mS∙cm⁻¹ (Table 1). Both charge and discharge voltages (U_cell) were in the range of 1.50 to 1.55 V and 1.21 to 1.23 V, respectively, and the voltage efficiency was 78% for I-V3.5-H₂SO₄ and 80% for I-V3.5-HCl and II-V3.5-HCl electrolytes (Figure 3). The small difference in the theoretical cell voltage, which was expected based on the cyclic voltammetry data, is correlated with the values of cell voltage U_cell obtained during cell operation. The cells exhibited coulombic and energy efficiencies of 90–97% and 72–77%, respectively (

Table 4. Parameters for the cells evaluated in the cell test for HCl (I-V3.5-HCl, II-V3.5-HCl) and H₂SO₄ (I-V3.5-H₂SO₄) vanadium electrolytes.

Parameter	I-V3.5-H2SO4		I-V3.5-HCl		II-V3.5-HCl
	Charge	Discharge	Charge	Discharge	Charge	Discharge
φA − φA, redox (V) **	−0.04 ± 0.01	0.06 ± 0.02	−0.05 ± 0.01	0.03 ± 0.01	−0.02 ± 0.01	0.02 ± 0.01
φC − φC, redox (V) **	0.03 ± 0.01	−0.02 ± 0.01	0.02 ± 0.01	−0.02 ± 0.01	0.03 ± 0.01	−0.03 ± 0.01
Ucell − (φC − φA) (V) **	0.09 ± 0.01	−0.10 ± 0.01	0.080 ± 0.01	−0.10 ± 0.01	0.10 ± 0.01	−0.11 ± 0.01
Capacity (Ah∙L−1) *	15.7	15.3	17.5	15.7	14.8	13.9
Energy (Wh∙L−1) *	24.1	18.5	26.5	19.0	22.7	17.1
Coulombic efficiency (%)	97		90		94
Energy efficiency (%)	77		72		75

* For the second cycle, ** the average value for three consecutive cycles.

). The values of coulombic efficiency were common for VRFB and consistent with the characteristics reported in the literature [9]. However, the values of coulombic and energy efficiency for the I-3.5-HCl were the lowest compared to II-V3.5-HCl and I-V3.5-H₂SO₄.

To further reveal the differences between the cells with HCl- and H₂SO₄-based electrolytes, the correlation between U_cell, half-cell, and redox potentials was considered. The cell voltage can be described as follows [26]:

U_cell = φ_C − φ_A + IR_cell + φ_memb

(6)

where I is the current, R_cell is the ohmic resistance of the cell, and φ_memb is the membrane potential.

The sum of IR_cell and φ_memb is therefore determined by the term U_cell − (φ_C − φ_A), as derived from Equation (6). If IR_cell is equal due to the same cell materials and chosen concentration of acids to target similar electrolyte conductivity, the difference in U_cell − (φ_C − φ_A) should be related to the membrane potential φ_memb. Table 4 shows that the U_cell − (φ_C − φ_A) values are similar for both of the HCl and H₂SO₄ systems during the charge and discharge step within the error of the experiment. The variance in the values given in Table 4 were due to variations in the cycles and correspond to a standard deviation from the average value.

Half-cell potentials can be expressed as follows:

φ_C = φ_C,redox + η_act(C) + η_con(C)

(7a)

φ_A = φ_A,redox + η_act(A) + η_con(A)

(7b)

where η_con(C) and η_con(A) are the concentration overpotentials for the cathode and anode, respectively; and η_act(C) and η_act(A) are the activation overpotentials for cathodic and anodic reactions, respectively. The sum of the overpotentials η_act + η_con for the cathode and anode can then be estimated from the terms (φ_C − φ_C,redox) and (φ_A − φ_A,redox) in Equations (7a) and (7b), respectively, which were calculated as average values from the charge-discharge cycle, which are shown in Table 4. (φ_C − φ_C,redox) is about 20 to 30 mV for the charge and discharge processes for all the cells, i.e., with HCl and H₂SO₄ electrolytes.

As for the anodic reaction, the term (φ_A − φ_A,redox) reaches values of 30–60 mV for charge and discharge steps in the case of I-V3.5-H₂SO₄ and I-V3.5-HCl electrolytes, i.e., higher than II-V3.5-HCl. Since η_con(C) and η_con(A) mostly depend on the cell construction and operation mode, which were the same for the HCl and H₂SO₄ systems, the major contribution to the difference between the (φ_A − φ_A,redox) term was expected from the anodic activation overpotential (η_act(A)). This may indicate the slower kinetics of the anodic reaction (V(III)/V(II) conversion) in the case of I-H₂SO₄ and I-HCl systems compared with II-HCl. Higher peak-to-peak separation for the V(III)/V(II) couple in I-V3-H₂SO₄ and I-V3-HCl than in II-V3-HCl (Table 3) is correlated with the (φ_A − φ_A,redox) values.

Except for the differences in the activation for the anodic reaction, the charge-discharge characteristics of the cells with HCl and H₂SO₄ electrolytes used in this work reveal no considerable difference.

Further cyclic performance of the cells was tested under conditions that are more closely related to common protocol. The cells were charged and discharged at the current density of 50 mA∙cm⁻¹, though we did not maintain the cell at open-circuit potential between the charge-discharge cycles. The end-of-charge and discharge voltages were set to 1.65 V and 0.8 V, respectively, for all cells, though the expected open-circuit cell voltage (from the formal potentials, Table 3) for I-V3.5-H₂SO₄ system should have been 20–40 mV higher than for the cell with I-V3.5-HCl and II-V3.5-HCl electrolytes. The capacity of the cell with II-V3.5-H₂SO₄ degraded to 50% of its initial value after 19 cycles under these conditions (Figure 4) and crossover of electrolyte occurred, leading to a variation in the electrolyte volumes in the anodic and cathodic tanks.

Higher capacity loss during continuous charge-discharge cycling i.e., to 30% of the initial value, was also detected in the case of the I-V3.5-HCl electrolyte. The crossover of the electrolyte was more pronounced and traces of an orange solid substance, probably vanadium pentoxide, were observed at the membrane after the cell was disassembled. Compared to the II-V3.5-HCl electrolyte, the I-V3.5-HCl electrolyte retained only 30% of the initial capacity and a large amount of the orange deposit was observed on the membrane at the end of the test. As the main difference between the I-V3.5-HCl and II-V3.5-HCl systems is in their conductivity (at constant total vanadium concentration), the stronger capacity decay in the case of I-V3.5-HCl can be assigned to the lower protons or free acid concentration in this electrolyte samples compared to II-V3.5-HCl or I-V3.5-H₂SO₄.

2.5. Thermal Stability

The poor thermal stability of the catholyte at state-of-charges (SoC) higher than 80% is a well-known issue for electrolytes based on sulfuric, methane sulfonic acids, or mixtures of both [5]. As for the vanadium electrolyte containing HCl, there are two concerns about chlorine evolution: (1) electrochemical chlorine evolution from the supporting electrolyte, especially if the cell is overcharged, and (2) chemical generation of chlorine from HCl due to its oxidation by V(V). The electrochemical chlorine evolution depends on the cell operation protocol and should not appear unless a large overpotential is applied. To avoid misinterpretation and to assess the thermal stability of V(V) electrolytes at elevated temperatures, it was first necessary to check its chemical stability at room temperature. Otherwise, a possible chemical reduction of V(V) in the case of reducible supporting electrolyte anions would lead to a decrease in the SoC of the catholyte and to an increase in induction time for the thermally-induced precipitation of the V(V) species. For this purpose, we analysed the V(V) samples of electrolytes (I-V5-H₂SO₄, I-V5-HCl, and II-V5-HCl), which were prepared by electrolysis (Table 1), by titration after storage at room temperature for several weeks and after the sample was kept at 45 °C. The titration data of the I-V5-H₂SO₄ sample remained unchanged at 98% of SoC at room temperature for least six weeks (

Table 5. Titration results for the catholyte samples with a HCl and H₂SO₄ matrix.

Sample	Conditions	Total V Concentration (M)	Molar Content (%)		Comment
			V(IV)	V(V)
I-V5-H2SO4	initial	1.61	1.5	98.5
	after 3 weeks at RT	1.62	1.4	98.6	unchanged
	after 6 weeks at RT	1.62	1.6	98.4	unchanged
	after 1 day at 45 °C	1.62	2	98.0	traces of precipitate
	after 5 weeks at RT and 2 days at 45 °C	1.40	1.8	98.2	precipitation
I-V5-HCl	initial	1.59	1.7	98.3
	after 3 weeks at RT	1.60	3.1	97.0	SoC shift
	after 6 weeks at RT	1.59	4.0	96.0	SoC shift
	after 1 day at 45 °C	1.55	4.1	95.9	precipitation
	after 5 weeks at RT and 2 days at 45 °C	1.40	6.0	94.0	precipitation
II-V5-HCl	initial	1.54	1.7	98.3
	after 3 weeks at RT	1.52	5.4	94.6	SoC shift
	after 5 weeks at RT	1.53	7.3	92.7	SoC shift
	after 1 day at 45 °C	1.54	9.0	91.0	SoC shift
	after 4 weeks at RT and 2 days at 45 °C	1.53	11.4	88.6	SoC shift

). After exposure of I-V5-H₂SO₄ to 45 °C for one day, only traces of precipitate formation could be observed; however, the total vanadium concentration and SoC remained unchanged. The decrease in the total vanadium concentration to 1.4 M and precipitation was observed in the sample after a two days exposure to 45 °C.

The reference sample of I-V5-HCl, which was stored at room temperature, showed a slight shift in SoC (by 1% during the three weeks and 2% after six weeks). The precipitation of vanadium pentoxide at 45 °C, with a measurable decrease in the total vanadium concentration, and shift in the SoC from 98 to 96% occurred in the sample I-V5-HCl after one day. Further decrease in SoC and more considerable formation of vanadium pentoxide occurred when the sample was kept over two days at 45 °C after storage at room temperature for five weeks. In comparison, the process of the V(V) reduction is faster in II-V5-HCl. However, there was no change in the total vanadium concentration and no deposition of vanadium pentoxide.

The SoC for II-V5-HCl at room temperature decreased to 94.6% (by 4%) over three weeks. The SoC decreased to 91% when the freshly prepared sample II-V5-HCl was heated to 45 °C. The II-V5-HCl sample, which was stored five weeks at room temperature, exhibited a further SoC shift by almost 10% after exposure to 45 °C for a week. If the decrease in the SoC can be attributed to the reduction of V(V) species by chloride ions, it should be dependent, among others, on the formal potential of the V(V)/V(IV) couple and on the amount of free acid, i.e., the activity of protons. The I-V5-HCl and II-V5-HCl samples, which were characterized by different conductivities (and presumable proton concentrations), exhibited different stabilities after exposure to elevated temperature. The fact that the reduction of V(V) to V(IV) proceeded faster is correlated with the redox potentials of the V(V)/V(IV) couple in HCl electrolyte of series I and II (Table 3). The absence of precipitate in II-V5-HCl electrolyte is likely due either to the decrease in SoC or to higher free acid concentration , rather than to the nature of the counter ion.

Generally, to address the question of V(V) chemical stability in HCl at given concentrations, it is necessary to consider the difference in the redox potentials of the V(V) reduction and chloride oxidation under real battery relevant conditions. The precise values of activity coefficients or information about the chemical state of reacting species in an electrolyte solution should be known to find the thermodynamic conditions required for electrolyte stability. Otherwise, the limit of the SoC of less than 100% in the case of HCl-based electrolytes (at least at 1.6 M vanadium and 6–7.6 M total chloride concentrations) appears to be the easiest condition to avoid chlorine evolution due to the reaction of V(V) with chloride ions. A VRFB operated with H₂SO₄-containing electrolyte is also known to have a limit of less than 80% SoC due to the thermal aging of the catholyte at elevated temperatures [4].

For electrolytes with 1.6 M total vanadium concentration, the application of HCl as an electrolyte matrix requires the careful control of the free acid concentration. The increase in the concentration of free acid can result in the shift of the formal potential of the V(V)/V(IV) redox couple to higher values and favor the chemical oxidation of chloride ions. Low concentrations of free acid lead to the precipitation of vanadium pentoxide at elevated temperatures, similar to the behavior of vanadium electrolyte in sulfuric acid. Regarding the electrochemical parameters of the cell operation, both electrolyte matrixes, HCl and H₂SO₄, were comparable, provided that the total vanadium concentration and conductivity of the electrolyte samples were equal.

3. Materials and Methods

3.1. Chemicals

Sulfuric acid (95%, supra) and hydrochloric acid (32%, p.a.) were Carl-Roth (Karlsruhe, Germany) reagents. Vanadium pentoxide (98%) and vanadyl sulfate hydrate (technical grade), purchased from Chempur (Karlsruhe, Germany), were used without purification. The amount of crystallization water in VOSO₄∙xH₂O was determined by potentiometric titration. Vanadium(III) chloride (97%) was used as a Sigma-Aldrich and Merck (EMD Millipore, Burlington, MA, USA) reagent. The commercial standard solution for conductivity measurements was obtained from Carl-Roth (Karlsruhe, Germany).

3.2. Electrolyte Preparation

The two pathways used to produce the electrolyte samples in various oxidation states are shown in Figure 5. V(III) and V(V) oxidation states of electrolyte in H₂SO₄ were prepared by the electrolysis of vanadyl sulfate dissolved in H₂SO₄ (sample I-V4-H₂SO₄). In the case of HCl, both Scheme I and II were used. For both series of samples in HCl matrix, the preparation procedure was based on the comproportionation of V(III) and V(V). For this reaction, VCl₃ was first dissolved in 3 M HCl solution with subsequent addition of stoichiometric amounts of V₂O₅ and concentrated HCl (

Table 6. The amounts of reagents used for the preparation of initial vanadium electrolyte solutions.

Sample	VCl3	V2O5	HCl		Targeted Concentration
			HCl 3 M	HCl 32% w/w	Vanadium
I-V4-HCl	0.4 moles	0.2 moles	0.5 L (1.5 moles)	37 mL (0.4 moles)	1.6 M
II-V3.5-HCl stock *	0.85 moles	0.13 moles	0.5 L (1.5 moles)	25 mL (0.26 moles)	2.2 M
Sample	VOSO4∙5.1H2O		H2SO4 95% w/w	H2O	Targeted concentration
					Vanadium
I-V4-H2SO4	204 g (0.8 moles)		84 ml (1.5 moles)	to 0.5 L	1.6 M

* II-V3.5-HCl was prepared by dilution of the II-V3.5-HCl stock with 3 M HCl.

). V₂O₅ was used in an amount so that the resulting solution contained either V(IV) (series I, I-V4-HCl) or mixture of V(III) and V(IV) in a 1:1 molar ratio (series II, II-V3.5-HCl). The total concentration of chloride in the samples I-V4-HCl and II-V3.5-HCl was determined gravimetrically.

The V(IV) solutions in series I (I-V4-HCl, I-V4-H₂SO₄) were subjected to electrolysis to prepare the V(III) (I-V3-HCl, I-V3-H₂SO₄) and V(V) (I-V5-HCl, I-V5-H₂SO₄) forms. The electrolyte in V^3.5+ form in series I (I-V3.5-HCl, I-V3.5-H₂SO₄), used for the charge-discharge trial, was obtained by mixing the V(III) and V(IV) electrolyte samples in a 1:1 v/v ratio in HCl or H₂SO₄ matrix, respectively (Figure 5).

The electrolyte in V^3.5+ form in series II (II-V3.5-HCl) was first converted into V(III) (II-V3-HCl) and V(IV) (II-V4-HCl) forms, then to V(II) (II-V2-HCl) and V(V) forms (II-V5-HCl) by charging the cell (Figure 5). After this charging step, the charge-discharge cell test was performed for the electrolyte series II cell.

A 40 cm² cell with the same design as described for the cell test (Section 3.4) was used for electrolysis. The electrolysis was carried out in galvanostatic-potentiostatic mode with a maximal voltage of 1.65 V and a maximal current of 3 A.

For the cyclic voltammetry investigations in the anodic and cathodic regions of potential, V(III) and V(IV) samples of the electrolyte were used.

3.3. Electrolyte Characterization

3.3.1. Titration and Gravimetric Analysis

The total vanadium concentration and the molar content of vanadium in various redox states in electrolyte samples were determined by potentiometric cerimetric titration. The titration was carried out with 0.1 M cerium(IV) sulfate standard solution using a Titrator T70 (Mettler Toledo Int. Inc., Giessen, Germany).

The concentration of chloride ions in chemically prepared HCl-containing electrolyte samples was determined by gravimetric analysis with silver nitrate.

3.3.2. Conductivity Measurement

We measured the electrolyte conductivity in a four-electrode glass cell with a cell constant of 7.3 mS∙cm⁻¹ and glassy carbon electrodes. The cell was calibrated with a standard conductivity solution (500 mS∙cm⁻¹). The AC impedance technique was used in the frequency range from 10 to 100 kHz with AC of 10 mA and 0 DC.

3.3.3. Optical Absorption Spectra (UV-Vis)

The absorbance of vanadium electrolyte samples was recorded with a Shimadzu (Duisburg, Germany) 1650PC ultraviolet-visible (UV-vis) double beam spectrophotometer. These measurements were recorded in matched quartz cuvettes with 0.2 mm optical path length, so that dilution of the samples was not necessary.

3.3.4. Cyclic Voltammetry (CV)

Cyclic voltammetric measurements were recorded in a three-electrode glass cell at room temperature using a Gamry Reference (Warminster, PA, USA) 3000 potentiostat. A platinum plate served as a counter electrode. The voltammograms were recorded and are shown using different scales of potential against reference electrodes containing chloride and sulfate ions to avoid the problem of liquid-junction potential. All potentials in the HCl matrix solutions were measured and reported versus a Ag/AgCl/3 M NaCl reference electrode. A Hg/Hg₂SO₄/K₂SO₄(sat.) reference electrode was used to measure the voltammetric response in sulfate-containing electrolyte samples. A glassy carbon disc working electrode (1 mm diameter, ALS, Tokyo, Japan) was polished using diamond pastes (6, 1, and 0.25 μm) and thoroughly washed with water. The working electrode was additionally electrochemically pre-treated after conventional polishing. The electro-oxidative pretreatment of a glassy carbon electrode was completed in a separate three-electrode cell filled with 2 M H₂SO₄ solution by application of six cyclic potential scans in the range from 0 to 2.0 V against a Ag/AgCl reference electrode at 20 mV∙s⁻¹. The final applied potential was 0 V.

3.3.5. Thermal Stability

For thermal stability trials, a climate chamber conditioned at 45 °C (Weiss WKL 64, Reiskirchen-Lindenstruth, Germany) was used to keep 1-mL electrolyte samples in closed vials for a predefined period of time. The molar content of V(V) and V(IV) and total vanadium concentrations were determined afterward by titration. If necessary, the liquid phase was separated from precipitate formed at 45 °C by filtration.

3.4. Cell Test

A charge-discharge cell test was performed in a 40 cm² cell, which was assembled using a Fumasep FAP-450 anion exchange membrane (FuMa-Tech GmbH, Bietigheim-Bissingen, Germany) as a separator, GFA6 graphite felt (SGL Group, Meitingen, Germany) and FU 4369 graphite bipolar plates (Schunk Kohlenstofftechnik GmbH, Heuchelheim, Germany) were used as electrode and current collector materials, respectively (for details of the cell design, see Noack et al. [27]). The felts were pre-treated for one hour at 400 °C in an oven before the cell assembly. The electrolyte flow rate was 75–80 mL/min.

The high-frequency resistance of the cells was determined before the charge-discharge test using AC impedance with an amplitude of 10 mA at open circuit voltage at a SoC of 0%.

The charge-discharge cell test was performed using a Gamry Reference (Warminster, PA, USA) 3000 potentiostat in galvanostatic mode at 50 mA cm⁻². The end-of-charge voltage was set to 1.65 V and the end-of-discharge voltage was 0.8 V. To monitor the cathodic (φ_{C, redox}) and anodic (φ_{A, redox}) redox potentials and cathodic (φ_C) and anodic (φ_A) half-cell potentials, the cell was equipped with additional glassy carbon electrodes and reference electrodes (Ag/AgCl or Hg/Hg₂SO₄, respectively) incorporated into the catholyte and anolyte circuit (Figure 6). Redox potentials and half-cell potentials were measured using additional electrometer channels of the Gamry (Warminster, PA, USA) potentiostat.

4. Conclusions

A comparison of 1.6 M vanadium electrolytes, prepared from V(IV) in sulfuric acid (4.7 M total sulfate) and from V(IV) or V^3.5+ in hydrochloric acid (6.1, 7.6 M total chloride), led us to the following conclusions:

(1)

The electrolyte preparation procedures should be considered to differentiate between the proton effect of an acidic matrix and the complexing effect or nature of the counter ions on the electrochemical properties of V(III)/V(II) and V(V)/V(IV) couples, and on the stability of the fully charged V(V) electrolyte at elevated temperatures.

(2)

Provided that the electrolyte samples have the same conductivity and total vanadium concentration, the characteristics of the cell operated with these electrolytes (capacity, energy efficiency, and activation losses) are similar. The cyclability of the cell with sulfuric acid electrolyte is better compared to the cells with hydrochloric acid.

(3)

V(V) in catholyte at an SoC of 98% in hydrochloric acid undergoes reduction by chloride ions, which can proceed slower or faster depending on the amount of free acid in the electrolyte and the temperature.