Enhanced Surface Area Carbon Cathodes for the Hydrogen–Bromine Redox Flow Battery

: The hydrogen–bromine redox ﬂow battery is a promising energy storage technology with the potential for capital costs as low as 220 $ kWh − 1 and high operational power densities in excess of 1.4 W cm − 2 . In this work, enhanced surface area bromine electrodes incorporating carbon black (CB) and graphene nanoplatelets (GnPs) on carbon paper and carbon cloth substrates were investigated, and the effect of electrolyte concentration on performance of the electrodes was studied. Carbon-black-modiﬁed electrodes are found to possess the largest electrochemically active surface areas, i.e., up to 11 times that of unmodiﬁed materials, while GnP electrodes are shown to have superior kinetic activity towards the bromine electrode reaction. In terms of performance, lower electrolyte concentrations are found to favour the improved kinetic parameters associated with graphene nanoplatelet electrodes, while highly concentrated electrolytes favour the larger electrochemically active surface area of carbon black electrodes. The optimal performance was achieved on a carbon-black-modiﬁed carbon cloth electrode in a 6 M HBr/2 M Br 2 electrolyte concentration, with polarisation current densities approaching 1.6 A cm − 2 at overpotentials of ± 400 mV, and mean overpotentials of 364 mV during oxidation and 343 mV during reduction, resulting from bromine oxidation/reduction cycling tests at ± 1.5 A cm − 2 .


Introduction
Energy storage technologies (ESTs) are advantageous to energy systems in several ways. The applications of ESTs for electrical infrastructure include the deferment of network reinforcement investment through load levelling, peak shaving services, and grid frequency regulation [1,2]. ESTs can also facilitate the implementation of electricity generation from renewable sources. Many renewable energy sources are intermittent and/or unpredictable, leading to gaps between demand and supply which can be managed by using ESTs [3,4].
Redox flow batteries (RFBs) are one type of EST that demonstrate promise for the applications outlined above. These have a number of desirable characteristics, including design flexibility, a long lifetimes, high efficiency, rapid response times, potentially low costs, and a lack of geographical requirements for deployment [3][4][5][6][7][8]. The hydrogen-bromine RFB was first reported in 1980 [9]. The operation of the H 2 -Br 2 RFB involves the reduction of protons and evolution of gaseous hydrogen at the anode during charge and the oxidation of hydrogen during discharge, as shown in Reaction (1). At the cathode, bromide ions are oxidised to bromine during charge, and the reverse reaction occurs upon discharge (Reaction (2)). The electrode potentials are given vs. the standard hydrogen electrode (SHE), and the overall cell Reaction (3) possesses a potential of 1.09 V at 298 K [10][11][12][13].

Electrode Materials and Preparation
Carbon paper (SGL Carbon, Sigracet 29AA, thickness of 190 µm, 88% porosity) and carbon cloth (AvCarb, HCB1071, thickness of 356 µm, 50% porosity) were used as the base electrode materials. Henceforth, carbon paper electrodes are referred to as 29AA, and carbon cloth as HCB1071. These were modified with carbon black (CB, Cabot Corp., PBX51, 1300-1550 m 2 g −1 ) or graphene nanoplatelets (GnP, Sigma-Aldrich, 750 m 2 g −1 ), using a simple drop-casting method as follows. Firstly, a 10 mL dispersion containing a 9:1 mass ratio of the carbon black or GnP active material to Nafion (Alfa Aesar, D520 5% w/w) was prepared in a 1:1 volume ratio of de-ionised (DI) water to isopropyl alcohol (IPA, Sigma-Aldrich, 99.9%). This was then sonicated for 15 min before homogenisation for 5 min at 28,000 rpm, using a Cole-Parmer Labgen series 7 homogeniser. The resultant dispersion was then drop cast onto 2 cm × 2 cm segments of carbon paper or carbon cloth in 8 µL aliquots to achieve a loading of 0.3 mg cm −2 active material and dried for 24 h at ambient temperature. Figure 1 provides a schematic overview of this process. Hereafter, carbon-black-and graphene-nanoplatelet-modified electrodes are denoted as CB and GnP, respectively. To prepare the electrode materials for experimentation, 5 mm × 5 mm segments were cut and then mounted onto a graphite/PVDF composite current collector (Eisenhuth, BMA5), using PTFE tape (3 M, T6013, 0.1 mm thickness), exposing an active electrode area of 0.1 cm 2 . tion/reduction cycling. Linear sweep voltammetry (LSV) was employed to determine kinetic parameters of the bromine electrodes, and the effect of different electrolyte concentrations on their comparative performance was investigated.

Electrode Materials and Preparation
Carbon paper (SGL Carbon, Sigracet 29AA, thickness of 190 µm, 88% porosity) and carbon cloth (AvCarb, HCB1071, thickness of 356 µm, 50% porosity) were used as the base electrode materials. Henceforth, carbon paper electrodes are referred to as 29AA, and carbon cloth as HCB1071. These were modified with carbon black (CB, Cabot Corp., PBX51, 1300-1550 m 2 g −1 ) or graphene nanoplatelets (GnP, Sigma-Aldrich, 750 m 2 g −1 ), using a simple drop-casting method as follows. Firstly, a 10 mL dispersion containing a 9:1 mass ratio of the carbon black or GnP active material to Nafion (Alfa Aesar, D520 5% w/w) was prepared in a 1:1 volume ratio of de-ionised (DI) water to isopropyl alcohol (IPA, Sigma-Aldrich, 99.9%). This was then sonicated for 15 min before homogenisation for 5 min at 28,000 rpm, using a Cole-Parmer Labgen series 7 homogeniser. The resultant dispersion was then drop cast onto 2 cm × 2 cm segments of carbon paper or carbon cloth in 8 µL aliquots to achieve a loading of 0.3 mg cm −2 active material and dried for 24 h at ambient temperature. Figure 1 provides a schematic overview of this process. Hereafter, carbonblack-and graphene-nanoplatelet-modified electrodes are denoted as CB and GnP, respectively. To prepare the electrode materials for experimentation, 5 mm × 5 mm segments were cut and then mounted onto a graphite/PVDF composite current collector (Eisenhuth, BMA5), using PTFE tape (3 M, T6013, 0.1 mm thickness), exposing an active electrode area of 0.1 cm 2 .

Electrochemical Methods
All electrochemical experimentation was conducted by using a Bio-Logic SP-300 potentiostat and EC-Lab software. A three-electrode half-cell setup was used, employing a 100 mL glass cell (Pine Research, Durham, NC, USA, AKCELL1). An Ag/AgCl reference electrode in 4 M KCl with AgCl solution (Pine Research, RREF0021), graphite rod counter electrode (Alfa Aesar, 10mm dia., 99.997%), and a working electrode that was prepared as described in Section 2.2 were utilised.
Linear sweep voltammetry (LSV) was conducted at a potential scan rate of 1 mV s −1 to overpotentials, ƞ, of ±400 mV vs. open circuit potential (OCP). Polarisation curves were obtained by galvanodynamic current control at 1 mA s −1 to potentials of ±400 mV vs. OCP.

Electrochemical Methods
All electrochemical experimentation was conducted by using a Bio-Logic SP-300 potentiostat and EC-Lab software. A three-electrode half-cell setup was used, employing a 100 mL glass cell (Pine Research, Durham, NC, USA, AKCELL1). An Ag/AgCl reference electrode in 4 M KCl with AgCl solution (Pine Research, RREF0021), graphite rod counter electrode (Alfa Aesar, 10 mm dia., 99.997%), and a working electrode that was prepared as described in Section 2.2 were utilised.
Linear sweep voltammetry (LSV) was conducted at a potential scan rate of 1 mV s −1 to overpotentials, η, of ±400 mV vs. open circuit potential (OCP). Polarisation curves were obtained by galvanodynamic current control at 1 mA s −1 to potentials of ±400 mV vs. OCP. Galvanostatic oxidation/reduction cycling was conducted at current densities between ±500 mA cm −2 and ±1.5 A cm −2 for periods of 10 min for both oxidation and reduction. During polarisation and cycling tests, the electrolyte was stirred at 1500 rpm, using a Camlab MS-H280-Pro magnetic stirrer and a PTFE stir bar in order to minimise diffusion limitations. To estimate the ECSA of working electrodes, cyclic voltammetry (CV) was carried out at eight potential sweep rates between 20 mV s −1 and 200 mV s −1 , with potential limits of ±50 mV vs. OCP. All experimentation was completed at a temperature of 298 K.

Material Characterisation
Surface morphologies of the electrode materials under investigation were characterised by using a FEI Quanta FEG 650 SEM, operated at an accelerating voltage of 20 kV and a working distance of 10 mm.

SEM Characterisation
The surface morphologies of the carbon paper (29AA) and carbon cloth (HCB1071) were examined by SEM in as-received condition and after the application of 0.3 mg cm −2 carbon black (CB) or graphene nanoplatelets (GnP). As seen in Figure 2a-c, the as-received 29AA consists of carbon fibres that are around 7 µm in diameter, with carbon filler impregnated in between, while the HCB1071 consists solely of carbon fibres, again around 7 µm in diameter (Figure 3c), arranged in a plain weave structure, with warp and weave thicknesses between 360 µm and 500 µm ( Figure 3a). Galvanostatic oxidation/reduction cycling was conducted at current densities between ±500 mA cm −2 and ±1.5 A cm −2 for periods of 10 min for both oxidation and reduction. During polarisation and cycling tests, the electrolyte was stirred at 1500 rpm, using a Camlab MS-H280-Pro magnetic stirrer and a PTFE stir bar in order to minimise diffusion limitations. To estimate the ECSA of working electrodes, cyclic voltammetry (CV) was carried out at eight potential sweep rates between 20 mV s −1 and 200 mV s −1 , with potential limits of ±50 mV vs. OCP. All experimentation was completed at a temperature of 298 K.

Material Characterisation
Surface morphologies of the electrode materials under investigation were characterised by using a FEI Quanta FEG 650 SEM, operated at an accelerating voltage of 20 kV and a working distance of 10 mm.

SEM Characterisation
The surface morphologies of the carbon paper (29AA) and carbon cloth (HCB1071) were examined by SEM in as-received condition and after the application of 0.3 mg cm −2 carbon black (CB) or graphene nanoplatelets (GnP). As seen in Figure 2a-c, the as-received 29AA consists of carbon fibres that are around 7 µm in diameter, with carbon filler impregnated in between, while the HCB1071 consists solely of carbon fibres, again around 7 µm in diameter (Figure 3c), arranged in a plain weave structure, with warp and weave thicknesses between 360 µm and 500 µm ( Figure 3a).  TEM analysis previously reported by Youssry et al. [27]. The micrometre-scale structures observed here by SEM must therefore be due to the coalescence of the nanoscale carbon black particles during the drop casting/drying process. When modified with 0.3 mg cm −2 GnP, boulder-type structures are again observed on the carbon fibres (Figures 2f,g and 3f,g), but these tend to be larger than for carbon black, i.e., in excess of 5 µm in some cases (Figures 2g and 3g). This is likely due to the coalescence of the larger-diameter GnP particles, <2 µm [28], compared to carbon black.

Electrochemically Active Surface Area
Estimates of ECSA were made by determining double-layer capacitances, Cdl, using CV at eight potential scan rates between 20 mV s −1 and 200 mV s −1 in a potential region of ±50 mV vs. open circuit potential (OCP) in a 1M KCl electrolyte where no faradaic current occurs. Anodic and cathodic charging currents are taken at 0 V vs. OCP and are plotted against scan rate, ν, to ascertain double-layer capacitance, Cdl, using the relationship shown in Equation (4) [29]: With the addition of 0.3 mg cm −2 carbon black (CB), additional material is observed on both carbon paper (29AA) (Figure 2d,e) and carbon cloth (HCB1071) (Figure 3d,e), with boulder-type structures adhered to the carbon fibres up to 2 µm in diameter (Figures 2e and 3e). PBX51 carbon black has an average particle size of 20 nm, as confirmed by the TEM analysis previously reported by Youssry et al. [27]. The micrometrescale structures observed here by SEM must therefore be due to the coalescence of the nanoscale carbon black particles during the drop casting/drying process. When modified with 0.3 mg cm −2 GnP, boulder-type structures are again observed on the carbon fibres (Figures 2f,g and 3f,g), but these tend to be larger than for carbon black, i.e., in excess of 5 µm in some cases (Figures 2g and 3g). This is likely due to the coalescence of the larger-diameter GnP particles, <2 µm [28], compared to carbon black.

Electrochemically Active Surface Area
Estimates of ECSA were made by determining double-layer capacitances, C dl , using CV at eight potential scan rates between 20 mV s −1 and 200 mV s −1 in a potential region of ±50 mV vs. open circuit potential (OCP) in a 1 M KCl electrolyte where no faradaic current occurs. Anodic and cathodic charging currents are taken at 0 V vs. OCP and are plotted against scan rate, ν, to ascertain double-layer capacitance, C dl , using the relationship shown in Equation (4) [29]: The ECSA is then found by Equation (5), where C dl is the double-layer capacitance of the material, and C s is the specific capacitance of the electrode material [29,30]: In order to establish an estimated value for C s , planar BMA5 graphite/PVDF plate (denoted as BMA5) is used as a reference, using Equation (6), where A is the geometric area of the working electrode (0.1 cm 2 ): Figure 4a provides the cyclic voltammograms obtained on a BMA5 graphite/PVDF electrode, with plots of i vs. ν shown in Figure 4b. As there are small variations in the double-layer capacitances obtained by using anodic and cathodic charging currents, the median value is taken to define C s , which becomes 29.4 µF cm −2 .
The ECSA is then found by Equation (5), where Cdl is the double-layer capacitance of the material, and Cs is the specific capacitance of the electrode material [29,30]: In order to establish an estimated value for Cs, planar BMA5 graphite/PVDF plate (denoted as BMA5) is used as a reference, using Equation (6), where A is the geometric area of the working electrode (0.1 cm 2 ): Figure 4a provides the cyclic voltammograms obtained on a BMA5 graphite/PVDF electrode, with plots of i vs. ν shown in Figure 4b. As there are small variations in the double-layer capacitances obtained by using anodic and cathodic charging currents, the median value is taken to define Cs, which becomes 29.4 µF cm −2 . Cyclic voltammograms at varying scan rates for all the electrode materials under investigation are provided in Figure 5, with the subsequent plots of i vs. ν displayed in Figure 4c,d. Table 1 shows the Cdl, roughness factor (RF), ECSA, and specific surface area for all electrodes. The RF is calculated by dividing the ECSA of the electrode under investigation by the ECSA of the planar BMA5 electrode. Specific surface areas were calculated by using the experimentally determined ECSA and mass of the electrodes. The coefficient of determination (R 2 ) values are provided as an indication of the linearity of i vs. ν data in Figure 4. In all cases, R 2 is at least 0.998, confirming very good linearity of the data. Cyclic voltammograms at varying scan rates for all the electrode materials under investigation are provided in Figure 5, with the subsequent plots of i vs. ν displayed in Figure 4c,d. Table 1 shows the C dl , roughness factor (RF), ECSA, and specific surface area for all electrodes. The RF is calculated by dividing the ECSA of the electrode under investigation by the ECSA of the planar BMA5 electrode. Specific surface areas were calculated by using the experimentally determined ECSA and mass of the electrodes. The coefficient of determination (R 2 ) values are provided as an indication of the linearity of i vs. ν data in Figure 4. In all cases, R 2 is at least 0.998, confirming very good linearity of the data.  In as-received condition, 29AA and HCB1071 display very similar ECSA values of around 0.28 cm 2 . HCB1071 possesses a slightly larger ECSA of 0.282 cm 2 compared to 0.276 cm 2 for 29AA (Table 1). This trend continues with the addition of either CB or GnP to the electrodes. GnP-modified electrodes have ECSA values of 2.36 cm 2 in the case of HCB1071,  In as-received condition, 29AA and HCB1071 display very similar ECSA values of around 0.28 cm 2 . HCB1071 possesses a slightly larger ECSA of 0.282 cm 2 compared to 0.276 cm 2 for 29AA (Table 1). This trend continues with the addition of either CB or GnP to the electrodes. GnP-modified electrodes have ECSA values of 2.36 cm 2 in the case of HCB1071, and 2.16 cm 2 for 29AA, increased by a factor of 8 over the as-received materials. CB-modified electrodes display the largest ECSA values, i.e., 3.11 cm 2 and 2.86 cm 2 , for HCB1071 and 29AA, respectively, representing increases by a factor of 11 for HCB1071 and a factor of 10 for 29AA, compared to the as-received materials. The observed increases in ECSA are concomitant with the stated surface areas of PBX51 carbon black (1300-1550 m 2 g −1 [31]) and GnPs (750 m 2 g −1 [28]).

Electrochemical Performance
Polarisation curves for the electrode materials under investigation are displayed in Figure 6a,b. In all cases, the 29AA and HCB1071 electrodes provide quite a similar performance, with the latter displaying marginally lower overpotentials for bromide oxidation, (η oxi ) and reduction of bromine (η red ) at current densities (j) of ±500 mA cm −2 and larger oxidation (j oxi ) and reduction (j red ) current densities at overpotentials (η) of ±400 mV ( Table 2). There is a clear improvement in performance with the addition of either CB or GnP to the electrodes. CB-modified electrodes display n oxi and n red values 31-39% lower at j = ±500 mA cm −2 than on unmodified materials, with j oxi and j red values 39-49% larger at η = ±400 mV ( Table 2). The lowest overpotential and the largest current densities are achieved on the GnP-modified electrodes, with n oxi and n red values 39-47% lower at j = ±500 mA cm −2 than on unmodified materials, and j oxi and j red values 69-87% larger at η = ±400 mV. Given the larger ECSA of CB-modified electrodes discussed in Section 3.2.1, it is surprising that GnP modification provides the better performance during electrochemical testing. Therefore, ECSA-adjusted LSV was employed to ascertain the charge transfer resistance, Rct; exchange current density, i0; and rate constant, k0, of the bromine electrode reaction on the electrode materials under investigation. The ECSA-adjusted voltammograms on CB-and GnP-modified 29AA and HCB1071 are shown in Figure 7. The potential responses during oxidation/reduction cycling of bromine electrodes are provided in Figure 6c,d, with the ratios between mean reduction potential, E red , and mean oxidation potential, E oxi , given in Table 2 as an indicator of voltaic efficiency in a full cell environment. The results of oxidation/reduction cycling follow the same trend as the polarisation data, with similar a performance displayed by the 29AA and HCB1071 electrodes in all cases. The addition of carbon black provides 46% and 48% increases in  Given the larger ECSA of CB-modified electrodes discussed in Section 3.2.1, it is surprising that GnP modification provides the better performance during electrochemical testing. Therefore, ECSA-adjusted LSV was employed to ascertain the charge transfer resistance, R ct ; exchange current density, i 0 ; and rate constant, k 0 , of the bromine electrode reaction on the electrode materials under investigation. The ECSA-adjusted voltammograms on CB-and GnP-modified 29AA and HCB1071 are shown in Figure 7. Given the larger ECSA of CB-modified electrodes discussed in Section 3.2.1, it is surprising that GnP modification provides the better performance during electrochemical testing. Therefore, ECSA-adjusted LSV was employed to ascertain the charge transfer resistance, Rct; exchange current density, i0; and rate constant, k0, of the bromine electrode reaction on the electrode materials under investigation. The ECSA-adjusted voltammograms on CB-and GnP-modified 29AA and HCB1071 are shown in Figure 7. At low overpotentials, i.e., <25 mV, the relationship between overpotential and current density can be approximated to be linear, and Rct can be calculated by using Equation At low overpotentials, i.e., <25 mV, the relationship between overpotential and current density can be approximated to be linear, and R ct can be calculated by using Equation (7), where j ECSA is the electrochemically active surface area adjusted current density. Subsequently, i 0 and k 0 may be found by using Equations (8) and (9), respectively, where R is the universal gas constant in J K −1 mol −1 , T is the temperature in Kelvin, n is the number of electrons involved in the reaction, F is Faraday's constant in A s mol −1 , and C 0 is the electrolyte concentration in mol L −1 [32][33][34]: i 0 = RT nFR ct (8) The kinetic parameters resulting from these equations for the enhanced-surface-area electrodes under investigation are shown in Table 3. As can be seen, GnP-modified electrodes have more favourable kinetic parameters towards the bromine electrode reaction, with 34-35% lower R ct and 52-54% larger i 0 and k 0 values, compared to the CB-modified electrodes. The improved electrochemical performance of GnP electrodes over their CBmodified counterparts during polarisation and cycling experimentation ( Figure 6) is then due to their superior kinetic activity for the bromine reaction, despite their inferior ECSA.

Effect of Electrolyte Concentration
The effect of varying electrolyte concentration on the comparative performance of CBand GnP-modified HCB1071 carbon cloth electrodes was studied by using polarisation and oxidation/reduction cycling of a bromine electrode reaction at current densities between 500 mA cm −2 and 1.5 A cm −2 . The resultant polarisation curves are provided in Figure 8a. These demonstrate enhanced current densities at η = ±400 mV as the electrolyte concentration increases (Table 4), due to the improved mass transport of bromide/bromine and the increasing conductivity of HBr solutions with concentration [35,36], with conductivity also being shown previously to improve with Br 2 concentrations up to 2 M in HBr solutions >3 M [37]. In the 3 M HBr/1 M Br2 electrolyte concentration, the GnP-modified electrode displays significantly larger current densities at ƞ = ±400 mV than the CB-modified electrode ( Table 4); this is in agreement with the results presented in Figure 6. However, as the electrolyte concentration is increased to 4.5 M HBr/1.5 M Br2, the difference between the electrodes diminishes; in fact, the CB electrode achieves slightly larger oxidation and reduc-  In the 3 M HBr/1 M Br 2 electrolyte concentration, the GnP-modified electrode displays significantly larger current densities at η = ±400 mV than the CB-modified electrode ( Table 4); this is in agreement with the results presented in Figure 6. However, as the electrolyte concentration is increased to 4.5 M HBr/1.5 M Br 2 , the difference between the electrodes diminishes; in fact, the CB electrode achieves slightly larger oxidation and reduction current densities of j oxi = 1454 mA cm −2 and j red = 1467 mA cm −2 compared to j oxi = 1429 mA cm −2 and j red = 1442 mA cm −2 on the GnP electrode. In the 6 M HBr/2 M Br 2 electrolyte concentration, this difference increases further, with the CB electrode displaying j oxi = 1575 mA cm −2 and j red = 1581 mA cm −2 compared to j oxi = 1539 mA cm −2 and j red = 1527 mA cm −2 on the GnP electrode (Table 4). It appears, therefore, that higher electrolyte concentrations favour the CB electrode over the GnP electrode. This trend is confirmed by the results of oxidation/reduction cycling provided in Figure 8b-d. At oxidation/reduction current densities of ±1.5 A cm −2 , for example, the GnP-modified electrode displays lower mean oxidation, n oxi , and reduction, n red , overpotentials than the CB electrode in 3 M HBr/1 M Br 2 . In 4.5 M HBr/1.5 M Br 2 , the mean overpotentials on the two electrodes are very similar, within 5 mV of each other, but the CB electrode outperforms the GnP electrode marginally (Table 4). In the 6 M HBr/2 M Br 2 electrolyte concentration, the CB electrode demonstrates a n oxi value that is 44 mV lower and an n red value that is 34 mV lower than that of the GnP electrode. This suggests that highly concentrated electrolytes favour the increased ECSA associated with CB-modified electrodes, while lower electrolyte concentrations favour the improved kinetic ability of GnP electrodes, although further work is required to elucidate this. Future work will also investigate the employment of CB-and GnP-modified electrodes in an H 2 /Br 2 flow cell environment, including long-term cycling to verify the performance and establish their stability in an H 2 /Br 2 flow battery.

Conclusions
Enhanced surface area bromine electrode materials were fabricated by the addition of carbon black (PBX51) and graphene nanoplatelets (GnP) to carbon paper (29AA) and carbon cloth (HCB1071) materials, using a simple drop-casting method. Electrochemically active surface areas (ECSAs) were estimated by obtaining double-layer capacitances, using cyclic voltammetry. GnP-modified electrodes were found to possess ECSAs eight times those of the unmodified materials, while carbon black electrodes displayed the largest ECSAs, between 10 and 11 times larger than unmodified electrodes. The kinetic ability of the modified electrodes towards the bromine reactions was assessed by using linear sweep voltammetry, with GnP electrodes demonstrating improved kinetic parameters over their carbon black counterparts.
The electrochemical performance was assessed by using polarisation and oxidation/reduction cycling tests, and the effect of electrolyte concentration on performance was investigated by using the same methods. The performance of carbon paper (29AA) and carbon cloth (HCB1071) electrodes was very similar in all cases. In 3 M HBr/1 M Br 2 electrolytes, GnP electrodes displayed superior performance. Polarisation curves on GnP electrodes demonstrated current densities at overpotentials of ±400 mV up to 87% larger than unmodified electrodes, compared to a maximum increase of 49% on carbon black electrodes. During oxidation/reduction cycling, carbon black (PBX51)-modified electrodes improved reduction/oxidation potential ratios increases up to 48% compared to unmodified materials, with increases of up to 63% for GnP electrodes.
Increasing the electrolyte concentration improved electrode performance in all cases. However, as the electrolyte concentration increased, carbon-black-modified electrodes outperformed GnP-modified electrodes. In the 6 M HBr/2 M Br 2 electrolyte concentration, the carbon-black-modified electrode displayed current densities approaching 1.6 A cm −2 at overpotentials of ±400 mV during polarisation tests, compared to ≈1.53 A cm −2 for GnP-modified electrodes. During oxidation/reduction cycling at 1.5 A cm −2 , carbon black electrodes demonstrated mean oxidation and reduction overpotentials that were 44 mV and 34 mV lower, respectively, compared to GnP electrodes. Lower-concentration electrolytes were shown to favour the superior kinetic ability of the GnP electrodes, while highly concentrated electrolytes favoured the larger ECSA of carbon black electrodes. The best performance achieved in this work was obtained by a carbon black (PBX51)modified carbon cloth (HCB1071) electrode in 6 M HBr/2 M Br 2 , with overpotentials of n oxi = 364 mV and n red = 343 mV at ±1.5 A cm −2 and current densities approaching 1.6 A cm −2 at η = ±400 mV.
Author Contributions: D.P.T., conceptualisation, data curation, formal analysis, investigation, methodology, validation, visualisation, writing-original draft, and writing-review and editing; X.L., conceptualisation, funding acquisition, project administration, supervision, validation, and writingreview and editing. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the European Union's Horizon 2020 research and innovation programme under the MELODY project, with grant agreement No. 875524, and the EPSRC Supergen Energy Storage Project (grant number: EP/P003494/1). The authors would also like to thank all MELODY project consortium members for valuable discussions and insights.