Chemical Stability of Graphite-Polypropylene Bipolar Plates for the Vanadium Redox Flow Battery at Resting State

Current collectors called bipolar plates (BPP) are important elements within the conversion unit of the vanadium redox ﬂow battery (VRFB). They are in direct contact with acidic electrolytes, containing vanadium species in different oxidation states. The inﬂuence of the state of charge (SOC) on the calendar aging of BPPs was examined. Graphite-polypropylene BPPs were immersed in positive and negative vanadium electrolytes at 0%, 20%, 80% and 100% SOC for 30, 90 and 190 days. H 2 gas evolution was observed as side reaction on the surface of the BPPs in the negative electrolyte. After electroless aging, scanning electron (SEM) and confocal microscopy measurements showed no signiﬁcant changes in the surface morphology. The electrical conductivities of the BPPs were not affected signiﬁcantly. However, contact angle ( θ ) measurements revealed that the positive electrolyte inﬂuenced the wettability of the BPPs. X-ray photoelectron (XP) spectroscopy showed progressing oxidation of the BPP surfaces in the positive electrolyte and adsorption or entrapment of vanadium ions in the pores at high SOC. Cyclic voltammograms (CV) provided evidence that the graphite was oxidized combined with an increase in effective surface area. ATR-FTIR measurements showed slight oxidation of pure polypropylene granulate in the positive electrolyte with 100% SOC. © The increasing energy power supply from intermittent renewable energy sources requires a rapid introduction of efﬁcient energy stor-ages. One promising technology is the vanadium redox ﬂow battery (VRFB), as it enables to scale the power and the storage capacity independently according to speciﬁc requirements. 1,2 In addition the VRFB is characterized by a fast response time and long electrolyte cycle life. 3 Each reaction unit in a VRFB stack is composed of two half-cells separated by a membrane consisting of an electrode in con- tact with a current collector called bipolar plate (BPP). 2 In a battery stack the BPPs are “non-active” components that conduct current from one cell to the other. They physically separate adjacent cells from each other while staying in contact with acidic half-cell electrolytes con- taining vanadium species in different oxidation states on each side. 4

The increasing energy power supply from intermittent renewable energy sources requires a rapid introduction of efficient energy storages. One promising technology is the vanadium redox flow battery (VRFB), as it enables to scale the power and the storage capacity independently according to specific requirements. 1,2 In addition the VRFB is characterized by a fast response time and long electrolyte cycle life. 3 Each reaction unit in a VRFB stack is composed of two half-cells separated by a membrane consisting of an electrode in contact with a current collector called bipolar plate (BPP). 2 In a battery stack the BPPs are "non-active" components that conduct current from one cell to the other. They physically separate adjacent cells from each other while staying in contact with acidic half-cell electrolytes containing vanadium species in different oxidation states on each side. 4 For brevity we call the V 2+ /V 3+ solution "negative electrolyte" and the VO 2+ /VO 2 + solution "positive electrolyte". While vanadium redox reactions mainly occur on the surfaces of the porous electrodes, they could also occur unintentionally on the surfaces of the BPPs. [5][6][7] Therefore, a good BPP should be characterized by a high chemical and mechanical stability, high electrical conductivity and impermeability to preclude leakage. 8 Metallic BPPs are usually not used in VRFB as they corrode in acidic environments and would need a protective layer. 5,9,10 Therefore, graphite based BPPs are commonly used as they possess a good electrical conductivity and a better chemical stability. 4 However, pure graphite plates are not favored as BPPs due to their high weight and cost as well as low mechanical strength and poor processability. 2,9,11,12 Instead, the BPPs in the VRFB usually consist of graphite-polymer composites, as they are less cost-intensive and possess higher mechanical stability than graphite. 13 In such composite materials, the polymer provides the mechanical stability while the carbon component, e.g. graphite, carbon black, carbon fibre and/or carbon nanotubes, provides the electrical conductivity. 5,12 Different composite materials were already investigated for the production of BPPs in the VRFB, including diverse graphite and carbon materials with polyethylene, 12,14-16 polypropylene (elastomer), 6,7 fluoropolymer, 4 polyphenylene sulfide, 5 phenol formaldehyde resin, 17 as well as epoxy resin. 9,10,[18][19][20][21] Looking only at the single components, polymers like polyethylene, polypropylene and polytetrafluoroethylene exhibit high chemical resistance toward concentrated sulfuric * Electrochemical Society Member. z E-mail: Carolina.Nunes-Kirchner@next-energy.de acid at ambient temperature. 22 However, carbon paper and graphite electrodes/current collectors showed evidence of corrosion, i.e. an increased content of oxygen-containing functional groups and degradation effects due to CO 2 evolution, after treatment in the positive halfcell electrolyte of a VRFB. 7,[23][24][25] Up to date there are several studies dealing with cyclic aging of BPPs for the VRFB. [4][5][6][7]9,10,14,16,18,20,21,26 For example Liu et al. showed that carbon-polyethylene composite BPPs were aging when an anodic polarization potential of 2.26 V vs. standard hydrogen electrode (SHE) was applied for 6 h in 2 M VOSO 4 in 2 M H 2 SO 4 . Their results indicate that new oxygen functional groups (C=O and O=C-OH) were formed on the surface of the BPP, delamination processes commenced and material was lost by corrosion. 16 Moreover, high overpotentials on the BPPs can lead to side reactions of the electrolyte such as evolution of H 2 gas in the negative half-cell and CO 2 , CO, and O 2 in the positive half-cell. 6,16,25 Electrochemical oxidation is caused by potential differences in the cells of a VRFB leading to damage of BPPs. 4 Corrosion may cause an increase in the porosity 7,16,21 leading to ion permeability, brittleness, loss in mechanical stability and a decreased electrical conductivity. It was reported that the shunt current might play the main role in this process whereas the main current is less dominant. 10 9 observed a weight loss in the graphite-epoxy BPP after treating it in vanadium electrolyte with oxidation state +5 at 80 • C for 7 days. It is evident that both treatments used acceleration of possible aging by means of higher temperatures than occurring in real VRFB. No long-term study of the effect of positive and negative vanadium electrolyte in different SOCs on a graphite-polypropylene BPP has been demonstrated up to date. This analysis is of a great importance if the battery shall be operated with intermittent renewable energy sources. The VRFB can remain idle at certain SOC until it operates again to charge or discharge. Thus, the resting state of the battery at a certain SOC might strongly influence the stability of the BPP. Therefore, the main focus of this study was to investigate the stability of BPPs during the resting state of a VRFB. For this purpose, BPP samples were submitted to calendar aging in the positive and negative electrolytes simulating the resting state of the VRFB at different SOC. The immersion time was varied and the BPP were electrically, chemically, morphologically and electrochemically analyzed before and after aging.

Experimental
Sample pretreatment.-The experiments were performed on injection moulded and sandblasted BPPs composed of 86% graphite and 14% polypropylene (PPG86, Eisenhuth GmbH & Co. KG, Germany). The BPPs had a thickness of 1.5 mm and were shaped to a disk of 16 mm diameter. All BPP samples were cleaned by successive immersion in deionized water, ethanol, acetone and deionized water for 15 min each. Then they were dried in a vacuum oven (Memmert, VO200, Germany) for 21 h at 60 • C. After cleaning the samples were characterized before they were exposed to the different electrolytes.
Electrolyte preparation.-The vanadium electrolyte was supplied by GfE (GfE Metalle und Materialien GmbH, Germany). It consists of 1.6 mol/L V n+ with a ratio of 50:50 V 3+ and VO 2+ , 2.0 mol/L H 2 SO 4 and 0.05 mol/L H 3 PO 4 . In this work both half-cell electrolytes were prepared in different SOCs in a conventional VRF cell setup at 40 mA/cm 2 using a potentiostat/galvanostat (Solartron Analytical Modulab Pstat potentiostat/galvanostat, UK). The negative and the positive electrolytes were simultaneously adjusted to SOCs of approximately 0%, 20%, 80% or 100%. The SOCs of the electrolytes were verified using a UV/Vis spectrophotometer (LAMBDA XLS+, PerkinElmer, USA). The spectra of the negative electrolyte were measured during the cell operation and the concentrations of V 2+ and V 3+ species were calculated using the absorbance values at 605 nm and 850 nm as described in the literature. 28,29 Due to the high absorbance and the nonlinear relationship to the concentration of VO 2 + , 30,31 the SOC values for the positive electrolyte were estimated based on the UV/Vis measurements of the negative electrolyte.
Aging of BPP.-The treatment of BPPs in the vanadium electrolytes was carried out in 250 mL Schlenk flasks. The Schlenk flasks were filled with 65 mL of the positive or negative electrolyte with different SOC (0%, 20%, 80% or 100%) and subsequently flushed with Argon. 30 BPPs were immersed in each flask and an overpressure of Argon was applied before they were closed tightly. A PTFE-sealing ring (Glindemann sealing ring, Germany) was used for tightening the glass stopper in order to prevent oxidation of the electrolyte by oxygen from air. The flasks were stored at room temperature (∼22 • C) and protected from light. For the investigation, ten BPPs were removed after a period of 30, 90 and 190 days of immersion. Then they were cleaned successively with deionized water, ethanol, acetone and deionized water for 15 min each and dried in a vacuum oven for 21 h at 60 • C.
Aging of polypropylene.-Pure granulated polypropylene for manufacturing of the BPPs was immersed separately in the positive electrolyte at 100% SOC for 120 days. After aging it was washed with deionized water.
Gas chromatography.-The gas composition in the Schlenk flask filled with the negative electrolyte was analyzed by gas chromatography (GC) using a GC-2014 (Shimadzu, Germany) with a micropacked column (ShinCarbon ST 100/120, Germany) and Ar as carrier gas. A blind measurement proving oxygen-free conditions was performed after deaeration of the GC.

Scanning electron microscopy (SEM).-Morphological charac-
terization of the BPP surfaces was carried out by a scanning electron microscopy (SEM, NEON 40, Zeiss, Germany) with 30 kV acceleration voltage and a SE2 detector.
Confocal microscopy.-White light confocal microscopy (Sensofar, PLu neox, Spain) measurements were performed to study the surface roughness of the BPPs. An area of 127.32×95.45 μm 2 with 100 x magnification was scanned with the software SensoScan. The z-scan was performed using a 0.20 μm step size. The root mean square surface roughness (Sq) was calculated with the same software after leveling the confocal microscopy image. In addition the developed interfacial area ratio (Sdr) was calculated with the software Gwyddion (Version 2.45). Sdr expresses the percentage of additional surface area introduced by the texture of the surface in comparison to a geometrical flat area. Ten different positions were measured, analyzed and averaged for each BPP.
Water contact angle measurement.-The static water contact angle (θ) was measured by the sessile drop method using 4 μL drops of deionized water with the contact angle system OCA 15plus (Data-Physics, Germany) equipped with a Hamilton 500 μL syringe, a CCD camera and the SCA20 software (version 2.0.0). The quoted values were obtained by calculating the average of six measurements for each investigated BPP.
X-ray photoelectron spectroscopy.-XP spectroscopy of the plates was performed with an ESCALAB 250 Xi (Thermo Fisher, UK) using a monochromated Al K α X-ray source, a pass energy of 10 eV, a dwell time of 50 ms, an energy step size of 0.02 eV and averaging of 10 scans. The XP spectra were analyzed by fitting the data using a mixed Gauss-Lorentz product function and a smart background function in the software Avantage (v.5.932, Thermo Fisher, UK).
Electrochemical measurements.-The electrochemical measurements were performed with a conventional three electrode set up inside a Faraday cage and connected to a potentiostat/galvanostat (Solartron Analytical Modulab UK). A sample holder (Metrohm Autolab B.V., Netherlands) was used to hold the BPP as working electrode by exposing a defined BPP surface area of 1 cm 2 to the solution. The Hg 2 SO 4 reference electrode was connected to the cell by a Luggin capillary filled with the electrolyte. All potentials were recalculated and quoted with respect to the SHE. The auxiliary electrode was a heat-treated graphite felt (400 • C in air for 18 h, GFD5, SGL Carbon Group, Germany) attached to a platinum wire. First linear sweep voltammetry (LSV) was carried out from open circuit potential (OCP) to 0 V at 20 mV/s at room temperature (∼22 • C) in 0.1 M H 2 SO 4 (>95%, 1.830 g/mL, Fisher Scientific, UK) and subsequently CVs were performed between 0 V and 1.05 V with 5 mV/s. Prior to the measurements the electrolyte was flushed with N 2 for 15 min and during the measurements the N 2 atmosphere was maintained. The double layer capacitance (C dl ) was estimated by taking into account the current values at approximately 0.4 V from the first CV, where no faradaic or pseudo-capacitive contributions were detectable.
Fourier transform infrared spectroscopy.-Chemical modifications of untreated and aged polypropylene were studied using an attenuated total reflection (ATR) equipment with a Fourier transform infrared spectrometer (FTIR, Spectrum 100, Perkin Elmer, USA). Three different samples of the treated and untreated polypropylene were scanned from 4000 to 650 cm −1 .
Electrical conductivity measurement.-The electrical resistivity measurements were performed using a four point probe equipment with a test unit (RM3-AR, Jandel, UK). Six measurements were performed at different locations on the surface of the BPPs and the specific electrical conductivity was calculated.

Results and Discussion
The transparent Schlenk flasks, in which the aging was carried out, were visually examined in regular time intervals. No visual changes were observed in the positive electrolyte. However, gas evolution was visible on the surface of the BPPs immersed in the negative electrolyte ( Figure 1). According to our observations the amount and size of gas bubbles was increasing with the SOC. A GC analysis of the head space in the Schlenk flask with 100% SOC revealed that the gas was hydrogen (H 2 ) with a retention time of around 1.5 min. A similar gas evolution effect had already been described in the negative vanadium electrolyte on carbon paper electrodes. 32 Therefore, UV/Vis measurements of the electrolyte were periodically recorded during the long-term experiment. The UV/Vis spectra of the negative electrolyte before and after 115 days of immersion of several BPPs are shown in Figure 2. Based on these measurements a decrease of about 4% in the SOC was estimated within 115 days. It can be suggested, that H 2 evolution occurred as a consequence of the oxidation of the V 2+ species according to the following reactions:  Thus it can be considered that an electroless H 2 evolution took place on the electrically conductive surface of the BPPs causing self-discharge of the electrolyte. A time-lapse movie generated from bottom view pictures of the Schlenk flask taken every 40 min (not shown) provided evidence that the gas evolution mainly took place in the vicinity of edges, scratches and irregularities on the BPP surfaces. Therefore, it is recommended to handle the BPP carefully while assembling the VRFB and to prevent them from defects. The gas evolution must not be underestimated during resting conditions. It might lead to an overpressure in the system and cause capacity losses in the real VRFB. 33 However, it was reported that this type of self-discharge happens generally in the reaction unit and is hardly affecting the separate electrolyte tanks where the electrolyte can be stored for a long time with only negligible self-discharge effects. 2 Chen et al. 34 showed that H 2 evolution caused an increase in porosity of graphite electrodes after applying negative polarization potentials in vanadium electrolyte containing V 3+ ions. It was assumed that the H 2 evolution in our study could also cause changes in the surface of the BPPs. SEM images (Figure 3) show that the surfaces of the BPP samples are a composite of irregularly shaped constituents. It consists of different domains including flat or elevated graphite plates and polypropylene material. No obvious changes could be detected by SEM between the BPPs treated for 190 days in the negative or positive electrolyte at 100% SOC compared to the pristine one. In addition, the averaged Sq obtained from confocal microscopy measurements were (4.09 ± 0.60) μm, (4.06 ± 0.76) μm and (3.98 ± 0.80) μm for the pristine BPP and the treated BPPs in the negative and positive electrolytes, respectively. The observed high roughness values are in good agreement with the SEM images and confirm the composite complex structure of the BPPs. The Sdr value of (296 ± 46) % for the pristine BPP shows that the surface area is in fact three times higher compared to the geometrical surface area in agreement with the high roughness values. BPPs treated in the negative and positive electrolytes show similar Sdr values of correspondingly (276 ± 63) % and (270 ± 74) %. The performed quantitative analysis of the confocal images indicates that no significant change in the surface texture occurred on the microscopic level during the calendar aging. Apparently the H 2 evolution in the negative half-cell electrolyte did not have a significant effect on the topology of the BPPs. However changes on the nm scale might be unverifiable due to the heterogeneity of the BPP surface. Figure 4 shows θ values of BPP samples before and after treatment. The untreated BPP has a θ value of 127 • ± 4 • , indicating a hydrophobic surface. In comparison, θ of pure polypropylene and graphite are ∼108 • and 75-95 • , respectively. 35,36 It can be noticed that the composite material has a lower wettability than the pure materials obtained from literature which might be caused due to different surface structures. The BPPs which were immersed in the negative electrolytes showed only slightly lower θ values in comparison to the untreated BPP, indicating that the surfaces remained hydrophobic after treatment. In contrast, the BPPs immersed in the positive electrolytes showed a decrease of θ with increasing SOC and immersion durations till 90 days indicating that the surface wettability and hydrophilicity was increased. However, BPPs immersed in the positive electrolyte with 80% and 100% SOC became significantly more hydrophobic at immersion times of 190 days. Moreover, a higher standard deviation of θ is apparent for the BPPs treated in the positive electrolyte with 100% SOC suggesting an increase of the inhomogeneity of the surface as a consequence of the aging. θ and thus the wettability depends on several factors, such as the chemical composition, the heterogeneity, the particle sizes and shapes, the roughness of the surface as well as on the porosity. 37,38 It should be kept in mind that the studied BPPs are composite materials consisting of two hydrophobic materials (graphite and polypropylene). The observed changes in the contact angle could be caused by water drop interaction with both materials. As confocal microscopy revealed that the surface roughness did not change significantly in the μm scale, we assume that the roughness is not the main factor for the change in wettability and the changes of the surface properties might be rather dynamic than monotonous.
In order to unravel the effects caused by the calendar aging on the wetting of the BPPs, XP spectroscopy was performed on plates aged in 100% SOC electrolytes and compared with the pristine BPP. The XP spectroscopy investigations refer to the whole composite material, i.e. graphite and polypropylene. The focus was laid on the O 1s and V 2p spectra ( Figure 5). Two O 1s signals at 533 eV and 531 eV were observed for all samples. The signal at 533 eV is attributed to C-O while 531 eV indicates C=O groups. 11,[39][40][41][42] The spectra of the pristine BPP and the BPP treated in the negative electrolyte exhibited C-O groups and only a small amount of C=O groups. The appearance of the oxygen functionalities in the pristine BPP might derive from ether, hydroxyl and keto groups introduced in the graphite structure [43][44][45] during the production of the raw material and/or the BPP. Moreover it has already been reported that oxygen functionalities might also be incorporated into the polymer matrix of polypropylene at ambient conditions. 46 The surface of the BPP treated in the positive electrolyte showed an increase in C=O functional groups. It is noticeable that the ratio between C=O and C-O groups was increasing with the storage time in the positive electrolyte indicating progressing oxidation of the BPP surface. It was already shown for electrode materials such as carbon paper, that a high SOC in the positive electrolyte caused oxidation of the electrode surface resulting in a changed composition of oxygen functional groups. 24 The C 1s spectra are not shown in this study because the changes were less obvious due to the high signals from graphitic carbon (see also 24 ). No traces of sulfur were identified in all spectra using XP spectroscopy indicating that no intercalation of sulfur took place. However the spectra (c) and (d) showed additional signals in the V 2p region with a binding energy of 517 eV which are attributed to traces of vanadium in the oxidation state +5. Only the V 2p 3/2 peak is visible in the spectra because the V 2p 1/2 peak (spin orbit splitting of 7.5 eV) is broader and cannot be separated from other background features and noise in the region around 523-526 eV. The signal with a binding energy of 530 eV can be assigned to a bond between oxygen and vanadium. [47][48][49] As the changes observed in the oxygen functional groups cannot be distinguished between the graphite and polypropylene by means of XPS, supporting analysis was conducted for the single components with additional methods.
In order to further investigate the effects of the calendar aging, CVs were performed in sulfuric acid electrolyte. Electrochemical measurements provided information only about the electrically conductive graphite area of the BPP. For comparison, the electrochemical behavior of the pristine BPP was also examined. The initial cycle of the untreated BPP in the forward potential sweep showed redox  activity in the potential range from 0.3 to 0.7 V and subsequent oxidation reaction at potentials more positive than 0.8 V ( Figure 6). However, a rather noisy and atypical behavior was observed reproducibly during the backward sweep at potentials more negative than 0.5 V for all pristine BPPs. The appearance of positive currents during negative potential sweeps was often reported on platinum during oxidation of methanol in acidic media. It can be assigned to oxidation of chemisorbed methanol after de-poisoning of oxidized platinum in the backwards potential sweep. 50,51 The same behavior observed in the initial scan for the pristine BPP could be related to removal of impurities from the surface. In addition, the hydrophobic nature of the BPP surface could aggravate the electrode-electrolyte interface. These effects were observed only during the initial scans and they diminished with subsequent cycling. After 100 scans pseudo-capacitive behavior in the potential range between 0.5-0.6 V was visible in good agreement with previously reported measurements on graphitic substrates. According to the literature, the pseudo-capacitance is attributed to the quinone/hydroquinone functional groups on graphite. 52,53 Most probably the graphitic surface was oxidized during the potential sweeping and the introduced surface groups increased the hydrophilicity of the BPP surface. The water θ measured on the plate after 100 scans (117.0 ± 1.2) • was significantly lower compared to the pristine surface. Hence, for the aged samples, only the initial cycles were compared and the results after 30 and 190 days of immersion are provided (Figures 7 and 8). The quinone/hydroquinone reaction was observed for all aged samples. However, the redox reaction in the potential range between 0.5-0.6 V was more pronounced for the BPP treated in the positive electrolyte with high SOC (Figure 8b) indicating the increased number of C=O groups in good agreement with the XP spectroscopy results. Bearing in mind the oxidative nature of VO 2 + and sulfuric acid, an oxidation of the graphite compound of the BPPs was likely to occur during the calendar aging. Choo et al. 53 obtained comparable results when they investigated the effects of the electrochemical oxidation of highly oriented pyrolytic graphite in sulfuric acid, where the resulting pseudo-capacitance was increasing with oxidation of the substrate. During the oxidation process of graphite, different oxygen functional groups might be introduced in the graphite structure. As proposed by Hofmann, Ruess and Scholz-Boehm 43-45 ether, hydroxyl and keto groups might be present in oxidized graphite structures in good agreement with the quinone/hydroquinone reaction of the aged BPP in our study. It was shown, that incorporated oxygen groups in the graphite structure are enlarging the spacing between graphene layers. 43,54,55 In addition, carbon-oxygen bonds can cause partial change of the hybridization of carbon atoms from sp 2 to sp 3 which can also contribute to an increase in the interlayer distance. 43 Those effects might induce defects and distortion in the graphite structure which may allow sulfate and vanadium ions and water molecules to intercalate between graphene sheets via interlayer or direct acid penetration. [56][57][58] The penetration of oxidizing agents might again cause augmented oxidation of the graphite resulting in an increasing amount of ketone groups accompanied by hydroxyl groups. [58][59][60] Recently, Deheryan et al. 61 reported that the magnitude of the capacitive currents can be correlated with θ for carbon nanosheets. Thus it is expected that C dl measured for the BPPs is directly coupled to the specific surface area of graphite, which was in contact with the sulfuric acid during CV measurement. The BPPs aged in the negative electrolyte with 0% SOC showed low C dl values in the range of 50 μF/cm 2 and no change of C dl as a function of the immersion time (Figure 9a). In contrast the C dl of the BPP treated at SOC 100% rises slightly with the immersion time up to 75 μF/cm 2 (Figure 9a). This value is comparable to C dl estimated for the pristine sample after 100 CV scans in 0.1 M H 2 SO 4 ( Figure 6). Higher C dl values of approximately 110 μF/cm 2 were calculated for the BPPs aged in the positive electrolyte with 0% SOC which did not change with treatment time (Figure 9b). However, samples aged in the positive electrolyte with 100% SOC showed significantly higher C dl values which increased with the aging time up to 686 μF/cm 2 (Figure 9b). This behavior reveals that the graphite surface area wetted by the sulfuric acid is continuously increasing as a function of immersion time in the oxidative vanadium electrolyte. This observation is caused by electrochemical reactions on the graphite surface and is not related to an increase in the geometrical surface area because no changes in the morphology were measured by the confocal microscope within its resolution. Thus the higher penetration of water and ions into the graphite structure due to increasing incorporation of oxygen functionalities 56,57 could explain the observed rise in the C dl for the BPPs treated in the positive electrolyte with 100% SOC.
In addition to the oxidation of the graphite structure and the intercalation of water, the diffusion of VO 2 + -ions into the graphite galleries  was also possible during the calendar aging. 59 This assumption is supported by additional redox features which were visible in the CVs at approximately 0.9 V only for samples aged in the positive electrolyte with high SOC (Figure 8b). Since the reaction appeared in the potential range close to the standard potential of VO 2+ /VO 2 + redox couple, the additional redox activity could be assigned to the presence of adsorbed or trapped vanadium species that could not be removed upon washing. 62 These results are also in good agreement with the observed vanadium redox features in the XP spectra which became more apparent at longer exposure times, showing that oxidation reaction and widening of the graphite structure are increasing the penetration of water molecules and vanadium ions.
It would be expected, that the BPP treated in the positive electrolyte with SOC 100% should exhibit the highest hydrophilicity after 190 days of immersion. Contrary to this expectation, the θ values for the BPP treated in the positive electrolyte with high SOC show an opposite behavior. The θ value for the sample immersed for 190 days in 100% SOC is comparable to the pristine one, in contrast to the high capacitance difference. Therefore an additional aging mechanism might be the generation of carboxylic groups and subsequent evolution of CO 2 at the edges of graphite structures during oxidation. 44 Also exfoliation of graphene oxide sheets might be possible, if the resulting CO 2 gas pressure would overcome the van der Waals forces between the graphene sheets. 63 Both, the CO 2 evolution and exfoliation of the graphene oxide, might explain the dynamic change in θ after long-term treatment in the positive electrolyte. Moreover, as mentioned before, C dl provides information about the graphite part of the composite material only. Since θ is influenced by both components of the BPP composite material and their different chemical and morphological properties, a direct correlation between θ and the C dl cannot be drawn for such composite materials.
Because the studied BPPs are composite materials, further investigations were needed in order to assess the aging of the polypropylene phase in the aggressive electrolyte environment. Although the general chemical resistance of polyolefins is well established, it had already been reported that polypropylene can be oxidized even at ambient conditions after which several oxygen-containing functional groups within the polymer matrix were identified by FTIR. 46 Moreover it was also shown that these functional groups could decrease θ. 64 In this study the ATR-FTIR signals of the polypropylene would be masked by the larger signals of the graphite phase if measuring the BPPs and therefore corresponding experiments were conducted with pristine polypropylene granulate used for manufacturing of the BPPs. It was immersed in the positive electrolyte with 100% SOC for 120 days and ATR-FTIR measurements were performed before and after exposition to the electrolyte (Figure 10a). Both spectra show characteristic bands of symmetric and asymmetric valence vibration of saturated hydrocarbons (3000 to 2860 cm −1 ) as well as CH 2 and CH 3 bending vibrations (1460 cm −1 and 1378 cm −1 ). 46 A zoom-in to the spectral region 1730 to 1710 cm −1 reveals differences in the C=O valence bands (Figure 10b). The bands shift to higher wavenumbers if electron withdrawing groups are in the vicinity of the C=O group. 64 The band obtained for the as-received polypropylene sample at 1730 cm −1 can be associated to a carbonyl in ester groups -C(=O)OC-, which could have been formed during natural aging under ambient atmosphere or which are originated from other impurities. The favorable oxidation site during natural aging is the tertiary C atom, 46 where an ester group within one or between two polymer chains might form. In the spectrum of the aged sample the ester carbonyl group is diminishing and a new band appears at 1710 cm −1 , which corresponds to a C=O in an acid, aldehyde or keto group. 46,64 Thus it can be suggested, that the -C(=O)OC-groups that were already present within the polymeric  structure were hydrolyzed in the acidic aqueous solution to carboxyland alcohol groups. Subsequently, the alcohol functionalities could be further oxidized to keto and aldehyde groups. In addition, due to the oxidative nature of VO 2 + it is possible, that new hydroxyl or keto functionalities were created in the polymer chain via slow autoxidation. [65][66][67] However it can be considered that the oxidation of the polypropylene under the present conditions was rather negligible as indicated by the low intensity of both carbonyl features observed in the ATR-FTIR spectra of the pristine and aged polymer. Therefore, it can be concluded that mainly the graphite compound of the BPP was attacked by the highly oxidizing positive electrolyte.
The electrical conductivity of the BPPs was measured by means of the four point probe technique in order to investigate if the changed surface chemistry had an influence on one of the main functional properties of the BPPs. The pristine BPP had a conductivity of (33.9 ± 4.5) S/cm (Figure 11), which is similar or even higher than values reported for comparable BPP composites. 6,9 The aged BPP had electrical conductivities in the same order of magnitude. Thus the electroless aging of the BPPs did not influence significantly the electrical conductivity of the BPPs.

Conclusions
The chemical stability of graphite-polypropylene BPPs in VRFB was examined by a variety of techniques after the BPP samples were immersed in the positive and negative vanadium electrolyte with 0%, 20%, 80% and 100% SOC for up to 190 days simulating the aging at rest conditions. In the negative electrolyte with high SOC electroless H 2 gas evolution was observed on the surface of the BPPs, which led also to self-discharge of the electrolyte but did not lead to significant changes in the BPP properties. In contrast, immersion in the positive electrolyte changed the hydrophilicity of the BPPs. XP spectroscopy and electrochemical measurements indicated a progressing oxidation of the graphite with rising SOC and immersion time. It is likely that sulfuric acid and the VO 2 + ions induced the oxidation of the graphitic compounds which propagated with time. As a consequence further penetration of water and vanadium ions into the graphite structure of the BPP was promoted. Using FTIR also a slight change of the pure polypropylene material was detected after long-term immersion in the positive electrolyte with 100% SOC. However, mainly the already introduced oxygen containing functionalities, such as esters, seemed to undergo a hydrolysis reaction. The performed SEM and confocal microscopy studies did not provide significant evidence for morphological changes during aging. Moreover, the electrical conductivity of the BPPs was not affected by the immersion in the electrolytes indicating that the observed modification of the surface composition was not necessarily affecting the functionality of the BPPs. Nevertheless, in order to maintain the lifetime and functionality of the BPPs, it is recommended to avoid storage in a fully charged positive electrolyte for a long time to limit the corrosion effects. Moreover the propagating penetration of VO 2 + species within the BPP bulk could be crucial especially for thin current collectors. These aging effects might be pronounced in battery stacks where the electric potential difference is increased between two BPP that are located distal to the core of the stack. Therefore, further investigations under operating conditions are necessary to describe the aging effects of BPPs in realistic cells more precisely.