A H₃PO₄ preswelling strategy to enhance the proton conductivity of a H₂SO₄-doped polybenzimidazole membrane for vanadium flow batteries

Abstract

A H₃PO₄ preswelling strategy is proposed to prepare H₂SO₄-doped polybenzimidazole (PBI) membranes for vanadium flow batteries (VFB). Before being immersed in 3.0 M H₂SO₄, PBI membranes are preswelled by immersion in concentrated H₃PO₄, which leads to a higher H₂SO₄ doping level, thereby dramatically reducing the area resistance of the PBI membrane to 0.43 Ω cm², which is close to that of Nafion 212 (0.35 Ω cm²) and much lower than that of Fumasep®FAP-450 (0.64 Ω cm²). Meanwhile, the substantially high selectivity is maintained. The VFB assembled with the H₃PO₄ preswelled PBI membrane displays high energy efficiencies (EE: 80.9–89.2%) over a current density range of 20–80 mA cm⁻², much higher than those of the non-preswelled PBI membrane (EE: 66.8–84.5%), Nafion 212 (EE: 63.1–75.6%) and Fumasep®FAP-450 (EE: 75.5–82.6%). The stable performance over 50 charge–discharge cycles demonstrates the good physicochemical stability of the preswelled PBI membrane. Considering the above results, the H₃PO₄ preswelling strategy proposed herein is facile and efficient for fabricating high-performance PBI membranes for VFB.

---

1. Introduction

Renewable energy offers versatile solutions to address the ever more urgent energy and environmental concerns, and the huge desire for green and sustainable energy worldwide propels their implementation and development. Unfortunately, the random and seasonal nature of renewable energy such as solar and wind energy leads to low-quality output electricity and hinders the effective utilization of renewable energy in the grid. Large-scale energy storage is the key technology to solve these problems since it can convert and store electricity to overcome the mismatch between generation and end-use.^1–4 Vanadium redox flow battery (VFB), originally proposed by Skyllas-Kazacos et al.,⁵ is well-suited for large-scale utility applications because of its technical advantages including flexible design, high efficiency, high safety, rapid response and long cycle life.^6,7

As the key part of a VFB, a membrane is responsible for isolating active species and transferring ions (H⁺, SO₄²⁻) to complete the electrical circuit.⁸ An ideal membrane should meet the tough requirements of high ionic conductivity, low vanadium permeability and water transfer rate, good chemical and mechanical stability and low cost.^7,9 The membrane commonly used in VFB is perfluorosulfonic acid polymers (e.g., Dupont's Nafion), as this kind of membrane possesses both high proton conductivity and excellent chemical stability. However the perfluorosulfonic acid membranes also suffer from serious vanadium cross-contamination¹⁰ and economical prohibited cost, these drawbacks hindered their further commercialization.^7,9,11 Hence, alternative low-cost separators have been developed and investigated for VFB, mainly including cation exchange membranes,^12–15 anion exchange membranes,^16–21 and porous membranes.^22–27 Although considerable progresses have been made such as reducing cost and improving ion selectivity, there are still lots of issues to be solved, such as improving chemical stability and coordinating trade-off between ion conductivity and selectivity.^13,28–30

Polybenzimidazoles (PBIs) are well-known for their excellent chemical stability, mechanical strength, thermal stability, durability and low cost.^31–34 Chemically, PBIs are basic polymers (pK_a = 5.23 as protonated) and can readily react with a strong acid through hydrogen bonding.^35,36 When doped with acids, PBI membrane becomes proton conductive. The acid doped PBI membrane seems to be the most successful system for high temperature polymer electrolyte membrane fuel cells (PEMFC) because of its high conductivity, good mechanical properties and excellent thermal stability at high temperature. It is expected to be an appropriate system for VFB. Because the working medium of PBI is H₂SO₄ solution in VFB, the membrane should be free from an issue of acid leakage. Zhou et al.³⁷ prepared a H₂SO₄-doped PBI membrane for VFB, by immersing a PBI membrane in 4 M sulfuric acid solution for 7 days. The PBI-based VFB exhibited a substantially high Coulombic efficiency (CE) which is attributed to low crossover of vanadium ions. However, the voltage efficiency (VE) or ionic conductivity needs to be further improved. The possible reason for the low ionic conductivity is that the H₂SO₄ doping level is too low to form continuous proton conductive channels in the membrane. The H₂SO₄ doping level is limited by the fairly dense microstructure of PBI membranes owing to the PBI's structure rigidity and strong hydrogen bonding.^33,34,38,39

In this work, a H₃PO₄ preswelling strategy was proposed to increase the H₂SO₄ doping level and then prepare a high-performance H₂SO₄-doped PBI membrane for VFB. Before immersed in 3.0 M H₂SO₄ solution, dry PBI membrane was preswelled by concentrated (85 wt%) phosphoric acid. During the preswelling process, compared with directly immersing PBI membrane in 3.0 M H₂SO₄ solution, more hydrogen bonds in PBI matrix are expected to be broken since more acid will react with PBI (higher acid doping level (ADL)). Besides, the PBI membrane will experience greater swelling and further create free volume at this stage. Then the preswelled membrane was transferred to 3.0 M H₂SO₄ solution immediately, where the H₃PO₄ in the membrane could be exchanged with H₂SO₄ and water molecules and result in higher ADL. It was proved that PBI membrane treated with this facile and practicable method exhibited high proton conductivity while excellent ion selectivity was retained. The detailed characteristics as well as VFB performance of the present acid-doped PBI membranes will be discussed in detail.

2. Experimental section

2.1 Materials

Poly(4,4′-diphenylether-5,5′-bibenzimidazole) (denoted as PBI) was purchased from Shanghai Shengjun Plastics Technology Co., Ltd. N,N-dimethylacetamide (DMAc) and phosphoric acid were bought from Tianjin Fuyu Fine Chemical Co., Ltd. and Damao Chemical Agents Company, Tianjin, China, respectively. The reference membranes, Fumasep®FAP-450 (denoted as FAP-450) and Nafion 212, were supplied by FuMA-Tech GmbH (Germany) and DuPont (USA), respectively.

2.2 Membrane preparation and modification

PBI powder was dissolved in DMAc at 50 °C with continuous stirring for 12 h to form a 5 wt% solution. The PBI membrane was cast by evaporating the DMAc solvent at 60 °C for one day and 120 °C for an additional day. Then transparent and homogeneous thin films (∼50 μm) were obtained. The membranes were washed with hot water for 12 h and dried in vacuum at 120 °C for 12 h to remove small traces of the solvent. Two methods were applied to prepare H₂SO₄-doped PBI membranes. The one was immersing PBI membranes directly in 3.0 M H₂SO₄ for different time (Scheme 1(I)); the other was immersing PBI membranes in 85 wt% H₃PO₄ for different time before immersing in 3.0 M H₂SO₄ (Scheme 1(II)). Treatment conditions of the prepared PBI membranes are listed in Table 1.

Scheme 1 The illustration of the H₃PO₄ preswelling method (II) vs. the traditional direct-immersion method (I).

Table 1 Treatment conditions of the prepared PBI membranes

Mark^a	Immersion time in 85 wt% H3PO4/h	Immersion time in 3.0 M H2SO4/h
a S represents 3.0 M sulfuric acid solution and P represents concentrated phosphoric acid solution; the number ahead represents the time (hour) PBI membranes were immersed in corresponding acid.
1.5P	1.5	0
72P	72	0
24S	0	24
1.5P–24S	1.5	24
72P–24S	72	24

---

2.3 Membrane characterization

ADL and swelling ratio. For the determination of ADL, defined as the molar ratio of acid to PBI repeat unit in each hybrid membrane, membranes were doped by immersing them in acid solutions for different time at 30 °C. After that, the membranes were wiped with a tissue to remove the extra acid on the membrane surface and dried at 80 °C until a constant weight was reached. On the assumption that the weight loss was due to water evaporation, the ADL was calculated according to eqn (1).where W_dry and W_acid are the weights of the PBI membrane before and after being soaked in acid and dried in an oven at 80 °C, respectively, M_acid and M_PBI are the molecular weights of acid and repeat unit of PBI, respectively.

After acid doping, the dimensional changes of each hybrid membrane were recorded. The area, thickness and volume swelling ratios are defined as eqn (2), (3) and (4), respectively.

A_dry, T_dry and V_dry are the area, thickness and volume values of the dry membrane before immersed in acid, respectively. A_acid, T_acid and V_acid are the area, thickness and volume values of the membrane after immersed in acid, respectively.

Field emission scanning electronic microscopy (FE-SEM) and energy dispersive X-ray spectrometer (EDS). Surface and cross-section morphologies of membranes were collected by FE-SEM (Nova NanoSEM 450). Dry samples were broken in liquid nitrogen to bare the cross-section surface and were coated with gold before the measurement. EDS spectra of prepared membranes were obtained by EDS detector attached to the FE-SEM.

FTIR analyses. The FTIR spectra of pristine and hybrid PBI membranes were measured using a FTIR spectrometer (Nicolet 6700) operating in the wavenumber range 600–4000 cm⁻¹.

Vanadium permeability. In a VFB, the permeability of different vanadium ions induces self-discharge and capacity decay, having great influence on battery performance. According to previous literature,⁴⁰ the diffusion coefficient of four kinds of vanadium ions were in the order of V²⁺ > VO²⁺ > VO₂⁺ > V³⁺. Because V²⁺ is easily oxidized by oxygen in air, the permeability of VO²⁺ was usually detected to characterize the ion selectivity of the membrane. Vanadium permeability of PBI membranes was identified according to literature²³ via a diffusion cell with a membrane in the middle as the barrier. The left cell was filled with 60 ml of 1.5 M VOSO₄ in 3.0 M H₂SO₄ and the right one was filled with 60 ml of 1.5 M MgSO₄ in 3.0 M H₂SO₄ to equalize ionic strength and minimize the effect of osmotic pressure. Magnetic stirrers were used in both cells to mitigate concentration polarization. Samples from the right cell were collected every 12 h. The concentration of VO²⁺ in sample solution was measured by UV-vis spectrometer. The vanadium permeability is calculated according to the Fick's diffusion law, shown as follows:V_B is the volume of MgSO₄ solution; A and L are the effective area and thickness of the membrane; P is the permeability of vanadium ion; C_A is the vanadium ion concentration on the VOSO₄ side; C_B(t) is the vanadium ion concentration on the MgSO₄ side as a function of time.

Area resistance (AR). The AR of membranes were measured by the method described elsewhere.²⁴ A conductive cell was separated by the sample membrane into two compartments filled with 3.0 M H₂SO₄ solution. The effective area (S) of the cell was 3.2 cm². r₁ and r₂ are the electric resistances of the cell with and without a membrane respectively. They were measured by electrochemical impedance spectroscopy (EIS) over a frequency range from 1 MHz to 1 Hz. The AR is calculated by eqn (6):

AR = (r₁ − r₂)S (6)

Mechanical properties. The mechanical property of fully hydrated membrane was measured at room temperature with a tensile tester (CMT8502, Shenzhen SANS Testing Machine Co. Ltd, China) at a constant tensile speed of 5 mm min⁻¹.

VFB single cell performance. A VFB single cell was assembled by sandwiching a membrane with two carbon felt electrodes (Gansu Hongwei Carbon Ltd, China), clamped by two graphite polar plates. The carbon felt electrode was oxidized in an air atmosphere at 400 °C for 6 h to improve electrochemical activity and hydrophilicity. All these components were fixed between two epoxy fiber glass plates. 40 ml of 1.5 M V²⁺/V³⁺ in 3.0 M H₂SO₄ and 1.5 M VO²⁺/VO₂⁺ in 3.0 M H₂SO₄ solutions were used as the negative and positive electrolytes, respectively. The electrolytes were cyclically pumped through the corresponding electrodes in pipe lines. The effective area of the separator was 9 cm². Charge–discharge cycling tests were conducted using a LANHE battery tester (CT2001A, 5V/1A) under a constant current density of 50 mA cm⁻². The cut-off voltages for charge and discharge were set at 1.65 V and 0.8 V, respectively, to avoid the corrosion of carbon felts and graphite polar plates.

CE, VE and energy efficiency (EE) are three important indexes to evaluate cell performance. CE is defined as the ratio of a cell's discharge capacity divided by its charge capacity. It is an indicator of the ability of preventing vanadium transfer. VE is defined as the ratio of discharge voltage divided by charge voltage. It is an indicator of the ion conductivity of membranes. EE, which is regarded as the most important efficiency, is defined as the ratio of discharge energy divided by charge energy. They are calculated according to eqn (7), (8) and (9) respectively.

where I_d is the discharging current, I_c is the charging current, V_d is the discharging voltage and V_c is the charging voltage.

Chemical stability. The chemical stability of the prepared membrane was investigated by an ex situ immersion method. A membrane sample was immersed in 100 ml electrolyte solution (1.5 M VO₂⁺ in 3.0 M sulfuric acid solution) at room temperature for 30 days. The weight, AR and vanadium permeability of the sample were recorded before and after the test to determine the changes.

3. Results and discussion

3.1 ADL and swelling ratio

Immersion of a PBI membrane in an acid solution results in an increase in weight and volume. These weight gains are due both to the water uptake and to acid doping. After separate the water content by drying hybrid membranes at 80 °C under vacuum, the membrane weight reaches a constant value and the amounts of acid in the doped membranes are obtained. Fig. 1a shows the ADL increase versus time for PBI membranes after being immersed in 85 wt% H₃PO₄ and 3.0 M H₂SO₄, respectively. In concentrated H₃PO₄ solution, the ADL increases rapidly with immersion time for the first 12 h. After that, the ADL of PBI membranes increases slightly. A doping level around 7.2 can be achieved by immersing PBI membranes in concentrated H₃PO₄ for 72 h at 30 °C. In 3.0 M H₂SO₄ solution, it is seen that an equilibrium time of only 3 h is needed to reach saturation at 30 °C. The ADL of PBI membranes achieved in concentrated H₃PO₄ is much higher than that of H₂SO₄-doped membrane in 3.0 M H₂SO₄ solution. That's because the concentration of acid solution is the main factor controlling the ADL of PBI membranes and high concentration of acid solution improves the ADL as reported elsewhere.^35,41

Fig. 1 Effect of immersion time on the (a) ADL and (b) swelling ratio of PBI membranes in 85 wt% H₃PO₄ and 3.0 M H₂SO₄.

Fig. 1b shows the dimensional variations of acid-doped PBI membranes. The swelling ratios are found to be anisotropic between in-plane and thickness directions. The swellings in plane direction for both H₃PO₄-PBI membranes and H₂SO₄-PBI membranes are very low while the swelling in the thickness direction are significant, which is also reported by H. Dai et al. previously,⁴² suggesting that the molecular chain of PBI may have specific orientations along the membrane plane. Besides, H₃PO₄-doped membrane demonstrates a much higher dimensional change than H₂SO₄-doped membrane owing to its higher ADL. In this work, the immersion time of 1.5 h and 72 h in 85 wt% H₃PO₄, representing short and long preswelling time respectively, are selected for further study and the corresponding membranes are referred to as 1.5P and 72P, respectively (Fig. 1a).

During the immersion in 85 wt% H₃PO₄, plenty of hydrogen bonds in PBI matrix are broken meanwhile acid anions are linked to the polymer by hydrogen bonding. The membranes experience greater swelling and further create free volume at this stage, when 1.5P and 72P are transferred to 3.0 M H₂SO₄ solution immediately to exchange the H₃PO₄ in the membrane with H₂SO₄ and water in the H₂SO₄ solution. The ADL and dimensional variations of 1.5P and 72P in 3.0 M H₂SO₄ solution with the immersion time are also investigated and are showed in Fig. 2. It is worth noting that since the molecular weights of H₃PO₄ and H₂SO₄ are the same, the ADL could also be calculated by the weight method although the type of the doped-acid is unclear.

Fig. 2 Effect of immersion time (x) on the (a) ADL and (b) swelling ratio of 1.5P and 72S membranes in 3.0 M H₂SO₄.

As can be seen from Fig. 2a, the ADL of sample membranes decrease sharply initially and then increase in a certain degree as the immersion time prolonged. In the beginning, the concentration of H₃PO₄ in the PBI membrane is extremely high. When it is immersed in 3.0 M H₂SO₄, a large amount of H₃PO₄ flows out immediately, of which the rate is much higher than the inflow rate of H₂SO₄, hence the ADL drop rapidly. Afterwards, a slight improvement of ADL might be ascribed to the concentrating effect of PBI with H₂SO₄ (i.e., the acid concentration in the membrane is higher than that of the surrounding medium), which takes a period of time as reported elsewhere.⁴³ After 24 h, the ADL in both cases increase slowly and reach equilibrium. The ADL of 1.5P–24S (immersing 1.5P membrane in 3.0 M H₂SO₄ for 24 h) is approximately 2 per repeat unit of PBI, whereas that of 72P–24S (immersing 72P membrane in 3.0 M H₂SO₄ for 24 h) increases to 2.5. The ADL and volume swelling ratios of 24S (immersing PBI membrane in 3.0 M H₂SO₄ for 24 h), 1.5P–24S and 72P–24S membranes are summarized in Table 2. It can be noticed that the ADL is improved by preswelling membrane in concentrated phosphoric acid and prolonged preswelling time benefits ADL improvement. Swelling ratios follow the same law (Fig. 2b). On the basis of the ADL of 24S membrane, ADL increases by 25.0% and 56.2% for 1.5P–24S and 72P–24S membranes, respectively. It is expected to enhance the ionic conductivity of the membrane, which will be confirmed by AR in the Section 3.3. In the following sections, the properties of 24S, 1.5P–24S and 72P–24S will be further investigated.

Table 2 Thickness, ADL and swelling ratio of hybrid PBI membranes, Nafion 212 and FAP-450

Membrane	24S	1.5P–24S	72P–24S	Nafion 212	FAP-450
a Measured after the membranes are immersed in 3.0 M H2SO4 for 24 h at room temperature.
Wet thickness/μm	61	70	76	60	60
ADL/mol acid	1.6	2.0	2.5	—	—
Volume swelling ratio/%	36.5	46.8	70.9	32.5^a	53.8^a

---

3.2 Membrane morphology and composition

SEM images are recorded to investigate the effect of the different treatment conditions on the hybrid membranes. The surface and cross section of pristine PBI membrane are smooth as shown in Fig. 3a–c. By contrast, small pores are formed on the surface (Fig. 3d) and some wrinkled areas emerge in the cross section (Fig. 3e and f) after doped with H₂SO₄. In the cases of 1.5P–24S (Fig. 3g–i) and 72P–24S (Fig. 3j–l), the surfaces are smooth and some wrinkles show up in the cross sections. From the SEM images, it is suggested that although the swellings ratios are obvious, 1.5P–24S and 72P–24S membranes retain dense and integral morphologies and no evident pores could be observed. In addition, it is noted that the thicknesses of 1.5P–24S and 72P–24S membranes increase compared with 24S and pristine PBI membranes due to H₃PO₄ preswell and improved ADL.

Fig. 3 SEM images of membrane surfaces, cross-sections and magnification of cross-sections (from left to right in each row). (a–c) Pristine PBI; (d–f) 24S; (g–i) 1.5P–24S; (j–l) 72P–24S; EDS spectra of cross-sections of (m) 1.5P–24S and (n) 72P–24S.

In order to investigate which kind of acid (H₂SO₄ or H₃PO₄ or both) are doped in 1.5P–24S and 72P–24S, the EDS analysis is conducted and the spectra are shown in Fig. 3m and n. The EDS spectra of the cross-sections of 1.5P–24S and 72P–24S membranes display only the peaks assigned to sulfur element except for the elements in PBI matrix (nitrogen element doesn't appear due to its low content and the instrument's weak sensitivity to nitrogen), which indicates that the H₃PO₄-doped in the preswelling process are replaced by H₂SO₄ and water molecules after immersing 1.5P and 72P membranes in 3.0 M H₂SO₄ for 24 h.

The IR spectra of pristine PBI membrane and acid doped PBI membranes are also investigated to further confirm the elimination of H₃PO₄ and give more detailed information about the acid doping situation. The peaks in the region of 600–2000 cm⁻¹ are exhibited in Fig. 4 since the characteristic intense absorption bands of the anions center in the spectral region. In the IR spectra of 1.5P, two strong absorptions at 938 cm⁻¹ and 1081 cm⁻¹ appear as compared with pristine PBI membrane, which are assigned to the presence of H₂PO₄⁻ according to the report by Bouchet and Siebert.⁴⁴ The associated anion is H₂PO₄⁻ in most of the acid concentration range for H₃PO₄-doped PBI membranes based on previous studies. In the cases of 1.5P–24S and 72P–24S, the peaks assigned to H₂PO₄⁻ disappear, which further confirms the complete elimination of H₃PO₄. Whereas the bands at 1159 cm⁻¹, 1035 cm⁻¹ and 899 cm⁻¹ arise. These absorptions are attributed to HSO₄⁻, the dominant anion in PBI-H₂SO₄ membrane when the ADL is above 0.6.^43,44 What's more, the band located around 1386 cm⁻¹ appears in the IR spectra of 24S, 1.5P–24S and 72P–24S, which is assign to the symmetric stretching vibration of S=O₂ in H₂SO₄. It implies the presence of free H₂SO₄ in these H₂SO₄-doped PBI membranes, which could make a significant contribution to proton conductivity compared with the so-called “bonded acid”.⁴⁵

Fig. 4 IR spectra of pristine PBI membrane and acid doped PBI membranes in the region of 600–2000 cm⁻¹.

The investigation above, on the one hand, makes it possible to prepare PBI-H₂SO₄ membranes with improved ADL in 3.0 M H₂SO₄ solution (the supporting electrolyte for VFB in this work) by preswelling PBI membrane in concentrated phosphoric acid and the ADL could be adjusted by varying the preswelling time conveniently. On the other hand, it demonstrates that H₃PO₄-doped PBI membrane is not stable in conventional VFB system, since H₃PO₄ will be definitely washed away by cycling electrolyte and replaced by H₂SO₄ and a small amount of vanadium ions (Section 3.5).

3.3 Vanadium permeability, AR and mechanical strength

The permeation of vanadium ions through membrane could cause self-discharge and capacity decay. Hence high ion selectivity of a membrane is essential to obtain a good cell performance. Fig. 5a shows a linear relationship between VO²⁺ ion concentration and diffusion time for each sample and the slop reflects the diffusion rate of VO²⁺ ions. Apparently, Nafion 212 exhibits the highest VO²⁺ diffusion rate, because of its large ionic channels (at typically 4–5 nm in diameter) through the membrane.⁴⁶ FAP-450 demonstrates a lower VO²⁺ diffusion rate than Nafion 212, possibly due to the fixed positive charges in the membrane, which can repulse vanadium cations through the Donnan exclusion effect. All of the prepared PBI membranes show much lower VO²⁺ diffusion rate than Nafion 212 and FAP-450 attributed to the dense microstructure of PBI membranes, exhibiting their excellent abilities to prevent VO²⁺ cross over. Comparison of 24S, 1.5P–24S and 72P–24S reveals that the vanadium permeability slightly increases with the increase in ADL and the swelling ratio of the PBI membranes. As calculated by eqn (5), the vanadium permeabilities of 24S and 1.5P–24S are undetectable while the vanadium permeabilities of 72P–24S, Nafion 212 and FAP-450 are 0.30 × 10⁻⁷ cm² min⁻¹, 3.05 × 10⁻⁷ cm² min⁻¹ and 1.31 × 10⁻⁷ cm² min⁻¹, respectively.

Fig. 5 (a) Vanadium diffusion and (b) AR and ionic conductivity of the prepared PBI membranes, Nafion 212 and FAP-450.

Fig. 5b displays the comparison between the AR of the as-prepared PBI membranes, Nafion 212 and FAP-450 in 3.0 M H₂SO₄ solution. The AR decrease from 1.32 Ω cm² of 24S to 0.43 Ω cm² of 72P–24S owing to the increase of ADL from 1.6 to 2.5. For acid doped PBI membranes, the nature of anion is reported to be the main factor governing the conductivity.⁴⁴ For the PBI-H₂SO₄ system, the polymer is associated with HSO₄⁻ and H₂SO₄ for ADL > 1.5,⁴⁴ which is also confirmed by the IR spectra in Section 3.2. Hence, further increase in ADL corresponds to more free acid in the PBI membranes, which is expected to improve the proton conductivity of PBI-H₂SO₄ membranes significantly. The lowest AR is yielded by 72P–24S (0.43 Ω cm²), which is close to that of Nafion 212 (0.35 Ω cm²) and much lower than that of FAP-450 (0.64 Ω cm²). The ionic conductivity calculated from the AR is also presented in Fig. 5b to directly compare the intrinsic properties of the membranes. It can be seen that the ionic conductivity is dramatically enhanced by the H₃PO₄ preswelling method and the ionic conductivity of 72P–24S is even a little higher than that of Nafion 212. The results suggest that the relative higher AR of 72P–24S than that of Nafion 212 is attributed to the larger thickness of 72P–24S.

The swelling behavior and ADL of acid-doped PBI membranes could substantially influence the mechanical strength of the membrane. The mechanical characteristics of fully hydrated PBI-H₂SO₄ membranes and commercial membranes are measured at room temperature and displayed in Table 3. The PBI sample without acid loading exhibits a tensile strength of 93.2 MPa and an elongation at break of 48.1%. Compared with pristine PBI membrane, acid-doped membranes have higher elongation at break, attributed to the plasticizing effect of acid and water. In addition, this effect is more obvious in hybrid membranes with higher ADL and swelling ratio. However, the membrane strength reduces when doped with sulfuric acid. The tensile strength of 1.5P–24S and 72P–24S are respectively about 73 MPa and 64 MPa.

Table 3 Mechanical strength of hydrated PBI membranes, Nafion 212 and FAP-450

Membrane	PBI	24S	1.5P–24S	72P–24S	Nafion 212	FAP-450
Tensile strength/MPa	93.2	79.6	73.1	63.8	21.5	19.8
Elongation at break/%	48.1	64.1	84.3	96.3	263.1	321.3

---

Improving loading level of PBI membrane enhances its conductivity, which inevitably results in a sacrifice of its mechanical strength, originating from the weakening of intermolecular forces as a result of an extended distance between PBI backbones. Although the tensile strengths of hybrid membranes decrease compared with pristine PBI membranes, they are still much higher than that of commercial Nafion and fumasep membranes, which benefits cell operation.

3.4 Single cell performance

Single cell performances of VFBs assembled with 24S, 1.5P–24S, 72P–24S, Nafion 212 and FAP-450 operating under various current densities are shown in Fig. 6. As expected, the CEs of VFBs with acid-doped PBI membranes are higher than that of Nafion-based and FAP-450-based VFBs under entire test current densities due to the lower vanadium permeability. The CEs of 24S, 1.5P–24S and 72P–24S membranes reach almost 100% under high current densities. With preswell in concentrated phosphoric acid, the VE of the VFB assembled with 72P–24S membrane is improved dramatically, which is consistent with the AR values in Fig. 5b. This is due to the improvement of ADL and swelling ratios of 1.5P–24S and 72P–24S, which endows the membranes with more proton transfer sites and larger proton transfer pathway.^47–49

Fig. 6 VFB performances of 24S, 1.5P–24S, 72P–24S membranes vs. Nafion 212 and FAP-450 under different current densities.

The CEs increase with the increase in current density, assigned to shorter crossover time of vanadium species. Whereas the VEs vary inversely with the current density on account of the increased ohmic polarization. EE is a key parameter to evaluate energy loss during charge–discharge process. The VFB with 72P–24S displays the highest EE which is higher than 80% (80.9–89.2%) under entire test current densities. Compared with the commercial membranes, 72P–24S exhibits higher CE and EE under various current densities, demonstrating a promising prospect of acid-doped PBI membranes for VFB application. Comparison of 1.5P–24S and 72P–24S reveals that prolonged preswelling time is beneficial to improve cell performance. However, further prolonged preswelling time (e.g., 10 days) is inadvisable since it brings marginal improvement in cell efficiencies (not shown here) and makes the process time-consuming.

Charge–discharge curves of different membranes under various current densities are presented in Fig. 7. The charge–discharge overpotential of VRB single cell decreases obviously with the increase in ADL of PBI hybrid membranes and with the decrease in current density, which might lead to the VE enhancement in Fig. 6. Meanwhile, the charge–discharge capacities are greatly improved.

Fig. 7 Charge–discharge curves of VFB single cell with (a) 24S; (b) 1.5P–24S; (c) 72P–24S; (d) Nafion 212; (e) FAP-450.

The results above demonstrate that the H₃PO₄ preswelling strategy is effective in improving ADL and proton conductivity of acid-doped PBI membranes and thus improving cell performance of VFB.

3.5 Charge–discharge cycling test

The cycling test is carried out to verify the stability of acid-doped PBI membranes. As shown in Fig. 8, the VE, CE and EE of 72P–24S-based VFB show no obvious decay over 50 cycles at 50 mA cm⁻², indicating that the prepared PBI membranes are chemically endurable to survive highly acidic and oxidizing environment and retain stable selectivity and conductivity, which is further verified by the chemical stability test results in Table 4. After 30 days of immersion, the weight of the sample slightly increases, which might be attributed to the VO₂⁺ ions absorbed by the membrane. The AR and ion selectivity of the tested membrane are as good as those of the membrane before the test. These results show that the PBI membrane is apparently durable in this immersion test. This could originate from the inherent excellent chemical stability of the PBI materials, which can also be confirmed by the long term cycle test in a previous study.⁵⁰ The stable voltage efficiency also indicates that the effect of acid leakage in the PBI membrane could be negligible, since the supporting electrolyte for VFB is 3.0 M H₂SO₄ solution.

Fig. 8 The cycling performance of VFB assembled with 72P–24S at 50 mA cm⁻².

Table 4 Stability test results of 72P–24S

Items	Before test	After test
Weight/mg	154.1	162.7
AR/Ω cm^2	0.44	0.42
Permeability/cm^2 min^−1	0.30 × 10^−7	0.15 × 10^−7

---

The charge–discharge capacities of VFBs assembled with various membranes over cycling are displayed in Fig. 9a. Since the concentration and volume of electrolytes in each test are kept identical, the discharge capacity reflects the utilization ratio of active species. Although the charge capacities of Nafion-based VFB are much higher than those of the VFB with 72P–24S, the corresponding discharge capacities are much more inferior due to the severe vanadium cross contamination. It implies not only a low utilization ratio of active species but also a waste of electric energy. The discharge capacities of VFB with 72P–24S increase significantly as compared with that of 24S, due to its lower AR. Meanwhile, the rate of the capacity decay is not accelerated by the preswelling process. Moreover, the battery with 72P–24S shows much higher discharge capacities than those with commercial FAP-450 and Nafion 212 over cycling.

Fig. 9 (a) Charge–discharge capacity and (b) discharge capacity retention of the VFBs assembled with various membranes as a function of cycle number at 50 mA cm⁻².

Capacity decay over cycling is a concern for VFBs, which is usually attributed to the crossover of vanadium ions. However, Fig. 9b shows that the capacity retentions of 24S, 72P–24S are almost the same as that of Nafion 212, although the PBI membranes show higher abilities to prevent vanadium ion crossover than Nafion 212. Besides, the VFB with FAP-450 experiences the fastest capacity decay although the crossover of vanadium ions through FAP-450 is not the severest. The possible reason may be that the capacity fade mechanisms of different membranes are distinctly different.^51–54 As reported by Sun,⁵⁴ the capacity decay of VFB assembled with Nafion membranes mainly results from the crossover of vanadium ions, whereas the side reactions are the major factor for VX-20 (a kind of anion exchange membrane). The underlying capacity decay mechanism of the VFB assembled with H₂SO₄-doped PBI membranes might be different and needs to be further explored since they are not conventional ion exchange membranes which possess ion exchange groups.

SEM images of 72P–24S membrane after cycling test are shown in Fig. 10. The flat surfaces of the tested PBI membrane facing positive (Fig. 10a) and negative (Fig. 10b) half cells as well as the cross section (Fig. 10c) maintain an integrated internal structure, and no damage and reduction in thickness can be observed. EDS spectra of the tested PBI membrane are shown in Fig. 10a′′–c′′ and the corresponding atomic ratios are presented in Table 5. For comparison, the value of initial 72P–24S is also listed. As shown in Table 5, the atomic ratio of C/O/S in the cross section of the tested 72P–24S remains almost the same as that of the initial membrane, indicating its good physicochemical stability. Besides, the atomic ratios of C/O/S on the surfaces and in the cross section of the tested membranes are similar. It demonstrates the uniform distribution of sulfuric acid in PBI membranes, which is further confirmed by the uniform distribution of S element in the mapping (Fig. 10a′–c′). As it can be seen from Fig. 10a′′–c′′ and Table 5, vanadium element is observed in the tested PBI membrane. However, the content is extremely low. It suggests that it is difficult for vanadium ions to permeate into acid-doped PBI membranes, which would explain the minute vanadium crossover and good chemical stability of the PBI-based membranes in VFB.

Fig. 10 SEM images, mapping of S and EDS spectra of 72P–24S membrane after cycling test (a) (a′) (a′′) surface facing positive half cell; (b) (b′) (b′′) surface facing negative half cell; (c) (c′) (c′′) cross section.

Table 5 Atomic ratios of 72P–24S membrane before and after cycling test

Sample	C/O/S/V atomic ratio
a After cycling test.b Without cycling test.
Surface facing positive half cell^a	45.05 : 45.81 : 7.70 : 1.44
Surface facing negative half cell^a	52.12 : 40.15 : 7.08 : 0.66
Cross section^a	58.02 : 34.06 : 7.43 : 0.49
Cross section^b	53.80 : 38.85 : 7.35 : 0

---

4. Conclusions

In summary, PBI-H₂SO₄ membranes with improved ADL and proton conductivity are prepared by H₃PO₄ preswelling strategy and investigated in VFB applications. The ADL of the preswelled acid-doped PBI membranes are improved significantly while the excellent mechanical strength and ion selectivity are retained. EDS and FTIR analysis demonstrate that the H₃PO₄ doped in the preswelling process is eliminated completely and the acid doped in PBI membrane is only sulfuric acid at the final stage. AR and vanadium permeability test prove that H₃PO₄ preswelling strategy is effective to improve the proton conductivity, and in the meanwhile, sustain the excellent ion selectivity of PBI hybrid membranes. The AR is reduced from 1.32 Ω cm² of 24S to 0.43 Ω cm² of 72P–24S in 3.0 M H₂SO₄ solution, which is close to that of Nafion 212 (0.35 Ω cm²) and much lower than that of FAP-450 (0.64 Ω cm²). Meanwhile, the vanadium permeability obtained is 0.30 × 10⁻⁷ cm² min⁻¹, one order of magnitude lower than that of Nafion 212 (3.05 × 10⁻⁷ cm² min⁻¹) and FAP-450 (1.31 × 10⁻⁷ cm² min⁻¹). As a result, 72P–24S membrane exhibits EEs higher than 80% (80.9–89.2%) under entire test current densities (20–80 mA cm⁻²). This is a better performance compared with 24S (EE: 66.8–84.5%), Nafion 212 (EE: 63.1–75.6%) and FAP-450 (EE: 75.5–82.6%). The stable cycling performance of the single cell along with the SEM and EDS analysis of the tested 72P–24S membrane demonstrate an outstanding physicochemical stability of the prepared PBI membranes. Moreover, the discharge capacity of VFB assembled with 72P–24S is higher than those of VFBs assembled with 24S, Nafion 212 and FAP-450 over cycling and the obtained discharge capacity retention is also higher than that of FAP-450. This work provides a facile and efficient strategy to enhance the proton conductivity of H₂SO₄-doped PBI membranes and to prepare high-performance PBI membranes for VFB applications.

Acknowledgements

The authors thank the support of National Science Fund for Distinguished Young Scholars of China (Grant No. 21125628), National Natural Science Foundation of China (Grant No. 21406031 and 21476044), the State Key Laboratory of Fine Chemicals (KF1507), the Program for Liaoning Excellent Talents in University (LR2014003), and State Key Laboratory of fine chemicals (Panjin) project (Grant No. JH2014009).

References

1. H. Chen, T. N. Cong, W. Yang, C. Tan, Y. Li and Y. Ding, Prog. Nat. Sci., 2009, 19, 291.
2. S. Chu and A. Majumdar, Nature, 2012, 488, 294.
3. K. Gong, Q. R. Fang, S. Gu, S. F. Y. Li and Y. S. Yan, Energy Environ. Sci., 2015, 8, 3515.
4. P. Leung, X. Li, C. Ponce de León, L. Berlouis, C. T. J. Low and F. C. Walsh, RSC Adv., 2012, 2, 10125.
5. M. Skyllas-Kazacos, M. H. Chakrabarti, S. A. Hajimolana, F. S. Mjalli and M. Saleem, J. Electrochem. Soc., 2011, 158, R55.
6. W. Wang, Q. Luo, B. Li, X. Wei, L. Li and Z. Yang, Adv. Funct. Mater., 2013, 23, 970.
7. X. Li, H. Zhang, Z. Mai, H. Zhang and I. Vankelecom, Energy Environ. Sci., 2011, 4, 1147.
8. S. Gu, K. Gong, E. Z. Yan and Y. S. Yan, Energy Environ. Sci., 2014, 7, 2986.
9. B. Schwenzer, J. Zhang, S. Kim, L. Li, J. Liu and Z. Yang, ChemSusChem, 2011, 4, 1388.
10. S. Lu, C. Wu, D. Liang, Q. Tan and Y. Xiang, RSC Adv., 2014, 4, 24831.
11. T. N. L. Doan, T. K. A. Hoang and P. Chen, RSC Adv., 2015, 5, 72805.
12. D. Chen, S. Wang, M. Xiao and Y. Meng, Energy Environ. Sci., 2010, 3, 622.
13. C. Fujimoto, S. Kim, R. Stains, X. Wei, L. Li and Z. G. Yang, Electrochem. Commun., 2012, 20, 48.
14. S. Kim, J. Yan, B. Schwenzer, J. Zhang, L. Li, J. Liu, Z. Yang and M. A. Hickner, Electrochem. Commun., 2010, 12, 1650.
15. D. Chen, S. Kim, L. Li, G. Yang and M. A. Hickner, RSC Adv., 2012, 2, 8087.
16. D. Chen, M. A. Hickner, E. Agar and E. C. Kumbur, Electrochem. Commun., 2013, 26, 37.
17. M.-S. J. Jung, J. Parrondo, C. G. Arges and V. Ramani, J. Mater. Chem. A, 2013, 1, 10458.
18. Z. Mai, H. Zhang, H. Zhang, W. Xu, W. Wei, H. Na and X. Li, ChemSusChem, 2013, 6, 328.
19. F. Zhang, H. Zhang and C. Qu, J. Phys. Chem. B, 2012, 116, 9016.
20. H. Zhang, H. Zhang, F. Zhang, X. Li, Y. Li and I. Vankelecom, Energy Environ. Sci., 2013, 6, 776.
21. D. S. Zhang, X. M. Yan, G. H. He, L. Zhang, X. H. Liu, F. X. Zhang, M. M. Hu, Y. Dai and S. S. Peng, J. Mater. Chem. A, 2015, 3, 16948.
22. J. Cao, H. Zhang, W. Xu and X. Li, J. Power Sources, 2014, 249, 84.
23. X. Xi, C. Ding, H. Zhang, X. Li, Y. Cheng and H. Zhang, J. Power Sources, 2015, 274, 1126.
24. H. Zhang, C. Ding, J. Cao, W. Xu, X. Li and H. Zhang, J. Mater. Chem. A, 2014, 2, 9524.
25. H. Zhang, H. Zhang, X. Li, Z. Mai and W. Wei, Energy Environ. Sci., 2012, 5, 6299.
26. H. Zhang, H. Zhang, X. Li, Z. Mai and J. Zhang, Energy Environ. Sci., 2011, 4, 1676.
27. W. Xu, X. Li, J. Cao, Z. Yuan and H. Zhang, RSC Adv., 2014, 4, 40400.
28. D. Chen and M. A. Hickner, Phys. Chem. Chem. Phys., 2013, 15, 11299.
29. S. Kim, T. B. Tighe, B. Schwenzer, J. Yan, J. Zhang, J. Liu, Z. Yang and M. A. Hickner, J. Appl. Electrochem., 2011, 41, 1201.
30. Z. Yuan, X. Li, Y. Zhao and H. Zhang, ACS Appl. Mater. Interfaces, 2015, 7, 19446.
31. Q. Li, J. O. Jensen, R. F. Savinell and N. J. Bjerrum, Prog. Polym. Sci., 2009, 34, 449.
32. K. Y. Wang, T.-S. Chung and J.-J. Qin, J. Membr. Sci., 2007, 300, 6.
33. S. C. Kumbharkar, P. B. Karadkar and U. K. Kharul, J. Membr. Sci., 2006, 286, 161.
34. M. Sadeghi, M. A. Semsarzadeh and H. Moadel, J. Membr. Sci., 2009, 331, 21.
35. J. A. Asensio, E. M. Sanchez and P. Gomez-Romero, Chem. Soc. Rev., 2010, 39, 3210.
36. J.-H. Kim, H.-J. Kim, T.-H. Lim and H.-I. Lee, J. Power Sources, 2007, 170, 275.
37. X. L. Zhou, T. S. Zhao, L. An, L. Wei and C. Zhang, Electrochim. Acta, 2015, 153, 492.
38. S. C. Kumbharkar and U. K. Kharul, J. Membr. Sci., 2010, 357, 134.
39. S. C. Kumbharkar and K. Li, J. Membr. Sci., 2012, 415, 793.
40. C. Sun, J. Chen, H. Zhang, X. Han and Q. Luo, J. Power Sources, 2010, 195, 890.
41. Y. Kang, J. Zou, Z. Sun, F. Wang, H. Zhu, K. Han, W. Yang, H. Song and Q. Meng, Int. J. Hydrogen Energy, 2013, 38, 6494.
42. H. Dai, H. Zhang, H. Zhong, H. Jin, X. Li, S. Xiao and Z. Mai, Fuel Cells, 2010, 10, 754.
43. X. Glipa, B. Bonnet, B. Mula, D. J. Jones and J. Rozière, J. Mater. Chem., 1999, 9, 3045.
44. R. Bouchet and E. Siebert, Solid State Ionics, 1999, 118, 287.
45. Q. F. Li, R. H. He, R. W. Berg, H. A. Hjuler and N. J. Bjerrum, Solid State Ionics, 2004, 168, 177.
46. Z. Mai, H. Zhang, X. Li, C. Bi and H. Dai, J. Power Sources, 2011, 196, 482.
47. A. Eguizábal, J. Lemus, M. Urbiztondo, O. Garrido, J. Soler, J. A. Blazquez and M. P. Pina, J. Power Sources, 2011, 196, 8994.
48. J. Lobato, P. Cañizares, M. A. Rodrigo, J. J. Linares and J. A. Aguilar, J. Membr. Sci., 2007, 306, 47.
49. R. He, Q. Li, A. Bach, J. Jensen and N. Bjerrum, J. Membr. Sci., 2006, 277, 38.
50. Z. Yuan, Y. Duan, H. Zhang, X. Li, H. Zhang and I. Vankelecom, Energy Environ. Sci., 2016, 9, 441–447.
51. E. Agar, K. W. Knehr, D. Chen, M. A. Hickner and E. C. Kumbur, Electrochim. Acta, 2013, 98, 66.
52. B. Li, Q. Luo, X. Wei, Z. Nie, E. Thomsen, B. Chen, V. Sprenkle and W. Wang, ChemSusChem, 2014, 7, 577.
53. Q. Luo, L. Li, W. Wang, Z. Nie, X. Wei, B. Li, B. Chen, Z. Yang and V. Sprenkle, ChemSusChem, 2013, 6, 268.
54. J. Sun, X. Li, X. Xi, Q. Lai, T. Liu and H. Zhang, J. Power Sources, 2014, 271, 1.

---