Preparation of silica nanocomposite anion-exchange membranes with low vanadium-ion crossover for vanadium redox flow batteries

Electrochimica Acta 105 (2013) 584–592

Contents lists available at SciVerse ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Preparation of silica nanocomposite anion-exchange membranes with low vanadium-ion crossover for vanadium redox flow batteries P.K. Leung, Q. Xu, T.S. Zhao ∗ , L. Zeng, C. Zhang Department of Mechanical Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China

a r t i c l e

i n f o

Article history: Received 6 October 2012 Received in revised form 28 April 2013 Accepted 30 April 2013 Available online 20 May 2013 Keywords: Flow battery Vanadium redox flow battery Anion exchange membrane Silica sol–gel reaction

a b s t r a c t Crossover of vanadium ions through the membranes of all-vanadium redox flow batteries (VRFB) is an issue that limits the performance of this type of flow battery. This paper reports on the preparation of a sol–gel derived silica nanocomposite anion exchange membrane (AEM) for VRFBs. The EDS and FT-IR characterizations confirm the presence and the uniformity of the silica nanoparticles formed in the membrane via an in situ sol–gel process. The properties of the obtained membrane, including the ion-exchange capacity, the area resistance, and the water uptake, are evaluated and compared to the pristine AEM and the Nafion cation exchange membrane (CEM). The experimental results show that the permeability of the vanadium ions through the silica nanocomposite AEM is about 20% lower than that of the pristine AEM, and one order of magnitude lower than that of the Nafion CEM. As a result, the rates of self-discharge and the capacity fading of the VRFB are substantially reduced. The Coulombic and energy efficiencies at a current density of 40 mA cm−2 are, respectively, as high as 92% and 73%. © 2013 Elsevier Ltd. All rights reserved.

1. Introduction

maintain the charge balance and the stoichiometry as:

The redox flow battery (RFB) is considered to be an attractive energy storage device for load-leveling and other off-grid applications, such as integration with renewable energy sources [1–4], as it offers obvious advantages in cost, cycle-life, safety and flexibility compared to other energy storage technologies. Among different types of RFB, the all-vanadium redox flow battery (VRFB) is the most studied system and has been demonstrated globally at large scale. Typical charge–discharge reactions of an VRFB involve two vanadium redox couples, V(II)/V(III) and V(IV)/V(V), in the negative and positive half-cells, respectively. In a fashion similar to most batteries, electrons are transferred between the two electrodes through the external circuit during the charge and discharge processes [5]. At the negative electrode, the redox reaction between V(III) and V(II) during charge and discharge are:

VO2+ + H2 O − e−

V3+ + e−

Charge



V2+

Discharge

E−VE = −0.26 V vs. SHE

(1)

Similarly, V(V) and V(IV) active species as a form of VO2 + and are reduced and oxidized at the positive electrode. Water molecules and protons are involved in the cathodic reaction to VO2+

∗ Corresponding author. Tel.: +852 2358 8647. E-mail address: [email protected] (T.S. Zhao). 0013-4686/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.electacta.2013.04.155

Charge



VO2 + + 2H+

Discharge

E+VE = +1.0 V vs. SHE (2)

In a VRFB, the ion exchange membrane is a key component that not only provides an ionic conduction pathway between the two electrolytes but also prevents mixing of the negative and positive electrolytes. The crossover of ions through the membrane will result in self-discharge and thus the loss of the chemical energy. As reported elsewhere [6], the self-discharge of VRFB is mainly associated with the diffusion of vanadium ions from one half-cell to the other due to the concentration gradients between the two electrolytes. For instance, V(II) and V(III) ions diffused from the negative half-cell can reduce V(IV) and V(V) ions in the positive electrolyte, while the V(II) and V(III) ions in the negative half-cell are oxidized by V(IV) and V(V) ions from the positive side. Crossover of vanadium ions gives rise to a reduction in the Coulombic efficiency and an imbalance of state of charge (SOC) between the two half-cell electrolytes, resulting in a capacity fading after prolonged cycling. State-of-art VRFBs use perfluorosulfonic cation exchange (CEM) membranes (such as Nafion). However, the high cost (up to 40% of the entire cell cost [7]) and significant crossover of vanadium ions through Nafion membranes are the two barriers that hinder the commercialization of VRFB [8]. Recent studies suggested that the anion exchange membrane (AEM) is an attractive candidate to replace Nafion as the AEM tends to have lower permeability of

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

vanadium ions than Nafion does due to the electrostatic repulsion between the positive functional groups of the membrane and the vanadium cations in the electrolyte. As proposed by some recent works [9], AEMs could transport both proton and sulfate anions in an acidic VRFB system. Compared to VRBs using Nafion membranes, the VRFBs equipped with several AEMs were reported to have improved Coulombic and energy efficiencies [10–15]. Despite this, most AEMs in the market are used for alkaline electrolytes. In order to ensure a long-term stability, a recent AEM designed for acidic VRFB system (Fumasep FAP, Fumatech GmbH, Germany) was used in this work, which costs (D 0.12 cm−2 [16]) only 20% of that of Nafion membranes (D 0.57 cm−2 [17]). The objective of this work is to minimize the crossover of vanadium ions by modifying an AEM via in situ sol–gel synthesis method. The silica nanoparticles serve as physical barriers to block the permeation of vanadium ions across the membrane as some vanadium cations may still penetrate through the AEM [18]. Prior to this work, commercial AEMs (Tokuyama Corp., Japan) have been modified via in situ polymerization [19,20] and sol–gel modified Nafion CEMs have been used in various applications, including VRFBs [18,21–23]. As reported elsewhere [18], the VRFBs equipped with the Nafion/silica membranes exhibit improved selfdischarge performance and show higher Coulombic and energy efficiencies than the system using Nafion CEM. In this work, a silica nanocomposite AEM was tested and compared with Nafion CEM by measuring the ion-exchange capacity (IEC), the area resistance, the water uptake and the vanadium ion permeability. It is demonstrated that the self-discharge and the capacity fading of the VRFB with the prepared membrane are substantially reduced. The energy efficiency of the VRFB in a typical charge–discharge cycle is as high as 73% at 40 mA cm−2 .

585

water, into the channel network of the AEM. Since it is miscible with water and the TEOS solution, an efficient sol–gel reaction can be ensured. The soaked membrane was then immersed in a solution (30 cm3 ) with 50 vol.% methanol and 50 vol.% 0.2 mol dm−3 ammonia (Aldrich, USA) for a specified period of time under vigorous stirring to initiate the hydrolysis/condensation reaction for the formation of the silica nanoparticles inside the AEM. A small amount of ammonia was used as the catalyst and to provide a slightly alkaline medium. After that, the resulting silica nanocomposite AEM was dipped immediately in anhydrous methanol for several times to remove the surplus reactants and then dried under ambient air for 24 h. Prior to any testing, the prepared silica nanocomposite AEM was stored in deionized water to establish certain level of hydration [18]. The silica content (w) of the nanocomposite AEM is obtained from: w=

mSiO2 , dry mAEM

=

mSiO2 /AEM − mAEM mAEM

× 100%

(3)

where mAEM , mSiO2 /AEM and mSiO2 , dry are the mass of the pristine AEM, silica nanocomposite AEM and the dry silica nanoparticles, respectively. The distribution of the silica content in the nanocomposite AEM was examined by Energy-Dispersive X-ray spectroscopy mapping (EDS-mapping, JEOL-6300F). Fourier transform infrared (FT-IR) spectra of the membranes were obtained using attenuated total reflection (ATR) accessory. The silica xerogel was prepared in a similar approach by mixing the same sol–gel solutions used in silica nanocomposite AEM fabrication. The solutions were 4 cm3 MeOH/0.2 mol dm−3 ammonia solution (1:1 volumetric ratio) and 6 cm3 MeOH/TEOS solution (1:1 volumetric ratio). With a small addition of ammonium fluoride (0.15 g) as a catalyst, a wet gel was formed within 30 min. The silica xerogel was hence formed by removing the water content through vacuum drying at 110 ◦ C for 24 h.

2. Experimental 2.2. Membrane characterization 2.1. Silica nanocomposite AEM preparation The method for preparing a silica nanocomposite AEM was based on the in situ sol–gel synthesis method developed by Mauritz et al. in the 1990s [18], which has been used and further modified in various works. In this work, an AEM used for acidic media was Fumasep FAP, which was purchased from Fumatech GmbH (Germany) and designed for the all-vanadium redox flow battery. For comparison purpose, Nafion 115 cation exchange membrane (Dupont, USA) was also used. As-received AEM was pretreated by first cleaning in a 3 wt.% H2 O2 solution (Aldrich, USA) at 70–80 ◦ C for 30 min. Subsequently, the membrane was immersed in a 1 mol dm−3 H2 SO4 solution (Sigma–Aldrich, USA) at the same temperature for another 30 min and further rinsed with deionized water several times to remove any remaining trace of H2 O2 and H2 SO4 . It should be noted that the membrane may decompose when the temperature is higher than 80 ◦ C. Such pretreatment procedures are crucial to swell the membrane (linear swelling ratio c.a. 6.2%) and to allow easier absorption of the solutions during the fabrication process. Before the in situ sol–gel synthesis, the pretreated membrane was cut into the size of 4.5 cm × 2 cm (9 cm2 ) and the weight was measured accordingly. The silica nanocomposite AEM was prepared by a conventional in situ sol–gel approach using tetraethyl orthosilicate (TEOS, Si(OCH2 CH3 )4 ) (Sigma–Aldrich, USA) as a silicon alkoxide precursor to react with water in methanol solvent (Aldrich, USA). For the preparation of the silica nanocomposite AEM, the pretreated AEM was soaked in a 30 cm3 MeOH/TEOS solution (1:1 volumetric ratio) for 12 h. The use of methanol is to facilitate the permeation of the TEOS solution and the subsequent

2.2.1. Ion-exchange capacity The ion-exchange capacity (IEC) of the AEMs was measured by following the procedures reported elsewhere [10]. The dried AEM was first cut into the size of 1 cm × 1 cm and then accurately weighted. The AEM was initially immersed in 1 mol dm−3 KOH solution (Sigma–Aldrich, USA) for 48 h to exchange into OH− form, which was then immersed in a 0.05 mol dm−3 HCl solution (30 cm3 ) for another 48 h. The amount of the OH− exchanged by the AEM would further neutralize with the H+ ions available in the 0.05 mol dm−3 HCl solution. To determine the amount of OH− exchanged by the AEM, the HCl solution for soaking the membrane was back titrated with 0.05 mol dm−3 KOH solution using phenolphthalein (Sigma–Aldrich, USA) as a colorimetric indicator. The IEC value (mmol g−1 ) was then calculated as the ratio of the titrated amount of OH− to the weight of the dried membrane. Similarly, the IEC of the Nafion CEM was measured by immersing a dried membrane in a 30 cm3 0.05 mol dm−3 KOH solution first, followed by a back titration with 0.05 mol dm−3 HCl solution. 2.2.2. Area resistance The area resistance of the membrane was measured in a conventional method described elsewhere [18–24]. A conductivity cell with two compartments separated by a membrane was used, in which 4 mol dm−3 H2 SO4 solution (20 cm3 ) was filled in each compartment. The area of the membrane exposed to the solution, S, was 4.5 cm × 2 cm, while the two graphite electrodes were placed at a distance of 2 cm from each other. The ionic resistances of the conductivity cell with and without membrane, were determined by the real axis intercept of the Nyquist plot using electrochemical

586

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

impedance spectroscopy (EG&G Princeton, model 273A and model 1025) at frequency range from 100 KHz to 1 Hz with a 10 mV amplitude. The area resistance, r, is obtained from: r = (rwith membrane − rwithout membrane ) × S

(4)

2.2.3. Water uptake The water uptake was measured as the mass ratio of the absorbed water to that of the dry membrane based on: W.U. =

WW − WD × 100% WD

(5)

where W.U. is the mass of water absorbed in the membrane, while WW and WD represent the mass of the membrane in wet and dry conditions, respectively. It should be noted that the wet membrane has been immersed in deionized water for 48 h. Before any measurement, the water on the membrane surface was wiped by an absorbent paper. 2.2.4. Permeability of vanadium ions The permeability of vanadium ions across the membrane was evaluated using an in-house dialysis cell, which is similar to the conductivity cell mentioned in Section 2.2.2. The cell was separated into two compartments by a membrane with the effective area of 4.5 cm × 2 cm. One compartment was filled with 20 cm3 1.0 mol dm−3 vanadium sulfate (V(II), V(III), V(IV) and V(V) species) (ZhongTian Chemical Ltd., China) in 3.0 mol dm−3 sulfuric acid, while the other compartment was filled with 20 cm3 1.0 mol dm−3 magnesium sulfate (Fluka, USA) in 3.0 mol dm−3 sulfuric acid. The use of magnesium sulfate is to reduce the osmosis effect of the anions by incorporating some sulfate anions (SO4 2− ) in the deficiency side. In addition, some drops of oil were added to the vanadium sulfate solution to avoid the oxidation of vanadium species with air. In this experiment, we assume that the change in vanadium ion concentration in the vanadium enrichment compartment is negligible and a pseudo-steady state condition is used in the membrane. At a regular time interval, samples in the magnesium sulfate compartment were taken and analyzed for vanadium ion concentration using inductively coupled plasma mass spectrometry (ICP-MS). 2.2.5. VRFB single cell A VRFB single cell was charge–discharge cycled using a computer controlled battery test system (CT3008W, Neware Inc., China). The in-house designed flow battery is similar to those reported elsewhere [25,26]. The single cell consisted of two acrylic flow channels (5 mm thickness), four silicone gaskets (1 mm thickness) as well as the negative and positive electrodes separated by an ion-exchange membrane. The both negative and positive electrodes were made up of two pieces of carbon felts (6 mm thickness pre-compressed, Sigratherm® GFA-05, SGL Carbon, Germany), which were compressed to the membrane by clamping the graphite substrates (50 mm × 70 mm × 6 mm) on both sides. Electrolytes of 20 cm3 were fed into the compartments using the acrylic flow channels and were circulated to and from the reservoirs at 3 cm3 s−1 (mean linear flow velocity 3 cm s−1 ) using a 2-channel peristaltic pump (WT-600-2J, Longerpump, China). The electrolytes used in the negative compartment was 1.0 mol dm−3 vanadium(III) sulfate in 3.0 mol dm−3 sulfuric acid, while that used in the positive compartment was 1.0 mol dm−3 vanadium(IV) sulfate in 3.0 mol dm−3 sulfuric acid. The positive electrolyte was simply prepared by dissolving vanadyl sulfate powder (ZhongTian Chemical Ltd., China) in sulfuric acid solution, while the negative electrolyte was prepared initially as a form of vanadyl sulfate (V(IV) species), which was further electrochemically reduced to vanadium(II) sulfate by applying

a current density of 20 mA cm−2 until the electrolyte turned into a deep purple color in an electrolyzer equipped with platinized titanium mesh electrodes. Such vanadium solution was further oxidized to vanadium(III) species resulting as the starting electrolyte for the flow battery experiment. The operating conditions of the electrolysis process were at room temperature (c.a. 22.5 ◦ C) and under electrolyte flow rate of 3 cm3 s−1 . To avoid the oxidation of the vanadium(II) species with air, electrolytes were first boiled at 80 ◦ C to remove the dissolved oxygen and a few drops of oil were employed as a sealing agent [22]. Pure nitrogen was purged to the negative electrolyte using a bubble dispenser continuously throughout the experiment. During cycling, the upper and lower voltage limits were controlled at 2.0 V and 0.8 V, respectively. Coulombic, voltage and energy efficiencies of the VRFB are obtained from: Coulombic efficiency =

Voltage efficiency = and Energy efficiency =

Id td × 100% Ic tc

Vd × 100% Vc

 V I t dt  d d d × 100% Vc Ic tc dt

(6)

(7)

(8)

where V is the cell voltage and I is the applied current density, c and d denote the charge and discharge processes, respectively. 3. Results and discussion 3.1. Silica nanocomposite AEM fabrication In this work, the silica nanocomposite AEM was prepared via the in situ sol–gel reaction of TEOS and water. When TEOS reacts with water, a continuous crosslinking network of silicon oxide was formed and spanned across the absorbed sol–gel solution (TEOS/MeOH/water) within the membrane structure. With a longer reaction time, a more compact silicon oxide network is expected. Hence, the silica content of the nanocomposite AEMs can be controlled by varying the sol–gel reaction time. Fig. 1a shows the variation in silica content of the membrane prepared with the reaction time. The silica content increased rapidly in the first 6 min and remained steadily at about 7.5 wt.%. Since the ammonia catalyst was used to facilitate the sol–gel reaction, the reaction took place at a faster rate than the other works reported without catalysts [23]. However, the silica content of the silica nanocomposite AEM (c.a. 7–7.5 wt.%) was lower than the values of the Nafion/silica membrane (>10 wt.%) as reported elsewhere [18–23]. The discrepancy can be attributed to the differences in membrane structures and the compositions of the sol–gel solutions. As reported by Varcoe et al. [27], AEMs tend to have lower uptake of methanol and water than Nafion CEM, which can limit the amount of sol–gel solution absorbed in the membrane for silica formation. The mass density of the silica nanosized network should be comparable to those of typical silica xerogels (<0.65 g cm−3 ) [28,29], which were fabricated using a similar experimental approach. Assuming that such SiO2 nanosized network has lower density than water (1.0 g cm−3 ) [28,29], the silica weight content of the nanocomposite membrane should not be larger than its water uptake value (22%), which is the mass content of water in a soaked membrane. The properties, including thickness, water uptake, ion-exchange capacity, area resistance of the silica nanocomposite AEM are compared with those of the pristine AEM and Nafion in Table 1. It can be seen that the water uptake and ion-exchange capacity of the pristine AEM are similar to those of the Nafion CEM. However,

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

587

Table 1 Comparison of the general properties between silica nanocomposite AEM (30 s, 1 min sol–gel reactions), pristine AEM and Nafion CEM. Membrane

Thickness (␮m)

Silicon content (wt.%)

Area resistance (m cm2 )

Water uptake (wt.%)

IEC (mmol g−1 )

AEM Sol–gel 30 s AEM Sol–gel 1 min AEM Nafion 115 CEM

60 60 60 127

5.60 6.46 0 0

1088 1510 700 987

20 20 22 26

1.13 1.07 1.16 0.91

after silica modification, a slight decrease in these two values is observed due to the filling of silica nanoparticles in the polar clusters. Due to the fact that the thickness of the pristine AEM (60 ␮m) is nearly half of that of the Nafion CEM (125 ␮m), the area resistance of the AEM (700 m cm2 ) is even lower than the Nafion CEM one (987 m cm2 ) in a typical VRFB electrolyte. With a longer sol–gel reaction time, the area resistance of the silica modified AEM increases as shown in Fig. 1b. The area resistance of the modified AEM increased and stabilized at about 1500 m cm2 after 3 min of sol–gel reaction. Therefore, the modified membrane used for VRFB testing was prepared for 30 s to yield a reasonable area resistance of about 1100 m cm2 , which is comparable to those of the Nafion 115 CEMs (987 m cm2 in this work) and in the literature (c.a. 1000 m cm2 [22,30,31]). In order to confirm the formation of the silica nanoparticles during the sol–gel reaction, FT-IR and EDS were used to characterize the presence of the silica network and the silica distribution in the membrane, respectively. Fig. 2 illustrates the FT-IR spectra of the silica modified AEM (c.a. 7.2 wt.% by 10 min sol–gel reaction), a pristine AEM and a pure silica xerogel, which were determined using attenuated total reflection (ATR) accessory. Although the strong

absorption bands of the AEM dominate the region of spectrum between 760 and 1600 cm−1 , the distinct peaks at 950 cm−1 and 1050 cm−1 in both spectra of the silica nanocomposite AEM and the pure silica xerogel confirms the nanostructure of the silicate network. The FT-IR spectra of the silica xerogel is particularly similar to that obtained by Muralidharan et al. [32]. The FT-IR absorption band at 800 cm−1 and 1050 cm−1 represents the asymmetric stretching vibrations of the Si O Si groups, while the peak at 950 cm−1 corresponds to the Si OH group in the silica nanostructure [32]. Fig. 3a shows the SEM image of the surface morphology of the resulting membrane, in which the distribution of such silicate nanoparticles was further examined by EDS. As shown in Fig. 3b, the bright red dots in the EDS mapping image indicate that the Si element spreads uniformly throughout the membrane, which suggests that the sol–gel reaction of TEOS and water can take place throughout the channel networks of the AEM. 3.2. Permeability of vanadium ions The permeability of the four vanadium ions at different oxidation states: V(II), V(III), V(IV) and V(V), across the silica nanocomposite AEM, the pristine AEM and the conventional Nafion CEM, were measured individually with the dialysis cell as described in Section 2.2.4. The silica nanocomposite AEM was prepared under sol–gel reaction of 30 s, which was reported to have an area resistance of c.a. 1088 m cm2 in Section 3.1. The effective areas of all the membranes were 9 cm2 . After some time, the vanadium ions diffuse through the membrane toward the deficiency side under a concentration gradient. The permeability of vanadium ions is determined by: VD

dcD (t) P = A (cE − cD (t)) L dt

(9)

where P is the permeability of the vanadium ions tested (m2 s−1 ), VD is the volume of the deficiency side (m3 ), A is the effective area

% Reflectance

Silica nanocomposite AEM Pristine AEM Silica xerogel

600

800

1000

1200

1400

1600

-1

Wavenumber / cm Fig. 1. The silica content (a) and the area resistance (b) of the silica nanocomposite AEM obtained at different sol–gel reaction time.

Fig. 2. FT-IR spectra of (a) silica nanocomposite AEM, (b) pristine AEM and (c) pure silica xerogel.

588

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

0,12 Silica modified AEM Pristine AEM Nafion CEM

ln [CE/ (CE - CD)]

0,10 0,08 0,06 0,04 0,02 0,00 5

10

15

20

25

30

35

Time / h Fig. 4. Concentration change of V(IV) ion in the magnesium sulfate compartment across silica nanocomposite AEM, pristine AEM and Nafion CEM.

Fig. 3. SEM image and EDS mapping image of the surface morphology of the silica nanocomposite AEM: (a) SEM image and (b) EDS mapping for the element Si. (For interpretation of the references to color in the text, the reader is referred to the web version of this article.)

of the membrane (m2 ), L is the thickness of the membrane (m), cE and cD are the vanadium ion concentrations (mol dm−3 ) in the enrichment and deficiency sides, respectively. It is assumed that the volumes of the electrolytes in the both sides are constant and the change in vanadium ion concentration in the enrichment side is negligible. Eq. (10) can be rearranged to [33]: dcD (t) P =A dt VD L cE − cD (t)

CD −

P d(cE − cD (t)) =A VD L cE − cD (t)

0

ln

t dt

(11)

0

and



(10)

works of Teng et al. [18] and Sun et al. [33]. It is found that the permeability of vanadium ion decreases significantly with the use of AEM and further reduces with silica impregnation. With the use of silica nanocomposite AEM, the permeability of V(V) ion is decreased by up to 74% compared to that of using Nafion CEM. The lower permeability of the AEM can be attributed to the positive charged groups within the membrane, which tends to repel vanadium cations by mean of electrostatics. Further diffusion of vanadium ion was blocked by the silica nanoparticles within the channel network or hydrophilic clusters of the modified AEM. Similarly, the permeability of the other vanadium ions, V(II), V(III), V(V), was tested and is presented in Fig. 5a and b. Generally, the permeability of most vanadium species is about an order of magnitude lower than that using AEMs. Despite this, the permeability of vanadium species of the Nafion CEM has a different trend with those using AEMs. For instance, the permeability of vanadium ions across Nafion CEM was measured with a decreasing order of V(II) > V(IV) > V(V) > V(III), while that of the AEM and the modified AEM was V(II) > V(V) > V(IV) > V(III). The trend of the Nafion CEM is in agreement with the data reported by previous workers [18,33], while that of the silica modified AEM is similar to the results of Tang et al. [6] which used Selemion AMV membrane as an AEM. The difference in such permeability rates can be possibly due to the repellent effect of vanadium cations across the AEMs. After silica modification, the overall permeability of vanadium ions was further reduced by another 20% as shown in Fig. 5b. Due to the relatively high vanadium ion permeability of the Nafion CEM, the effect of silica impregnation can be more pronounced on a pristine Nafion CEM as reported elsewhere [18]. For instance, Vijayakumar et al. [23] reported that their silica modified Nafion CEM could reduce the diffusivity of V(IV) ion by about 50%.

cE cE − cD



3.3. Self-discharge experiment =A

P t VD L

(12)

As shown in Fig. 4, the change in V(IV) ion concentration across the three membranes are plotted as a form of ln(cE /(cE − cD )) vs. time, where the gradient of this linear relationship corresponds to the value of (PA/VD L). Therefore, the permeability of V(IV) was calculated and was found to be in a decreasing order of Nafion CEM (1.62 × 10−6 cm2 min−1 ) > pristine AEM (5.24 × 10−7 cm2 min−1 ) > silica modified AEM −7 (4.24 × 10 cm2 min−1 ). In this work, the value obtained for the Nafion CEM is in the same order with those reported in the

In the previous section, the permeation rates of the vanadium ions across the three membranes: silica modified AEM (30 s reaction), the pristine AEM and Nafion CEM, have been identified. The permeation of different ions has a significant effect on the self-discharge of the VRFB. In order to evaluate the degree of selfdischarge, an open-circuit voltage (OCV) measurement of a dialysis cell was employed. The dialysis cell was similar to the one used in Section 3.2. Negative electrolyte was 1.0 mol dm−3 V(II) sulfate in 3 mol dm−3 sulfuric acid, while the positive electrolyte was 1.0 mol dm−3 V(V) sulfate in 3 mol dm−3 sulfuric acid. Both electrolytes were over-charged in an electrolyzer to ensure that the

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

(Permeability / 10-7 ) /cm2 min-1

(a)

589

30 25

V(II) V(III) V(IV) V(V)

20 15 10 5 0

Pristine AEM

(Permeability 10-7 ) / cm2 min-1

(b)

Nafion CEM

10 Fig. 6. Comparison of the OCV values between dialysis cells with silica nanocomposite AEM, pristine AEM and Nafion CEM.

V(II) V(III) V(IV) V(V)

8 6 4 2 0

Silica nanocomposite AEM

Pristine AEM

Fig. 5. Comparison of the permeability of the four vanadium ions (V(II), V(III), V(IV) and V(V) ions) through the membrane between (a) the Nafion CEM and the pristine AEM (b) and between the silica nanocomposite AEM and the pristine AEM.

state-of-charge of the cell was at 100%, which resulted in deep purple and orange colors in the negative and positive electrolytes, respectively. The OCV can be expressed as a function of vanadium ion concentrations according to Nernst equation: Eeq = E +

cV(V) cV(II) c 2 +

RT H ln cV(III) cV(IV) F

(13)

After some time, the concentrations of the V(V) and V(II) ions decreases due to the permeation of vanadium ions across the membrane. In the positive half-cell, the diffusion of V(II) and V(III) ions from the negative half-cell can lead to the following self-discharge reactions [6]: V2+ + 2V5+ → 3V4+

(14)

V3+ + V5+ → 2V4+

(15)

and V2+ + V4+ → 2V3+

(16)

Similarly, V(IV) and V(V) ions from the positive side can react with the active species in the negative half-cell as [6]: V5+ + 2V2+ → 3V3+

(17)

V4+ + V2+ → 2V3+

(18)

V5+ + V3+ → 2V4+

(19)

Due to the depletion of the charged ions in both half-cells, the OCVs of a dialysis cell with the silica modified AEM, the pristine AEM and Nafion CEM decrease gradually with time as shown in Fig. 6. After a prolonged period of time, a rapid drop in OCV to c.a. 0.8 V is observed when the capacity of either each half-cell (or both) has been fully self-discharged. It can be seen that the OCV with the silica nanocomposite AEM maintained at more than 1 V for about 116 h, while those of the pristine AEM and the Nafion CEM are about 100 h and 74 h, respectively. This indicates that the self-discharge has been extended effectively by about 12% and 58%, respectively. The lower self-discharge rate of using the silica modified AEM is mainly due to its lower vanadium ion permeability as determined in the previous section (Section 3.2). This results show that the combination of AEM and silica nanoparticles is a promising strategy to reduce the permeation of vanadium ions and to extend the self-discharge time compared to the conventional approach of using Nafion CEM. 3.4. Performance of the VRFB single-cell The charge–discharge performance of the VRFB single cell with the silica nanocomposite AEM (30 s reaction), the pristine AEM and the conventional Nafion CEM, were evaluated under the experimental procedures as described in Section 2.2.5. The effective areas of both electrodes and membrane were ensured to be 4.5 cm × 2 cm (9 cm2 ). Pure nitrogen gas was purged to the negative electrolyte continuously to minimize the possible oxidation of vanadium species with air. Fig. 7 shows the charge–discharge curves of the VRFB with the aforementioned membranes at 40 mA cm−2 , in which the duration of a typical charge–discharge cycle was about 3 h. The Coulombic efficiencies of the Nafion CEM obtained in this work (i.e. 87% at 40 mA cm−2 ) are comparable to other works reported elsewhere [9,18,21,25,34–36]. Among the three membranes tested, the discharge capacity and the Coulombic efficiency of the VRFB were the highest with the silica nanocomposite AEM, primarily due to its lower permeability of vanadium ions as determined in Section 3.2. The discharge capacity (476 mA h) was about 7.5% higher than those of the pristine AEM and the Nafion CEM (c.a. 440 mA h). However, the voltage efficiency of the pristine AEM is similar to that of the Nafion CEM. The voltage efficiency of the silica nanocomposite AEM is found to be slightly lower than those of the pristine AEM and Nafion CEM due to the increased area resistance after silica modification. The system efficiencies of the VRFB with three kinds of membranes at

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

0

2.4

Discharge capacity / mA h 120 240 360 480 600

Charge capacity / mA h 0 120 240 360 480 600

2.2

Charge

Discharge

Cell potential / V

2.0 1.8 1.6 1.4

(a) 100 % Coulombic efficiency

590

90

80 Silica nanocomposite AEM Pristine AEM Nafion CEM

70

1.2 Nafion CEM Pristine AEM

1.0 .8 .6

60 10

Silica nanocomposite AEM

20

30

40

50

60

70

80

90

-2

Current density / mA cm 0

20

40

60

80

100 120 140 160 180

(b)

100

different current densities are summarized in Fig. 8a–c. Regardless of which membrane was used, the Coulombic efficiency of the VRFBs is observed to increase with current densities. This can be attributed to the shorter charge–discharge time achieved at higher current densities [6]. Owing to the fact that energy efficiency is the product of Coulombic and voltage efficiencies, a downward trend is observed with increased current density due to its significant change of voltage efficiencies (Fig. 8b). However, in practice, a VRFB may not discharge immediately after being charged. The energy loss during the standing by can be significant for conventional VRFBs. In this work, the discharge capacities (at 40 mA cm−2 ) of the charged vanadium flow batteries with the three membranes were evaluated after being charged at 40 mA cm−2 for 1.5 h and then standing by for 10 h. Such a long period of time allows more vanadium ions to permeate from one to the other, resulting in capacity loss. As shown in Fig. 9, membranes with low vanadium ion permeability, particularly AEMs, show much larger capacity than Nafion CEM. The resulting Coulombic and energy efficiencies of the silica modified AEM (Coulombic : 75%, energy : 60%) are higher than those of Nafion CEM (Coulombic : 55%, energy : 46%) and the pristine AEM (Coulombic : 66%, energy : 55%). As shown in Fig. 10, a gradual decrease in capacity during cycling is observed in the VRFBs with the three membranes. As reported by Teng et al. [6], capacity fading during cycling is mainly attributed to the differential rates of the gaseous evolutions and the different permeability of the vanadium ion, which can lead to an imbalance of state of charge (SOC) between the two half-cell electrolytes. Due to the lower permeability of vanadium ions, the capacity loss per cycle for the silica nanocomposite AEM was c.a. 3.1 mA h cycle−1 , which is still lower than those with Nafion CEM (c.a. 6.3 mA h cycle−1 ) and the pristine AEM (c.a. 3.3 mA h cycle−1 ) after 16–20 cycles. To evaluate the stability of these membranes in a longer term, each of the three membranes was soaked in a pure 1.5 mol dm−3 vanadium(V) solution for 120 h (c.a. 5 days). Apparently, the three membranes did not suffer from oxidation caused by the V(V) ions, as no V(IV) ion from the associated reduction reaction was detected in the solution by the UV–vis spectroscopy. Despite this, the permeability of both the pristine AEM and silica modified AEMs was observed to be higher after prolonged soaking in the V(V) electrolyte (Fig. 11). The permeability of V(IV) ion with the use of

Silica nanocomposite AEM Pristine AEM Nafion CEM

90

80

70

60 10

20

30

40

50

60

70

80

90

80

90

-2

Current density / mA cm

(c)

80 75

% Energy efficiency

Fig. 7. Charge–discharge curves of the VRFB equipped with silica nanocomposite AEM, pristine AEM and Nafion CEM at 40 mA cm−2 .

% Cell potential efficiency

Time / min

70 65 60

Silica nanocomposite AEM Pristine AEM Nafion CEM

55 50 10

20

30

40

50

60

70 -2

Current density / mA cm

Fig. 8. Effect of current density on (a) Coulombic efficiency, (b) voltage efficiency and (c) energy efficiency of the VRFB with silica nanocomposite AEM, pristine AEM and Nafion CEM, respectively.

silica nanocomposite AEM increases from 4.24 × 10−7 cm2 min−1 to 5.48 × 10−7 cm2 min−1 , while that with the pristine AEM increase from 5.24 × 10−7 cm2 min−1 to 5.78 × 10−7 cm2 min−1 . As suggested elsewhere [23], the increase in the permeability of the silica nanocomposite membranes after soaking in vanadium solutions can be attributed to several possibilities: (1) the SiO2 nanoparticles leaching out of the membrane; (2) the irreversible

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

591

0,10

(a) 1.70 Open circuit for 10 h

0,08 Nafion CEM Pristine AEM Silica nanocomposite AEM

1.60

ln [CE/ (CE - CD)]

Cell potential / V

1.65

1.55 1.50

0,06 0,04 Pristine AEM, before soaking SiO2-AEM, before soaking

0,02

1.45 0

2

4

6

8

10

Pristine AEM, after soaking SiO2-AEM, after soaking

0,00 5

10

Time / h

(b)

20

25

30

35

Time / h

1.4 Fig. 11. Permeability of V(IV) ions across the silica nanocomposite AEM and pristine AEM before and after soaking in 1.5 mol dm−3 V(V) solution for 120 h.

Discharge after 10 h open circuit 1.2

Cell potential / V

15

1.0 4. Conclusions

.8 .6 .4 .2 10.0

Nafion CEM Pristine AEM Silica nanocomposite AEM 10.2

10.4

10.6

10.8

11.0

11.2

Time / h Fig. 9. Cell voltage curves of the fully charged VRFBs equipped with silica nanocomposite AEM, pristine AEM and Nafion CEM under open circuit for 10 h following a constant current discharge at 40 mA cm−2 .

binding of the SiO2 material via water physisorption along the side walls of the channel; (3) the shrinkage of the SiO2 agglomerates caused by the condensation of the silica network in a highly acidic environment. However, the mechanisms of these issues need to be further investigated to ensure the high performance of this membrane for long term operation.

In this work, a silica nanocomposite AEM has been prepared via in situ sol–gel reaction, in which silicon nanoparticles were formed uniformly throughout channel networks of the membrane. The function of silica nanoparticles is to reduce the crossover of vanadium ions between the two compartments. The permeability of vanadium ions across the silica nanocomposite AEM was significantly lower than those of the pristine AEM and Nafion CEM. With the use of the silica nanocomposite AEM, the rates of self-discharge and capacity loss of the VRFB were substantially reduced. In a typical charge–discharge cycle, an energy efficiency of 73% was obtained at a current density of 40 mA cm−2 . Compared to the pristine AEM and Nafion CEM, this membrane was found to have the highest discharge capacity and Coulombic efficiency at a range of current densities (20–80 mA cm−2 ). Due to its low cost and reduced vanadium ion permeability, the silica nanocomposite AEM can be an alternative candidate for VRFB application. Furthermore, it should be mentioned that the long term chemical stability of the silica nanocomposite AEM and its associated aging properties in VRFB electrolyte need to be addressed in the future. Acknowledgements The work described in this paper was fully supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. 622712). References

Fig. 10. Discharge capacities of the VRFBs with silica nanocomposite AEM and Nafion CEM over 20 charge–discharge cycles at 40 mA cm−2 .

[1] C. Ponce de León, A. Frías-Ferrer, J. González-García, D.A. Szánto, F.C. Walsh, Redox flow cells for energy conversion, Journal of Power Sources 160 (2006) 716–732. [2] A.Z. Weber, M.M. Mench, J.P. Meyers, P.N. Ross, J.T. Gostick, Q.H. Liu, Redox flow battery: a review, Journal of Applied Electrochemistry 41 (2011) 1137–1164. [3] M. Skyllas-Kazacos, M.H. Chakrabarti, S.A. Hajimolana, F.S. Mjalli, M. Saleem, Progress in flow battery research and development, Journal of the Electrochemical Society 158 (2011) R55–R79. [4] P. Leung, X. Li, C. Ponce de León, L. Berlouis, C.T.J. Low, F.C. Walsh, Progress in redox flow batteries, remaining challenges and their applications in energy storage, RSC Advances 2 (2012) 10125–10156. [5] M. Skyllas-Kazacos, M. Rychcik, P.G. Robins, A.G. Fane, M.A. Green, New allvanadium redox flow cell, Journal of the Electrochemical Society 133 (1986) 1057–1058. [6] A. Tang, J. Bao, M. Skyllas-Kazacos, Dynamic modelling of the effects of ion diffusion and side reactions on the capacity loss for vanadium redox flow battery, Journal of Power Sources 196 (2011) 10737–10747.

592

P.K. Leung et al. / Electrochimica Acta 105 (2013) 584–592

[7] S. Eckroad, Handbook of Energy Storage for Transmission or Distribution Applications, Electric Power Research Institute Report 1007189, California, USA, 2002. [8] X.F. Li, H.M. Zhang, Z.S. Mai, H.Z. Zhang, I. Vankelecom, Ion exchange membranes for vanadium redox flow battery (VRB) applications, Energy & Environmental Science 4 (2011) 1147–1160. [9] D. Chen, M.A. Hickner, E. Agar, E.C. Kumbur, Selective anion exchange membranes for high coulombic efficiency vanadium redox flow batteries, Electrochemistry Communications 26 (2013) 37–40. [10] T. Mohammadi, M. Skyllas-Kazacos, Modification of anion-exchange membranes for vanadium redox flow battery applications, Journal of Power Sources 63 (1996) 179–186. [11] G.J. Hwang, H. Ohya, Crosslinking of anion exchange membrane by accelerated electron radiation as a separator for the all-vanadium redox flow battery, Journal of Membrane Science 132 (1997) 55–61. [12] G.J. Hwang, H. Ohya, Preparation of anion-exchange membrane based on block copolymers: Part 1. Amination of the chloromethylated copolymers, Journal of Membrane Science 140 (1998) 195–203. [13] J.Y. Qiu, M.Y. Li, J.F. Ni, M.L. Zhai, J. Peng, L. Xu, H.H. Zhou, J.Q. Li, G.S. Wei, Preparation of ETFE-based anion exchange membrane to reduce permeability of vanadium ions in vanadium redox battery, Journal of Membrane Science 297 (2007) 174–180. [14] D.B. Xing, S.H. Zhang, C.X. Yin, B.G. Zhang, X.G. Jian, Effect of amination agent on the properties of quaternized poly(phthalazinone ether sulfone) anion exchange membrane for vanadium redox flow battery application, Journal of Membrane Science 354 (2010) 68–73. [15] S.H. Zhang, C.X. Yin, D.B. Xing, D.L. Yang, X.G. Jian, Preparation of chloromethylated/quaternized poly(phthalazinone ether ketone) anion exchange membrane materials for vanadium redox flow battery applications, Journal of Membrane Science 363 (2010) 243–249. [16] Fumasep FAP, FuMA-Tech GmbH, http://www.fumatech.com/NR/rdonlyres/ 6B681BD0-C95C-4978-87D8-FA10C7794F3E/0/fumasepFAP.pdf [17] Nafion® perfluorinated membrane, Nafion® 115, http://www.sigmaaldrich. com/catalog/product/aldrich/541346?lang=en®ion=HK [18] X.G. Teng, Y.T. Zhao, J.Y. Xi, Z.H. Wu, X.P. Qiu, L.Q. Chen, Nafion/organically modified silicate hybrids membrane for vanadium redox flow battery, Journal of Power Sources 189 (2009) 1240–1246. [19] T. Sata, Properties of composite membranes formed from ion-exchange membranes and conducting polymers. 4. Change in membrane resistance during electrodialysis in the presence of surface-active agents, Journal of Physical Chemistry 97 (1993) 6920–6923. [20] T. Sata, T. Yamaguchi, K. Matsusaki, Preparation and properties of composite membranes composed of anion-exchange membranes and polypyrrole, Journal of Physical Chemistry 100 (1996) 16633–16640. [21] J.Y. Xi, Z.H. Wu, X.G. Teng, Y.T. Zhao, L.Q. Chen, X.P. Qiu, Self-assembled polyelectrolyte multilayer modified Nafion membrane with suppressed vanadium

[22]

[23]

[24]

[25]

[26] [27]

[28]

[29] [30]

[31]

[32]

[33]

[34]

[35]

[36]

ion crossover for vanadium redox flow batteries, Journal of Materials Chemistry 18 (2008) 1232–1238. X.G. Teng, Y.T. Zhao, J.Y. Xi, Z.H. Wu, X.P. Qiu, L.Q. Chen, Nafion/organic silica modified TiO2 composite membrane for vanadium redox flow battery via in situ sol–gel reactions, Journal of Membrane Science 341 (2009) 149–154. M. Vijayakumar, B. Schwenzer, S.W. Kim, Z.G. Yang, S. Thevuthasan, J. Liu, G.L. Graff, J.Z. Hu, Investigation of local environments in Nafion-SiO2 composite membranes used in vanadium redox flow battery, Solid State Nuclear Magnetic Resonance 42 (2012) 77–80. H.Z. Zhang, H.M. Zhang, X.F. Li, Z.S. Mai, W.P. Wei, Silica modified nanofiltration membranes with improved selectivity for redox flow battery application, Energy & Environmental Science 5 (2012) 6299. P. Qian, H.M. Zhang, J. Chen, Y.H. Wen, Q.T. Luo, Z. Liu, D.J. You, B.L. Yi, A novel electrode-bipolar plate assembly for vanadium redox flow battery applications, Journal of Power Sources 175 (2008) 613–620. P.K. Leung, C. Ponce-de-Leon, C.T.J. Low, A.A. Shah, F.C. Walsh, Characterization of a zinc-cerium flow battery, Journal of Power Sources 196 (2011) 5174–5185. J.R. Varcoe, R.C.T. Slade, E.L.H. Yee, S.D. Poynton, D.J. Driscoll, Investigations into the ex situ methanol, ethanol and ethylene glycol permeabilities of alkaline polymer electrolyte membranes, Journal of Power Sources 173 (2007) 194–199. C. Alié, F. Ferauche, R. Pirard, A.J. Lecloux, J.P. Pirard, Preparation of lowdensity xerogels by incorporation of additives during synthesis, Journal of Non-Crystalline Solids 289 (2001) 88–96. M.A. Einarsrud, Light gels by conventional drying, Journal of Non-Crystalline Solids 225 (1998) 1–7. Z. Mai, H.-M. Zhang, X.F. Li, C. Bi, H. Dai, Sulfonated poly(tetramethydiphenyl ether ether ketone) membranes for vanadium redox flow battery application, Journal of Power Sources 196 (2011) 5174–5185. N.F. Wang, S. Peng, D. Lu, S.-Q. Liu, Y.-N. Liu, K.-L. Huang, Nafion/TiO2 hybrid membrane fabricated via hydrothermal method for vanadium redox battery, Journal of Solid State Electrochemistry 16 (2012) 5174–5185. M.N. Muralidharan, C.A. Rasmitha, R. Ratheesh, Photoluminescence and FTIR studies of pure and rare earth doped silica xerogels and aerogels, Journal of Porous Materials 16 (2009) 635–640. C.X. Sun, J. Chen, H.M. Zhang, X. Han, Q.T. Luo, Investigations on transfer of water and vanadium ions across Nafion membrane in an operating vanadium redox flow battery, Journal of Power Sources 195 (2010) 890–897. P. Zhao, H.M. Zhang, H. Zhou, J. Chen, S. Gao, B. Yi, Characteristics and performance of 10 kW class all-vanadium redox-flow battery stack, Journal of Power Sources 162 (2006) 1416–1420. D. Chen, S. Wang, M. Xiao, Y. Meng, Preparation and properties of sulfonated poly(fluorenyl ether ketone) membrane for vanadium redox flow battery application, Journal of Power Sources 195 (2010) 2089–2095. D. Chen, S. Wang, M. Xiao, Y. Meng, Synthesis and characterization of novel sulfonated poly(arylene thioether) ionomers for vanadium redox flow battery applications, Energy & Environmental Science 3 (2010) 622–628.