Nafion blend membrane for direct methanol fuel cell (DMFC)

Nafion blend membrane for direct methanol fuel cell (DMFC)

Electrochimica Acta 50 (2004) 583–588 Characteristics of PVdF copolymer/Nafion blend membrane for direct methanol fuel cell (DMFC) Ki-Yun Choa , Ji-Y...

265KB Sizes 1 Downloads 31 Views

Electrochimica Acta 50 (2004) 583–588

Characteristics of PVdF copolymer/Nafion blend membrane for direct methanol fuel cell (DMFC) Ki-Yun Choa , Ji-Yong Eoma , Ho-Young Junga , Nam-Soon Choia , Yong Min Leea , Jung-Ki Parka,∗ , Jong-Ho Choib , Kyung-Won Parkb , Yung-Eun Sungb a

Department of Chemical and Biomolecular Engineering, Korea Advanced Institute of Science and Technology, 373-1, Guseong-Dong, Yuseung-Gu, Daejon 305-701, Korea b Department of Materials Science and Engineering, Kwangju Institute of Science and Technology, Gwangju 500-712, Korea Received 2 June 2003; received in revised form 29 March 2004; accepted 29 March 2004 Available online 20 August 2004

Abstract For direct methanol fuel cell, blends of vinylidene fluoride-hexafluoropropylene copolymer (P(VdF-co-HFP)) and Nafion were prepared the different equivalent weight of Nafion. The investigations of the blend morphology were performed by means of permeability test, uptake measurement, differential-scanning calorimetry (DSC), and scanning electron microscopy. In the blend membranes, many pores were created as the content of Nafion in blend increased. Then, the methanol uptake was sharply increased. But the methanol permeability was not sharply increased because the methanol permeation through blend membranes is diffusioncontrolled process. The methanol permeability of N10 (low equivalent weight) series was similar to that of N11 series (high equivalent weight). The proton conductivity of N10 series was around one and a half times higher than that of N11 series. The cell performance of the blend was much enhanced when the equivalent weight of Nafion was 1000. © 2004 Elsevier Ltd. All rights reserved. Keywords: DMFC; Solid polymer electrolytes; PVdF copolymer; Equivalent weight of Nafion; Methanol permeability

1. Introduction Polymer electrolyte membrane fuel cells are extremely attractive as power sources for the transportation, the distributed power, and the portable electrical devices. The most attractive materials for the membranes used in fuel cell are DuPont’s Nafion® and Dow’s membranes because they exhibited a number of desirable properties, such as a high ionic conductivity, mechanical strength, and chemical/thermal stability. However, they are responsible for the high cost of fuel cell and are permeable to methanol and hydrogen, which lowers fuel efficiency. These limitations have stimulated many ∗

Corresponding author. Tel.: +82 42 869 3925; fax: +82 42 869 3910. E-mail address: [email protected] (J.-K. Park).

0013-4686/$ – see front matter © 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2004.03.063

efforts in the development of alternative membrane materials [1–6]. J. Lin et al. attempted to blend Nafion with small amount of vinylidene fluoride-hexafluoropropylene copolymer (P(VdFco-HFP)) in order to reduce the methanol crossover and the cost with retaining essential proton conductivity [7]. Even though the amount of the Nafion content in the blends was large, it showed a sharp drop in the proton conductivity by around two orders of magnitude compared with native Nafion membrane. In this study, we prepared the blend membrane used a small amount of the Nafion content and investigated the effect of the equivalent weight (E.W.) on electrochemical properties of the PVdF copolymer/Nafion blend. In all case of the blends, it is expected that the PVdF copolymer becomes a

584

K.-Y. Cho et al. / Electrochimica Acta 50 (2004) 583–588

continuous phase and the Nafion forms a dispersed phase. Acetone was used for a solvent of PVdF copolymer at the temperature of 25 ◦ C.

2. Experimental 2.1. Materials P(VdF-co-HFP) with a vinylidene fluoride content of 85% (Kynar Flex 2751, Mn = 380,000) was supplied from Atofina Chemicals Inc. Nafion 117 membrane (1,100 E.W., sulfonic acid form) and 20 wt.% Nafion (1,100 E.W., and 1,100 E.W., sulfonic acid form) solutions, in isopropanol, n-propanol, and water mixture were purchased from E. I. Dupont de Nemours & Co. The Nafion 117 membrane was pretreated as the reported procedures [8] prior to use. In brief, the membrane was pretreated by boiling in 0.5 M H2 O2 , rinsing in boiling water, boiling in 0.5 M H2 SO4 , and rinsing in boiling water, finally. The Nafion solution was used as received. Acetone and methanol were purchased from Merck and used without further purification. 2.2. Preparation of blend membranes Solution blends of P(VdF-co-HFP) with the Nafion were prepared in a following procedure. At first, P(VdF-co-HFP) powder was fully dissolved in acetone and the appropriate amount of Nafion solution was added to this solution. Then, the blend of P(VdF-co-HFP) and the Nafion was vigorously stirred to be a homogeneous transparent solutions. The resulting solution was cast onto a flat glass surface with a doctor blade device gapped at 1 mm, then left in an oven at 25 ◦ C for about 10 h to remove the solvents. The obtained films were peeled from the plate in a deionized water bath. The notation of the prepared membranes is shown in Table 1. 2.3. Characterization 2.3.1. Measurements of proton conductivity All the prepared blends were kept in distilled deionized water at 80 ◦ C for 1 day for the complete swelling of waTable 1 The notation of fabricated samples Notation

Nafion solution

PVdF-co-HFP

Equivalent weight

wt.%

N109505 N109010 N108515 N108020

1000

5 10 15 20

95 90 85 80

N119505 N119010 N118515 N118020

1100

5 10 15 20

95 90 85 80

ter. Then, the hydrated blend membrane was sandwitched between two stainless steel electrodes (12 mm in diameter). The ionic conductivities of the films were measured by complex impedance analysis using a Solartron 1,255 frequency response analyzer coupled to an IBM PS/2 computer over a frequency range of 100 Hz–10 MHz. An AC pertubation of 10 mV was applied to the cell. The real and imaginary parts of the complex impedance were plotted and the ionic conductivity (σ) could be obtained from the bulk resistance found in complex impedance diagram. 2.3.2. Methanol permeability measurement Methanol permeability of the samples was measured by using a diffusion cell followed the design in the literatures [9]. The diffusion cell was consisted of two compartments that were separated by vertical membrane. One compartment of the cell was filled with a mixture solution of methanol and deionized water (5/95 (w/w)). The other compartment filled with deionized water was directly connected to a differential refractive index detector. The compartments were stirred continuously during permeability measurement. 2.3.3. Water uptake and methanol uptake The membranes were soaked for a day in deionized water and methanol to determine the uptake contents, respectively. Weight of a dried membrane was measured after drying the sample overnight in vacuum at 80 ◦ C. The uptake was calculated by water uptake (%) =

wwet − wdry × 100 wwet

where wdry is the mass of dried sample and wwet is the mass of wet sample, respectively. 2.3.4. Thermal analysis Differential-scanning calorimetric study was carried out using a TA instruments’ Differential Scanning Calorimeter 2010 at a heating rate of 5 ◦ C/min to characterize thermal properties of the blend membrane. Each sample was heated from −30 ◦ C to 30 ◦ C under nitrogen atmosphere. 2.3.5. Electron microscopy Morphologies of the blends were investigated by using scanning electron microscopy (Phillips SEM 535M). All the specimens were sputter-coated with gold to be visualized under scanning electron microscope and SEM micrographs were taken with the electric field strength of 20–30 kV. 2.3.6. Cell performance Cell performance was evaluated by using a DMFC unit cell with a 2 cm2 cross-sectional area and measured with a potentiometer (WMPG-1000) which recorded the cell potential from the circuit voltage under constant current conditions. The catalysts used at the anode and the cathode were applied on carbon paper (TGPH-090, Toray) by brushing. The catalyst loadings at the anode and the cathode based on

K.-Y. Cho et al. / Electrochimica Acta 50 (2004) 583–588

585

the catalyst weight without polymers were 3 and 5 mg/cm2 , respectively. 2 M methanol at the cathode, was supplied by a Masterflex liquid micro-pump at 1 mL/min while the cathode was fed with dried O2 at a rate of 500 mL/min by a flow meter.

3. Results and discussion 3.1. Methanol permeability The methanol permeability, which is the product of the diffusion coefficient and the sorption coefficient, is often used to describe the transport of a permeant through polymers. The diffusion coefficient is a kinetic term which shows the effect of the surrounding environment on the molecular motion of the permeant and the sorption coefficient is an equilibrium term correlating the concentration of a component in the fluid phase with its concentration in the membrane polymer phase [10]. Fig. 1 shows the methanol permeabilities for N10 and N11 series. The permeability increased gradually with the increase of Nafion content. It showed the difference of the two types of blends. The methanol permeability of the blend membrane would be suppressed remarkably by about one order, compared with that of the native Nafion membrane (2.32 × 10−6 cm2 /s). To investigate an equilibrium factor that affects the methanol permeability, methanol uptake experiment was carried out. Fig. 2 showed the plot of methanol uptake versus the Nafion content in the blends. The methanol uptake value of N10 and N11 series was increased rapidly with the increase of Nafion content in the blend membrane. However, the methanol uptake of N108020 (71 wt.%) was higher than that of the native Nafion (57 wt.%). It is doubtful that the methanol permeability was quite suppressed in the blends, even though the methanol uptake of the blend was higher than that of the native Nafion. It is plausible to say that methanol transport was mainly influ-

Fig. 1. Methanol permeabilities of N10 and N11 series with variation of blend composition and that of the Nafion.

Fig. 2. Methanol uptake of N10 and N11 series with variation of blend composition.

enced by the kinetic factor instead of the equilibrium factor in the blends. Therefore, the methanol permeation through the PVdF copolymer/Nafion blends was diffusion-controlled process. 3.2. Proton conductivity Fig. 3 shows the proton conductivity of fully hydrated blend membranes at room temperature. Proton conductivity for two blend types sharply increased with the increase of Nafion content in the blend membrane. The proton conductivity of N10 series was higher than that of N11 series because the N10 series have higher number of charge carriers, which are the sulfonic acid groups than that of the N11 series [11]. And it was remarkable that the conductivity of N108020 was 1.52 × 10−3 S/cm although the Nafion content in blend was only 20 wt.%. The native Nafion 117 membrane showed the proton conductivity of 1.78 × 10−2 S/cm. Considering their quite low values of the methanol permeability, it showed the possibility of application for DMFC.

Fig. 3. Proton conductivity of N10 and N11 series with variation of blend composition and that of the Nafion.

586

K.-Y. Cho et al. / Electrochimica Acta 50 (2004) 583–588

Fig. 4. Water uptake of N10 and N11 series with variation of blend composition and that of the Nafion.

3.3. Water uptake Water uptake was defined as the weight ratio of the water absorbed by the blend film to the dried film. Fig. 4 showed the water uptake for N10 and N11 series. The water uptake of N10 and N11 series increased with increasing the Nafion content. The water uptake generally showed lower value comparing with the methanol uptake. It might be originated from the decreased compatibility between water and PVdF copolymer in comparison with methanol as expected from difference in solubility parameters.

The water uptake of a sulfonated polymer membrane at a given temperature increased with the degree of sulfonation and it strongly influenced their proton conductivity. The native Nafion membrane had more sulfonic acid groups than that of N108020. And the proton conductivity of the native Nafion was higher than the proton conductivity of N108020. However, the water uptake (26.4 wt.%) of N108020 was higher than that of the native Nafion membrane (24.7 wt.%). In the blends, the degree of water absorption, on a mass basis, does not correlate well with proton conductivity. Because of the proton conductivity through the PVdF copolymer/Nafion blends might be diffusion-controlled process like the methanol permeation. It was because the state of water in blends membrane could be different from that of the native Nafion membrane. The fully hydrated membranes contained the water of three different types associated with polymer: (1) non-freezing water bound to the sulfonic acid groups, (2) freezable bound water weakly bound to the ionic groups and the polymer matrix and (3) free water that is not intimately bound to polymer chain and behaves like bulk water [12]. The water uptake in the membranes must be the following two factors. One was related to the increase in osmotic driving force for water sorption with the increase of the Nafion content (non-freezing water). The other is related to the volume portion of the pores generated in fabrication procedure (free water). According to the SEM images of the surface of N10 series shown in Fig. 5, the N10 series using

Fig. 5. SEM images of N10 series with variation of blend composition and that of the Nafion.

K.-Y. Cho et al. / Electrochimica Acta 50 (2004) 583–588

Fig. 6. DSC thermograms of N10 series with variation of blend composition and that of the Nafion.

acetone exhibited a porous and coarse surface morphology. However, the native Nafion membrane exhibited a compact and smooth surface. In the blends, the water in the pore was not intimately bound to polymer chain and behaves like bulk water. So the ratio of free water might increase. Thermograms showing the melting endotherms of water in the samples were shown in Fig. 6. Nafion did not show bimodal melting peak. This meant the freezable water was not detected. However, the N10 series showed broad bimodal melting peak from −10 ◦ C to 5 ◦ C. These peaks were originated to the freezable water [13]. With the increase of Nafion content in the blend membranes, the blend membranes had many pores in the SEM images. The presence of freezable water might be due to pores. The freezable water was related with the methanol permeability in proton exchange membranes [14]. Although the content of the freezable water in the blend membranes was higher than that of the native Nafion, the permeability of the blend membranes was lower than the permeability of the native Nafion. It was because the freezable water in blend membranes existed in the pore. The pores might be not interconnected well. Therefore, the methanol permeability and the proton conductivity of the blend membranes were lower than that of the native Nafion although the methanol uptake and water uptake of the blend membranes were high compared with that of the native Nafion. Then, the methanol and proton transport may be mainly influenced by the kinetic factor instead of the equilibrium factor in the blend membranes. 3.4. Cell performance The polarization test of the DMFC for N108020, N118020 and Nafion was shown in Fig. 7 for 2 M methanol solution at 30 ◦ C. We coated the samples with Nafion solution to enhance the interfacial compatibility of the membrane with the electrode. The same hot-pressing conditions for the fair per-

587

Fig. 7. Polarization curves of N108020 and N118020 at 30 ◦ C and 2 M methanol from anode while the cathode was fed with dry O2 at a rate of 500 mL/min.

formance of the native Nafion 117 were applied for the blend membranes. In this experiment, the cell performance of the blend membrane was lower than that of the native Nafion 117 and the blend membranes exhibited a steep ohmic loss because higher bulk resistance and poor physical contact made such a steep decrease in the cell potential in comparison with the native Nafion 117. The performance of N108020 was superior to that of N118020. The proton conductivity of N108020 (1.52 × 10−3 ) was higher than that of N118020 (5.86 × 10−3 S/cm). However, the methanol permeability of N108020 (1.24 × 10−7 cm2 /s) was higher than that of N118020 (9.14 × 10−7 cm2 /s). Below the proton conductivity of Nafion, the cell performance might be predomintantly dependent on the proton conductivity.

4. Conclusions In this study, the blend membranes of P(VdF-co-HFP) with Nafion were prepared with the different equivalent weight of Nafion. The proton conductivity of the blend membranes was lower than that of the native Nafion. However, the methanol permeability of the blend membranes was lower than that of the native Nafion. In the blend membranes, the methanol and proton transport did not be mainly influenced by the equilibrium factor but by the kinetic factor in the blends. The water uptake and methanol uptake were sharply increased with the increase of the content of Nafion but the methanol permeability and proton conductivity were not sharply increased. The uptakes increased because the generation of pores with increasing the content of Nafion. However, the methanol and the water in the pores were not well correlated with the methanol permeability and proton conductivity. Although the small amount of the content of Nafion, the proton conductivity was good and the methanol permeability was suppressed in the blend membranes. Then, the cell per-

588

K.-Y. Cho et al. / Electrochimica Acta 50 (2004) 583–588

formance of the blend membrane was smiliar to that of the native Nafion. Acknowledgement This work was supported by the Brain Korea 21 Project. Y.-E. Sung acknowledges support by KOSEF through the Research Center for Energy Conversion and Storage. References [1] L. Leung, C. Bailly, J.F. O’Gara, D.J. Williams, F.E. Karasz, W. MacKnight, J. Polym. Commun. 28 (1987) 20. [2] S.M.J. Zaidi, S.D. Mikhailenko, G.P. Robertson, M.D. Guiver, S. Kaliaguine, J. Membr. Sci. 185 (2001) 73. [3] C. Genies, R. Mercier, B. Sillion, N. Cornet, G. Gebel, M. Pineri, Polymer 42 (2001) 359.

[4] R. Nolte, K. Ledjeff, M. Bauer, R. Mulhaupt, J. Membr. Sci. 83 (1993) 211. [5] F. Wang, M. Hickner, Y.S. Kim, T.A. Zawodzinski, J.E. McGrath, J. Membr. Sci. 197 (2002) 231. [6] F. Wang, J. Mecham, J.W. Harrison, J.E. MeGrath, Polym. Prepr. 40 (2000) 180. [7] J. Lin, M. Ouyang, J.M. Fenton, H.R. Kunz, J.T. Koberstein, M.B. Cutlip, J. Appl. Polym. Sci. 70 (1998) 121. [8] E.A. Ticianelli, C.R. Derouion, A. Redondo, S. Srinicasan, J. Electrochem. Soc. 135 (1998) 2209. [9] V. Tricoli, J. Electrochem. Soc. 145 (1998) 3798. [10] Z. Ogumi, T. Kuroe, Z. Takehara, J. Electrochem. Soc. 1322 (1985) 2601. [11] V. Tricoli, J. Electrochem. Soc. 145 (1998) 3798. [12] G. Pourcelly, A. Oikonomow, H.D. Hurwitz, Electroanal. Chem. 287 (1990) 43. [13] L.E. Karlsson, B. Wesslen, P. Jannasch, Electrochimica acta 47 (2002) 3269. [14] Y.S. Kim, L. Dong, M.A. Hickner, T.E. Glass, V. Webb, J.M. McGrath, Macromolecules 36 (2003) 6281.