Materials Chemistry and Physics 114 (2009) 151–155

Contents lists available at ScienceDirect

Materials Chemistry and Physics journal homepage: www.elsevier.com/locate/matchemphys

Characterization of Teflon-like carbon cloth prepared by plasma surface modification for use as gas diffusion backing in membrane electrode assembly Chih-Ming Lee a , Yi-Hao Pai b , Jyh-Myng Zen c , Fuh-Sheng Shieu a,∗ a

Department of Materials Engineering, National Chung Hsing University, Taichung 402, Taiwan Institute of Photonics and Optoelectronics, and Department of Electrical Engineering, National Taiwan University, Taipei 106, Taiwan c Department of Chemistry, National Chung Hsing University, Taichung 402, Taiwan b

a r t i c l e

i n f o

Article history: Received 15 June 2008 Received in revised form 27 August 2008 Accepted 29 August 2008 Keywords: Plasma surface modification Gas diffusion backing Hydrophobic property Proton exchange membrane fuel cell

a b s t r a c t The hydrophobic property of carbon cloth was largely improved by plasma treatment and the Teflon-like property was effectively applied to fabricate a gas diffusion backing (GDB) for use in membrane electrode assembly (MEA). The surface morphology, hydrophilic/hydrophobic property and electron conductivity of the as-prepared GDB was fully characterized. The water contact angle and SEM microstructure image of the CF4 , CHF3 plasma-treated GDB were both indicated as ∼130◦ , and very few gas diffusion pores either sealed or blocked by excessive hydrophobic material residual. The measured resistivity values of CF4 plasma, CHF3 plasma, SF6 plasma and commercial carbon were 0.45, 0.5, 0.47 and 0.49, respectively, which indicates that the electrical resistivity of carbon cloth with CF4 plasma treatment was slightly lower than others. In cell performance test, the CF4 plasma-treated modules could also produce better property than those MEAs prepared with CHF3 plasma-treated GDB, SF6 plasma-treated GDB and commercially available GDB, leading to the highest fuel cell performance with an optimal power output of 350 mW cm−2 . © 2008 Elsevier B.V. All rights reserved.

1. Introduction Proton exchange membrane fuel cell (PEMFC) is a promising mainstream green power source for mobile and stationary applications. The porous gas diffusion layer (GDL), which can affect the diffusion of reactants and water as well as the conductivity, plays a crucial role among all components inside the PEMFC. The GDL consists of two parts, i.e., macro-porous gas diffusion backing (GDB) and micro-porous layer (MPL) [1]. Carbon cloth [1–3] and carbon paper [4,5] are the most common used macro-porous GDB in GDL to serve as current collector and physical support structure for the catalyst layer (CL). The function of GDB can be summarized as follows: (1) to provide a proper pore structure and hydrophobicity to allow for a better gas transport and water removal from the electroosmotic drag through the electrolyte membrane [6,7], (2) to condense water vapor from the humidified reactant feed [8–10], (3) to direct involve in the reduction reaction at the cathode [8] and (4) to minimize electric contact resistance with the adjacent CL. An ideal GDB should therefore be highly conductive, highly porous and wet-proofed [11,12]. It can then provide an efficient pathway for the gas reactants to be fed homogeneously and the produced

∗ Corresponding author. Fax: +886 4 2285 7017. E-mail address: [email protected] (F.-S. Shieu). 0254-0584/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.matchemphys.2008.08.092

liquid water to be removed rapidly without blocking the reactant gases [9,12]. An effective wet-proofing procedure manipulates the hydrophobic property of the porous GDB and subsequently affects the overall power output performance of the PEMFC. Typically, the wet-proofed GDBs are prepared by coating with hydrophobic materials (e.g., polytetrafluoroethylene (PTFE)) [13–15] or fluorinated ethylene propylene (FEP) [13,16,17] onto a raw GDB (e.g., carbon paper or carbon cloth) [12,18]. This treatment can lead to a large contact angle between liquid water and the coated GDB and so the porous channels inside the wet-proofed GDB are less blocked or flooded by the produced liquid water [18]. However, the surface characterization of the GDB depends not only on the loading of hydrophobic materials but also on the techniques utilized to coat the hydrophobic materials onto the GDB [15,19]. For example, Jaszewski et al. reported that the anti-adhesive property was very different for PTFE-like film prepared either by plasma polymerized or ion sputtered methods [20]. Ioroi et al. investigated the influence of PTFE loading on GDB and found a better performance with smaller PTFE loading. Nevertheless, a very poor fuel cell performance was observed for GDB without PTFE loading [21]. It is believed that the surface morphology and distribution of the hydrophobic coating materials on carbon paper or carbon cloth can largely affect the MEA performance [16]. The surface morphology and distribution problem arises from the fact that excessive liquid

152

C.-M. Lee et al. / Materials Chemistry and Physics 114 (2009) 151–155

hydrophobic materials may migrate (via capillary action) towards the carbon paper or carbon cloth surfaces and subsequently blocks or seals the surface pores during the drying processes of the wetproof procedures. Once the percentage of the sealed surface pores in the GDB gets higher, neither the removal of liquid water nor the feed of reactant gas can function well, and can thus considerably reduce the power output of fuel cell. Overall, the development of a new GDB wet-proof procedure that provides pore sealing free (SF) or blocking free GDB (i.e., homogeneous hydrophobic coating) has become a challenging issue in MEAs. In this work, the radio frequency (RF) plasma treatment method widely applied in polymer surface property modification [22,23] was adapted to improve the hydrophobic property of GDB. The chance of leaving excessive hydrophobic material residuals to block or seal the gas diffusion pores is expected to be very low by this surface plasma treatment. The as-formed surface morphologies and microstructure of the wet-proofed GDB can then enhance the MEA power output without additional usage of expensive hydrophobic materials (e.g., PTFE or FEP). 2. Experimental 2.1. Plasma-treated Teflon-like materials for GDB In recent years, there has been an increased interest in plasma deposition technology. Plasma is energetically the fourth state of matter, apart from the solid, liquid, and gas state. When a RF power is applied to a gas, it is excited into glow discharge condition through oscillation and collision. The elastic collision frequency () of gas under this condition is normally between 109 and 1011 collisions s−1 , which is much higher than of the applied RF [24] and the electron species in the excited gas experience much more collisions during each applied field cycle [25]. Due to the glow

discharge, the excited and globally neutral plasma phase contains highly reactive species, such as electrons, ions and excited molecules, originated from the mother gas. The RF power is continuously supplied so that the excited gas remains in the plasma phase for further operations. Physical characteristics and chemical compositions of the plasma are generally determined by gas species and system parameters as reported earlier [23]. The preparation procedure of the hydrophobic GDB is described as follows. First, a square carbon cloth with an area of 5 cm2 (BEAM ASSOCIATE, Taiwan) was cleaned by a reactive ion etcher system (Trion Phantom III) under the following working conditions: 10 sccm O2 mass flow, 50 W RF power, 5 × 10−3 Torr working pressure, and 10 min reaction time. After cleaning the carbon cloth, CF4 , SF6 , and CHF3 plasmas were used to treat the carbon surface in the present study. The Teflon-like materials were finally deposited in a high vacuum chamber under the following working conditions: 5 × 10−3 Torr base pressure, 25 sccm mass flow rate, 100 W RF power, and 40 min reaction time, in the reactive ion etcher system (Trion Phantom III).

2.2. Membrane electrode assembly fabrication The MEAs were fabricated by first immersing a Nafion 112 (Du Pont) polymer electrolyte membrane (PEM) in 5% H2 O2 at 80 ◦ C for 1 h to eliminate the membrane surface impurities. The PEMs were then sunk into 0.5 M H2 SO4 solution at 80 ◦ C for another 1 h. The PEMs were frequently washed with 80 ◦ C deionized water before use [3]. A 10 wt% Pt/C catalyst ink, containing chloroplatinic acid (as the metal precursor, Seedchem), carbon (XC-72) and 5 wt% Nafion solution, was prepared by the impregnation method [26]. The catalyst ink was then coated onto one side of the as-prepared PEMs [3], and dried at 80 ◦ C, 1 × 10−2 Torr vacuum atmosphere for 1 h. The opposite side of the membrane was treated similarly after the first side of the PEM was dried. The pre-treated PEM was sandwiched between two layers of Pt/C catalyst ink to serve as anode and cathode. The anode/PEM/cathode modules was designated as pre-MEAs. Note that the Pt loading was kept at a constant value of 0.4 mg cm−2 in the catalyst ink. Finally, under a vacuum atmosphere of 0.6 Torr, the GDLs and pre-MEAs were first pressed by 10 kgf cm−2 at 50 ◦ C for 10 s, followed by 500 kgf cm−2 at 140 ◦ C for 90 s.

Fig. 1. Images of (a) CF4 plasma treated, (b) CHF3 plasma treated, (c) SF6 plasma treated and (d) commercial carbon cloth.

C.-M. Lee et al. / Materials Chemistry and Physics 114 (2009) 151–155 2.3. Experiments on GDB surface characterization and MEA performance Morphology and microstructure analyses of the GDBs (more specifically the ACF carbon cloths) were studied by a JEOL 6700F SEM. The liquid water contact angle (WCA) at the surface of the GDB was measured by the sessile-drop method using a contact angle system FTA 200 (ACIL & First Ten Angstroms Inc.) at room temperature. Four samples with at least five spots per GDB were measured. The electrical resistivity (R) of the GDBs was measured by four-probe test fixture. The crystalline phase analyses of the catalyst ink layers were done by comparing the selected area diffraction (Zeiss 902 A TEM, SAD) patterns and X-ray diffraction (MAC MXT III, XRD) of the Pt/C catalysts to those of the standard compounds documented in the JCPDS [27]. The single cell is assembled by using the single-serpentine pattern of the graphite flow channel [28], and the MEAs mentioned above. Note that the singleserpentine pattern of the graphite flow channel with 1 mm wide and 1 mm deep consists of one continuous channel that proceeds through a series of alternating 180◦ turns. The cell performance curves were measured by a fuel cell system (BEAM ASSOCIATE Co., Ltd.) with the reactant stream being kept at water vapor saturation temperature of 60 ◦ C, back pressure of 0 psi and flow rate of 100 sccm. Note that another back pressure test condition was intentionally set at 25 psi to emphasize the excellent hydrophobic property of the GDB after the plasma treatment.

3. Results and discussion 3.1. Characterization of the GDBs Fig. 1 shows the results observed from the sessile-drop test for the processed carbon cloth (i.e., GDBs). The contact angle was measured by fitting a tangent to three-phase point where liquid surface touches the solid surface. The water contact angle

153

(WCA) for the CF4 plasma treated, CHF3 plasma treated, SF6 plasma treated and commercial carbon cloth was measured as 132.0 ± 0.2◦ , 130.1 ± 0.2◦ , 82.0 ± 0.2◦ and 131.9 ± 0.2◦ , respectively. Obviously excellent hydrophobic property was observed after the CF4 plasma and CHF3 plasma treatment; whereas, more hydrophilic functional groups were formed on carbon cloth when treated with SF6 plasma. Fig. 2 shows the electron microscopy of the plasma treated and commercial carbon cloth wet-proofed surfaces. The surface morphology of carbon cloth treated with CF4 and SF6 plasma reveals a rough surface caused by etching effect. The carbon cloth surface with CHF3 plasma treated seems to be coated with high crosslinking Teflon-like materials. As reported earlier [29], when a wet-proofed material was prepared by the plasma method, the material surface with different F/C stoichiometry can cause a network structure and high crosslinking with variable F/C ratio in the range of 2 ≥ F/C > 0. On the other hand, an ordered chain structure with surface group of –CF3 was observed at even higher F/C ratios. Note that the F/C stoichiometry depends on both gas source and gas flow condition. Our results are thus quite consistent with the reported phenomenon that the plasma-treated method (such as CF4 and CHF3 plasma) can indeed affect the carbon cloth surface morphology. Most importantly, the commercial carbon cloth did not have an even and homogeneous distribution on the fiber surfaces and almost all of the gas diffusion pores next to the cylindrical carbon single fibers were either covered or sealed by the hydrophobic polymer material. It is because excessive PTFE content may hinder the gas diffusion

Fig. 2. Electron microscopic images of (a) CF4 plasma treated, (b) CHF3 plasma treated, (c) SF6 plasma treated, and (d) commercial carbon cloth.

154

C.-M. Lee et al. / Materials Chemistry and Physics 114 (2009) 151–155

and water removal processes [16]. Overall, the advantage of the plasma treatment is very obvious. The electrical resistivity (i.e., in-plane conductivity) values measured were R = 0.45, 0.50, 0.47 and 0.49  for CF4 plasma, CHF3 plasma, SF6 plasma and commercial carbon cloth, respectively. Although the CF4 plasma-treated carbon cloth shows a slightly lower electrical resistivity, the difference is not obvious. Overall, the plasma-treated carbon cloth (or GDBs) can attain a desirable hydrophobic property and a slightly reduced electrical resistance without having the surface gas diffusion pores be sealed or blocked by excessive hydrophobic material residual. This result also suggests that the gaseous reactants are likely to have a better access to the inner electrode catalyst layers where the primary reactions proceed.

3.2. Microstructure and crystal structure of the Pt/C catalysts Fig. 3a shows the bright field morphology of the 20 wt% catalyst ink material. As can be seen, the Pt and CB particles appear in different contrast levels in the TEM image. It is as expected since the high-mass region may scatter more than the low-mass region of a given sample with a constant thickness throughout [30]. The Pt region or high-mass region on the sample exhibits a higher contrast; and vice versa for the CB particles. It also reveals a homogeneous distribution of Pt particle with an average particle diameter of 5 nm on the catalytic support. From the obtained SAD patterns (Fig. 3b), the d spacing was measured as 2.265 Å for the (1 1 1) reflection of the Pt phase. This result clearly indicates a polycrystalline structure of the Pt and CB particles. 3.3. Cell performance testing Fig. 4a shows the polarization results with 0 psi back pressure for various groups of MEAs chosen in this study. The open circuit voltages (OCV) were measured as ∼0.91 V, and the reversible potentials of the fuel cell electrochemical reactions were 1.23 V at standard conditions. The actual voltage is lower than the theoretical value due to irreversible losses [27]. The ohmic range of the MEAs was generally the same (i.e., 15–800 mA cm−2 ) except a relatively narrow of 15–180 mA cm−2 for the SF6 plasma-treated GDB.

Fig. 3. Bright field images (a) and SAD patterns (b) of the 20 wt% catalyst ink prepared by the impregnation method.

Fig. 4. (a) The cell performance results for the MEAs studied. All MEAs used in the test have a Pt loading of ∼0.4 mg cm−2 at a 0 psi back pressure. (b) The power density to current density diagram.

C.-M. Lee et al. / Materials Chemistry and Physics 114 (2009) 151–155

155

4. Conclusion We demonstrate here that the hydrophobic property of GDB was largely improved by the plasma treatment. The surface morphology indicates that the surface gas diffusion pores of the plasma-treated GDBs were less sealed or blocked by the excessive hydrophobic material residual. The water contact angles of the CF4 plasma and CHF3 plasma-treated GDBs were measured as 132.0 ± 0.2◦ and 130.1 ± 0.2◦ , respectively. Polarization and power density measurements were compared and the CF4 plasma-treated modules shows the best performance. Future work will be conducted to improve the CF4 plasma treatment with various processing conditions of temperature, reaction time and vacuum pressure, etc.

Acknowledgements The work is financially supported by the National Science Council (NSC) through grants NSC-093-2218-E-005-040 and NSC094-2218-E-005-007.

References

Fig. 5. (a) The cell performance results with a 25 psi back pressure for the four MEAs studied. (b) The power density to current density diagram.

Fig. 4b shows the P–I plot corresponding to the power output of the four cases. In accordance with the previous results of ohmic range, an optimal power output of 250 mW cm−2 with a current density of 650 mA cm−2 was observed and a much smaller values of 50 mW cm−2 and 120 mA cm−2 for the SF6 plasma-treated MEAs. Again it is due to the fact that the liquid water distribution in the SF6 plasma treatment MEAs is not well managed. To demonstrate the excellent hydrophobic property of the GDB after the plasma treatment, a higher back pressure test condition was studied. Fig. 5a and b shows the polarization results and P–I plot with a 25 psi back pressure. As can be seen, a much higher current density validates the excellent property of the GDB after the CF4 plasma treatment in fuel cell performance. Considering the fact that there is almost no difference in conductivity of the four GDBs, the mass transport through GDB is the key factor to the difference in fuel cell performance.

[1] X.L. Wang, H.M. Zhang, J.L. Zhang, H.F. Xu, Z.Q. Tian, J. Chen, H.X. Zhong, Y.M. Liang, B.L. Yi, Electrochim. Acta 51 (2006) 4909. [2] S. Escribano, J.F. Blachot, J. Ethève, A. Morin, R. Mosdale, J. Power Sources 156 (2006) 8. [3] Y.H. Pai, J.H. Ke, C.C. Chou, J.J. Lin, J.M. Zen, F.S. Shieu, J. Power Sources 163 (2006) 398. [4] X. Zhang, Z. Shen, Fuel 81 (2002) 2199. [5] C. Lim, C.Y. Wang, J. Power Sources 113 (2003) 145. [6] G. Karimi, X. Li, J. Power Sources 140 (2005) 1. [7] X. Ren, S. Gottesfeld, J. Electrochem. Soc. 148 (2001) A87. [8] U. Pasaogullari, C.Y. Wang, J. Electrochem. Soc. 151 (2004) A399. [9] N. Djilali, D. Lu, Int. J. Therm. Sci. 41 (2002) 29. [10] J. Chen, T. Matsuura, M. Hori, J. Power Sources 131 (2004) 155. [11] F. Barbir, PEM Fuel Cells: Theory and Practice, Elsevier/Academic Press, San Diego, 2005, p. 93. [12] S. Litster, G. McLean, J. Power Sources 130 (2004) 61. [13] D. Bevers, R. Rogers, M. von Bradke, J. Power Sources 63 (1996) 193. [14] V.A. Paganin, E.A. Ticianelli, E.R. Gonzalez, J. Appl. Electrochem. 26 (1996) 297. [15] L. Giorgi, E. Antolini, A. Pozio, E. Passalacqua, Electrochim. Acta 43 (1998) 3675. [16] C. Lim, C.Y. Wang, Electrochim. Acta 49 (2004) 4149. [17] D.S. Chan, C.C. Wan, J. Power Sources 50 (1994) 163. [18] J. Benziger, J. Nehlsen, D. Blackwell, T. Brennan, J. Itescu, J. Membr. Sci. 261 (2005) 98. [19] H.K. Lee, J.H. Park, D.Y. Kim, T.H. Lee, J. Power Sources 131 (2004) 200. [20] R.W. Jaszewski, H. Schift, B. Schnyder, A. Schneuwly, P. Groning, Appl. Surf. Sci. 143 (1999) 301. [21] T. Ioroi, T. Oku, K. Yasuda, N. Kumagai, Y. Miyazaki, J. Power Sources 124 (2003) 385. [22] D. Barton, J.W. Bradley, D.A. Steele, R.D. Short, J. Phys. Chem. B 103 (1999) 4423. [23] C. Riccardia, R. Barnia, E. Sellib, G. Mazzoneb, M.R. Massafrac, B. Marcandallic, G. Poletti, Appl. Surf. Sci. 211 (2003) 386. [24] R.F. Baddour, R.S. Timmins (Eds.), The Application of Plasmas to Chemical Processing, MIT Press, Cambridge, MA, 1967. [25] A. Grill, Cold Plasma in Materials Fabrication, Institute of Electrical and Electronics Engineers Inc., New York, 1994, pp. 24–45. [26] S.R. de Miguel, J.I. Vilella, E.L. Jablonski, O.A. Scelza, C. Salinas-Martinez de Lecea, A. Linares-Solano, Appl. Catal. A: Gen. 232 (2002) 237. [27] Y.H. Pai, H.F. Huang, Y.C. Chang, C.C. Chou, F.S. Shieu, J. Power Sources 159 (2006) 878. [28] S.F. Liu, H.S. Chu, P. Yuan, J. Power Sources 161 (2006) 1030. [29] P. Favia, R. d’Agostino, Surf. Coat. Technol. 98 (1998) 1102. [30] D.B. Williams, C.B. Carter, Transmission Electron Microscopy, Plenum Press, New York, 1996, pp. 351–354.