Chemically Tuned Anode with Tailored Aqueous Hydrocarbon Binder

Jun 1, 2009 - An anode for direct methanol fuel cells was chemically tuned by tailoring an aqueous hydrocarbon catalyst .... were purchased from Aldri...
0 downloads 0 Views 4MB Size
Download PDF
pubs.acs.org/Langmuir © 2009 American Chemical Society

Chemically Tuned Anode with Tailored Aqueous Hydrocarbon Binder for Direct Methanol Fuel Cells Chang Hyun Lee,†,‡ So Young Lee,† Young Moo Lee,*,† and James E. McGrath‡ †

School of Chemical Engineering, College of Engineering, Hanyang University, Seoul 133-791, Korea, and ‡Macromolecules and Interfaces Institute, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061 Received February 2, 2009. Revised Manuscript Received April 11, 2009

An anode for direct methanol fuel cells was chemically tuned by tailoring an aqueous hydrocarbon catalyst (SPI-BT) binder instead of using a conventional perfluorinated sulfonic acid ionomer (PFSI). SPI-BT designed in triethylamine salt form showed lower proton conductivity than PFSI, but it was stable in the catalyst ink forming the aqueous colloids. The aqueous colloidal particle size of SPI-BT was much smaller than that of PFSI. The small SPI-BT colloidal particles contributed to forming small catalyst agglomerates and simultaneously reducing their pore volume. Consequently, the high filling level of binders in the pores, where Pt-Ru catalysts are mainly located on the wall and physically interconnected, resulted in increased electrochemical active surface area of the anode, leading to high catalyst utilization. In addition, the chemical affinity between the SPI-BT binder and the membrane material derived from their similar chemical structure induced a stable interface on the membrane-electrode assembly (MEA) and showed low electric resistance. Upon adding SPI-BT, the synergistic effect of high catalyst utilization, improved mass transfer behavior to Pt-Ru catalyst, and low interfacial resistance of MEA became greater than the influence of reduced proton conductivity in the electrochemical performance of single cells. The electrochemical performance of MEAs with SPI-BT anode was enhanced to almost the same degree or somewhat higher than that with PFSI at 90 °C.

Introduction The polymer electrolyte membrane fuel cell (PEFC) is a promising energy source to directly convert the chemical energy of fuels such as hydrogen and methanol to the electric energy with high efficiency and low emission of pollutants.1 The electrochemical performances of the fuel cell are greatly affected by the key components of the fuel cell, including a polymer electrolyte membrane, catalysts for anode and cathode, and their assembly (membrane-electrode assembly, MEA).2 Another important component in the conventional MEA formation is the catalyst binder needed for the electrode formula. The catalyst binder acts not only as a proton conducting medium in both the electrodes but also as a mechanical supporter for catalyst agglomerates resulting from a strong physical interaction between the carbon substrate particles3 where nanosized catalysts are dispersed on their surface (bottom of Figure 1). For this purpose, perfluorinated sulfonic acid ionomers (PFSIs) including Nafion have been widely used, owing to their high proton conductivity (∼0.1 S cm-1 at 30 °C in liquid water), excellent chemical durability, and inertness to catalyst.2 Unlike the MEA based on PFSI membranes with similar chemical structures, the PFSI catalyst binder can cause severe problems in the electrochemical performance of a MEA based on hydrocarbon membranes (HC-MEA). Different surface and bulk properties between the hydrocarbon membranes and the PFSI binder usually induce their low compatibility and thus form *Corresponding author. Telephone: +82-2-2220-0525. Fax: +82-2-22915982. E-mail: [email protected]. (1) Grove, W. R. Philos. Mag. Ser. 3 1839, 14, 127–130. (2) Steele, B. C. H.; Heinzel, A. Nature 2001, 414, 15. (3) Borup, R.; Meyers, J.; Pivovar, B.; Kim, Y. S.; Mukundan, R.; Garland, N.; Myers, D.; Wilson, M.; Garzon, F.; Wood, D.; Zelenay, P.; More, K.; Stroh, K.; Zawodzinski, T.; Boncella, J.; McGrath, J. E.; Inaba, M.; Miyatake, K.; Hori, M.; Ota, K.; Ogumi, Z.; Miyata, S.; Nishikata, A.; Siroma, Z.; Uchimoto, Y.; Yasuda, K.; Kimijima, K.; Iwashita, N. Chem. Rev. 2007, 107, 3904.

Langmuir 2009, 25(14), 8217–8225

unstable interfaces in the HC-MEA. Though some hydrocarbon membranes exhibit excellent initial single cell performances owing to proton conductivities similar to or higher than those of PFSI membranes,4-7 their current-voltage polarization curves rapidly drop within several days and are accompanied by the delamination of the electrode layers from the membranes.8 This phenomenon can be clearly observed in the direct methanol fuel cell (DMFC) system where all polymeric MEA components, including the hydrocarbon membranes, are fully hydrated or dehydrated repeatedly and methanol molecules are easily transported with water molecules through the membranes.9 There have been a number of approaches to improve the compatibility between hydrocarbon membranes and catalyst binders in the HC-MEA. One is to fabricate partially fluorinated membranes using PFSI as a catalyst binder.10,11 C-F bonds in the membranes contributed to reduced interfacial resistances with the electrodes containing PFSI, leading to improved MEA lifetime. However, the relatively high cost of fluorinated monomers may make the merits of the fluorinated membranes less attractive.12-14 (4) Asano, N.; Aoki, M.; Suzuki, S.; Miyatake, K.; Uchida, H.; Watanabe, M. J. Am. Chem. Soc. 2006, 128, 1762. (5) Tsang, E. M. W.; Zhang, Z.; Shi, Z.; Soboleva, T.; Holdcroft, S. J. Am. Chem. Soc. 2007, 129, 15106. (6) Li, Y.; Roy, A.; Badami, A. S.; Hill, M.; Yang, J.; Dunn, S.; McGrath, J. E. J. Power Sources 2007, 172, 30. (7) Yamaguchi, T.; Zhou, H.; Hara, N. Adv. Mater. 2007, 19, 592–596. (8) Scott, K.; Taama, W. M.; Argyropoulos, P. J. Membr. Sci. 2000, 171, 119. (9) Day, T. J. F.; Schmitt, U. W.; Voth, G. A. J. Am. Chem. Soc. 2000, 122, 12027. (10) Wiles, K. B.; Diego, C. M.; Abajo, J.; McGrath, J. E. J. Membr. Sci. 2007, 294, 22. (11) Sankir, M.; Kim, Y. S.; Pivovar, B. S.; McGrath, J. E. J. Membr. Sci. 2007, 299, 8. (12) Kim, D. S.; Robertson, G. P.; Guiver, M. D.; Lee, Y. M. J. Membr. Sci. 2006, 281, 111. (13) Ghassemi, H.; McGrath, J. E.; Zawodzinski, T. A. Polymer 2006, 47, 4132. (14) Nieh, M. P.; Guiver., M. D.; Kim, D. S.; Ding, J.; Norsten, T. Macromolecules 2008, 41, 6176.

Published on Web 06/01/2009

DOI: 10.1021/la900406d

8217

Article

Lee et al.

Figure 1. Schematic representation of the polymeric MEA components used in this study and the anode structure composed of catalyst agglomerates showing different electrochemical behaviors.

A second approach used identical materials with the hydrocarbon membranes as a catalyst binder instead of PFSIs.15,16 This approach can guarantee excellent adhesion between the hydrocarbon membranes and the electrodes but may lead to insufficient fuel and oxygen supply through the electrodes, which limits the electrochemical catalytic reaction in the catalyst agglomerates.17,18 Furthermore, it is not easy to control the degree of sulfonation (DS) of the same polymeric materials used for both membranes and catalyst binders. The membrane materials should be insoluble in the fully hydrated state under fuel cell conditions, while the binder materials should be soluble or be present in a colloidal state in an aliphatic alcohol-water mixture for electrode ink formation. One way to meet this contradictory requirement was to add organic solvents in the conventional electrode formula for dissolving the hydrocarbon materials with a relatively low DS (60 mol %) for both membrane and catalyst (15) Jung, H. Y.; Cho, K. Y.; Sung, K. A.; Kim, W. K.; Park, J. K. J. Power Sources 2006, 163, 56. (16) Jung, H. Y.; Cho, K. Y.; Sung, K. A.; Kim, W. K.; Kurkuri, M.; Park, J. K. Electrochim. Acta 2007, 52, 4916. (17) Pivovar, B. S.; Kim, Y. S. J. Electrochem. Soc. 2007, 154(8), B739. (18) Sung, K. A.; Jung, H. Y.; Kim, W. K.; Cho, K. Y.; Park, J. K. J. Power Sources 2007, 169, 271.

8218 DOI: 10.1021/la900406d

binder fabrication, considering that the materials can be dissolved in hot water for catalyst ink formation.17 Unfortunately, the HCMEAs failed to show potential for stable and long-term operation due to their susceptibility to hydrolytic and chemical attack. Here, we report a facile approach to fabricate a HC-MEA with high catalyst utilization and low interfacial resistances between the membrane and electrodes without a loss in catalyst activity by designing an aqueous hydrocarbon binder for the anode. It is important to use an aqueous catalyst binder for maximum catalyst activity in the MEA. A sulfonated polyimide-silica nanocomposite (SPI-SiO2) membrane with excellent hydrolytic durability (∼1 year in hot water at 80 °C) was used as the membrane material.19 The aqueous hydrocarbon binder (SPIBT, Figure 1) was synthesized in sulfonated polyimide triethylammonium (Et3N) salt form with a chemical structure similar to that of the polymer matrix in a nanocomposite membrane. SPIBT was then blended with a PFSI binder using a different ratio. The changes in the physicochemical and electrochemical characteristics of the HC-MEA according to SPI-BT content in blends with PFSI were systematically investigated. Finally, the efficacy of the SPI-BT binder was verified on the basis of electrochemical performances of single cells.

Experimental Section 0

Materials. 4,4 -Diaminodiphenyl ether (ODA) and 3,5-diaminobenzoic acid (DBA) as diamines were purchased from Tokyo Kasei Co. (Tokyo, Japan) and used as received. ODA was then converted into 4,40 -diaminodiphenyl ether-2,20 -disulfonic acid (SDODA) via direct sulfonation using concentrated sulfuric acid (19) Lee, C. H.; Park, C. H.; Lee, S. Y.; Jung., B. O.; Lee, Y. M. Desalination 2008, 233, 210.

Langmuir 2009, 25(14), 8217–8225

Lee et al. (95%, Aldrich, WI) and fuming sulfuric acid (SO3, 30% Aldrich, WI).20 1,4,5,8-Naphthalenic tetracarboxylic dianhydride (NTDA, Tokyo Kasei Co, Tokyo, Japan) was used as a dianhydride and as received. Benzoic acid, triethylamine, and m-cresol were purchased from Aldrich Chemical Co (WI) and used as received. Aerosil 200 (BET surface area = 200 ( 25 m2 g-1 and average particle size = 12 nm) was obtained from Degussa Chemical Co. (Dusseldorf, Germany) and dried at 80 °C and 3-5 mmHg prior to use. Pluronics L64 (PEO13-PPO30-PEO13, BASF, Ludwigshafen, Germany), a commercial surfactant, was received and used as a dispersant for the homogeneous distribution of Aerosil 200. Nafion (5 wt % in water-isopropyl alcohol (IPA) mixture, SE5142, proton-form, EW = 1100) was purchased from Aldrich (WI) and used as a PFSI binder. Membrane Fabrication. Random SPI-SiO2 as a polymer matrix was fabricated by two consecutive steps: (1) thermal solution imidization in m-cresol at 80 °C for 4 h and 180 °C for 24 h using SDODA (1.6 mmol), DBA (2.4 mmol), and NTDA (4.0 mmol) and (2) direct mixing of Aerosil 200-L64 solution mixture in m-cresol (weight ratio: Aerosil 200/L64 = 1/3) with the SPI solution.19 The silica content was 1 wt % based on total polymer weight. The final solution (∼15 wt %) was cast on a glass plate, dried at 80 °C for 2 h, and heated at 180 °C for 10 h under vacuum for complete imidization via additional thermal activated reaction. After soaking in methanol at room temperature for 8 h for residual solvent removal, the SPI membrane with an average thickness of 30 μm was thoroughly acidified by treatment in 1 M hydrochloric acid at 25 °C for 24 h and washed in deionized water at 25 °C for 24 h. Finally, SPI-SiO2 membrane was dried in a vacuum oven at 110 °C for at least 24 h. The proton conductivity of the membrane was 9.27  10-2 S cm-1 at 30 °C and 95% relative humidity (RH). The measurement method is described in the Characterization section below. Preparation of Hydrocarbon Binder for Anode. SPI-BT was fabricated with an equivalent mole of SDODA and NTDA using the same thermal protocol as that of the random SPI as a polymeric material used for membrane fabrication. Considering the molar equivalent of sulfonic acid groups in the solution mixture, a constant Et3N content was added in the solution mixture. After the imidization reaction, additional m-cresol was added to dilute the highly viscous Et3N-salt SPI-BT solution and then the solution was cooled down to 110 °C. The SPI-BT solution was precipitated in cold acetone to eliminate unreacted monomers or oligomers with low molecular weight. The Et3N-salt-form SPIBT precipitate (intrinsic viscosity (IV) = 0.7 dL g-1 in NMP at 25 °C) was filtered, washed, and dried under vacuum at 120 °C. SPI-BT was used as a binder component for anode ink formation. Electrode Formation and MEA Fabrication. Cathode catalyst ink was prepared by dispersing a desirable amount of 5 wt % Nafion (PFSI) and Pt black (HISPEC 1000, Johnson Matthey Corp., London, U.K.) in an IPA-water mixture. In all MEAs, PFSI was used as cathode catalyst binder. For anode catalyst ink formation, SPI-BT content in blends with PFSI was varied from 0 to 100 wt % (Table 1). Pt-Ru black (HISPEC 6000, Johnson Matthey Corp., London, U.K.) was added to the IPAwater mixture containing the blends and sonicated for several minutes to obtain a well dispersed catalyst ink slurry. Here, IPA and water content (weight ratio of 3:1) was fixed in all ink compositions. Furthermore, the blends were prepared in membrane form via the solution casting method for their proton conductivity evaluation using four-point probe ac impedance spectroscopy.21 The cathode catalyst ink was sprayed on the random SPI-SiO2 membranes (catalyst-coated membrane method). After complete drying, the other sides of the membranes were coated with (20) Fang, J.; Guo, X.; Harada, S.; Wateri, T.; Tanaka, K.; Kita, H.; Okamoto, K. Macromolecules 2002, 35, 9022. (21) Lee, C. H.; Park, H. B.; Lee, Y. M.; Lee, R. D. Ind. Eng. Chem. Res. 2005, 44, 7617.

Langmuir 2009, 25(14), 8217–8225

Article Table 1. Proton Conductivity of the Catalyst Binder Blends and ECSA Values of Anode Obtained from CV Results SPI-BT content in the blends with PFSI [wt %]

proton conductivity [σ, S cm-1]a

ECSA [m2 g-1 Pt-Ru]b

0 0.08 15.09 50 0.06 75 0.04 10 0.03 21.10 a Measured in the membrane state without annealing process under 60 °C and 95% relative humidity. b Measured via CV in the MEA state.28

individual anode catalyst ink. Each MEA had an active area of 5 cm2. The loading content of each catalyst and catalyst binder was 3 and 0.3 mg cm-2, respectively. Both sides of the catalystcoated membranes then were covered with gas diffusion layers (GDL, standard wet proofing carbon paper, Toray Inc., Japan) and hot-pressed at 130 °C and 180 kg cm-2 for 3 min. Characterization. The ohmic resistance (R, Ω, active area = 1  4 cm2) of the blends of SPI-BT and PFSI as well as the SPI-SiO2 membrane was measured via a four-point probe ac impedance spectroscopic method using a combination of Solartron 1260 and 1287 instruments. The ohmic resistance was converted to proton conductivity (σ, S cm-1) by using the equation of σ = l/(R  S) (l = the distance between the reference electrodes, and S = the cross-sectional area of the membrane sample). The proton conductivity was measured in a hygro- and thermocontrolled chamber where all electromagnetic noises were thoroughly shielded. The SPI-BT particle dispersion in the IPA-water mixture was evaluated by using dynamic light scattering (DLS, angle = 90°) at 30 °C (Malvern Zetasizer Nano ZS, Malvern Instrument, U.K.). The concentrations of all samples were fixed at 0.5 wt/vol %. Each measurement was carried out repeatedly at least nine times. The anode surface morphology for each MEA was examined by using a scanning electron microscope (FE-SEM, JEOL JSF 6340F, Tokyo, Japan). The transmission electron microscope (HR-TEM, JEOL3010, Tokyo, Japan) was used to observe the catalyst distribution depending on SPI-BT content. The pore characteristics of the catalyst agglomerates were evaluated by means of both the bubble point test method (Perm-porometer, PMI, Ithaca, NY)22 and the nitrogen adsoprtion-desorption method (ASAP 2010 Micropore Analyzer, Micromeritics, Norcross, GA).23 The electrochemical active surface area (ECSA) of the anode was measured via cyclic voltammetry (CV, Autolab PGSTAT 30, Eco-Chemie Co., The Netherlands) in the MEA state at 25 °C.24,25 For the CV measurement, 1 standard cubic centimeter per minute (sccm) and 200 sccm of deionized water and dry hydrogen gas were fed to the anode (working electrode) and the cathode (reference/counter electrodes), respectively. The CV curve of the anode was obtained after scanning potential at 20 mV s-1 between 0 V and +1.3 V. The electrochemical single cell performances of the MEAs were evaluated for 48 h using a combination system of a single direct methanol fuel cell apparatus (CNLPEM005-1, CNL Energy, Korea) and an electric loader (WFCTS, WonATech Co., Ltd., Korea). The flow rates of 1 M methanol and humidified oxygen to the anode and cathode were 1 and 200 sccm under ambient pressure, respectively.

Results and Discussion The proton conductivity of a catalyst binder is an important factor in determining the fast proton transport through the (22) Kuroki, H.; Yamaguchi, T. J. Electrochem. Soc. 2006, 153, A1417. (23) Kim, J. H.; Kim, H. J.; Lim, T. H.; Lee, H. I. J. Power Sources 2007, 170, 275. (24) Kim, J. H.; Ha, H. Y.; Oh, I. W.; Hong, S. A.; Kim, H. N.; Lee, H. I. Electrochim. Acta 2004, 50, 801. (25) Liang, Z. X.; Zhzo, T. S. J. Phys. Chem. C 2007, 111, 8128.

DOI: 10.1021/la900406d

8219

Article

electrodes and is associated with high electrochemical MEA performances. Proton conductivity usually increases with high DS. However, high sulfonic acid (-SO3H) content caused the physicochemical and electrochemical instability. The conversion of protonated -SO3H groups to salt-form (-SO3M, M = Na+, organic, or metallic Lewis acids) has been shown to stabilize the same polymeric material.26,27 Among the salt-form catalyst binders, Et3N salt-form SPI-BT was chosen in this study, since it showed a similar proton conductivity (∼0.03 S cm-1 at 60 °C and 95% RH) and higher MEA performance compared to those using sodium-salt-form (-SO3Na) SPI-BT based on our measured data. Note that the SPI-BT was compatible with PFSI, showing no macrophase separation. The proton conductivity of the blend decreased with high SPI-BT content, owing to its lower conductivity than PFSI (Table 1). The solubility of a catalyst binder to specific solvents used for catalyst ink formula can influence the catalyst dispersion in the ink state and/or the electrodes, the catalyst activity and utilization, and the physicochemical and electrochemical stability of the MEAs.28,29 Considering that the solubility (δ) of a polymer is generally determined by a combination of inter- and intramolecular van der Waals force (δd), dipole-dipole interaction (δp), and hydrogen bond (δh),30 a triangular solubility diagram based on Hansen’s three-dimensional solubility parameters applicable to a large number of solvents may be useful to predict the solubility behavior of the polymer in the solvents.31,32 In the diagram, the soluble region of the polymer is referred to as “solubility envelope”. Sulfonated hydrocarbons used as catalyst binders are usually soluble in most polar aprotic solvents including dimethylacetamide (DMAc). Here, SPI-BT is not exceptional. Thus, the solubility envelope marked as the red dashed area in Figure 2 may be nearly overlapped in the sulfonated hydrocarbons. Also, the solubility envelope can be expanded at the elevated temperatures owing to the fast molecular movement of the polymer chains. When the solvents within the solubility envelope are used to disperse both catalyst particles including carbon-supported catalysts, and catalyst binders, severe problems in all the MEA components may occur owing to the side effects of the residual high-boiling-point solvents. The problems include a proton conductivity drop in either catalyst binders or membranes, pinhole formation in the membranes, and mass flux limitation in the corresponding electrodes for fuel or oxygen transport.18 For these reasons, it is very important to select appropriate solvents for catalyst ink formation. We took notice of the solvent composition of the commercial PFSI solution (IPA/water = 3:1, weight ratio), where PFSI is present in the colloidal state.33 Hence, SPI-BT was designed to dissolve in solvents such as in m-cresol within the solubility envelope for easy processing and, at the same time, to be in a colloidal state in an aqueous IPA solution for catalyst dispersion. In fact, SPI-BT was insoluble in IPA and methanol even at around their boiling temperatures. Therefore, the tie-line connecting IPA and methanol can become a boundary for the insoluble region of SPI-BT including the aqueous IPA and (26) Almeida, S. H. D.; Kawano, Y. J. Therm. Anal. Calorim. 1999, 58, 569. (27) Lage, L. G.; Delgado, P. G.; Kawano, Y. J. Therm. Anal. Calorim. 2004, 75, 521. (28) Uchida, M.; Fukuoka, Y.; Sugawara, Y.; Eda, N.; Ohta, A. J. Electrochem. Soc. 1996, 143, 2245. (29) Song, J. M.; Suzuki, S.; Uchida, H.; Watanabe, M. Langmuir 2006, 22, 6422. (30) Kinzer, K. E.; Lloyd, D. R.; Wightman, J. P.; McGrath, J. E. Desalination 1983, 46, 327. (31) Barton, A. F. M. CRC handbook of solubility parameters and other cohesion parameters; CRC Press: Boca Raton, FL, 1991. (32) Teas, J. P. J. Paint Technol. 1968, 40, 19. (33) Szajdzinska-Pietek, E.; Schlick, S. Langmuir 1994, 10, 2188.

8220 DOI: 10.1021/la900406d

Lee et al.

Figure 2. Triangular solubility diagram of SPI-BT at 25 °C. Here, Fi [%] = [δi/(δd + δp + δh)]  100 (i = d, p, and h). The red dashed area represents the solubility envelope of the sulfonated hydrocarbon polymers including SPI-BT.

Figure 3. SPI-BT colloidal particle size in IPA-water mixtures obtained via DLS measurement at 30 °C.

1 M methanol solution, which acts as a dispersion solvent and a fuel for ink formation and DMFC operation, respectively, in this study. The dispersion behavior of the PFSI and SPI-BT solutions was investigated. As reported in the literature,33,34 PFSI in the IPAwater mixture used for catalyst ink formation in this study exhibited bimodal distribution, indicating small colloidal particles (1000 nm, portion: ∼80%) derived from aggregation between PFSI interchains (Figure 3).34 The colloidal particle size of SPI-BT was between 18 and 210 nm in tested aqueous IPA solutions. Differing from the case of PFSI, SPI-BT showed a relatively uniform, monodisperse distribution, irrespective of water content. The SPI-BT size increased with water content in the aqueous IPA solutions, since the hydrophilic polymer chains can be swollen further. In particular, in the same IPA-water mixture (IPA/water = 3:1 or 75 wt % IPA solution), the colloidal particle size of SPI-BT was much smaller than the particle size of PFSI, as can be seen in Figure 3. The small SPIBT size originated from Et3N molecules which shielded the (34) Wang, S.; Sun, G.; Wu, Z.; Xin, Q. J. Power Sources 2007, 165, 128.

Langmuir 2009, 25(14), 8217–8225

Lee et al.

Figure 4. Volume frequency percent as a function of the Pt-Ru/ CB agglomerate diameter in the catalyst ink state at 30 °C.

intermolecular interaction between the -SO3H groups in the SPIBT chains via the formation of ionic complex of -SO3-Et3NH+ and hindered the chains from being spontaneously aggregated. Similar results have been reported even in a PFSI study, where the PFSI particle size was reduced up to 30-40 nm after the addition of LiCl, NaCl, or NaOH in the IPA-water mixture.34-36 Therefore, it was expected that SPI-BT changed the size of the agglomerates composed of carbon substrates deposited with PtRu catalyst nanoparticles, abbreviated as Pt-Ru/CB, in the IPA-water mixture. Figure 4 shows the Pt-Ru/CB agglomerate diameter and its distribution in the catalyst inks with different amounts of SPI-BT, measured by using DLS. The Pt-Ru/CB without any catalyst binders aggregated in the broad diameter ranges from 10 to 250 μm. After the incorporation of the catalyst binders, the agglomerate sizes became small, regardless of which catalyst binder was employed. This result suggests that the catalyst binders led to Pt-Ru/CB dispersion by weakening the physical interaction between Pt-Ru/CB particles. Note that the reduction of agglomerate size was very prominent in the large aqueous SPI-BT binder content, suggesting that SPI-BT changed the corresponding anode microstructures. It is interesting to investigate the relationship between the tuned anode microstructure and the electrochemical performances of single cells. The micropore structures in the fuel cell electrodes are classified into the primary pore (20-40 nm) in catalyst agglomerates and the secondary pore (40 nm to 1 μm) between the agglomerates (bottom of Figure 1).29,37 The catalyst nanoparticles (over 85%) are mainly located on the wall of the primary pore.37 Therefore, the catalyst binder should penetrate the catalyst agglomerates, as much as possible, in order to maximize the utilization of catalysts participating in the electrocatalytic reaction. Figure 5 shows FE-SEM images of the anode surface by varying aqueous SPI-BT content in the blends with PFSI where the loading contents of the catalyst binder blends (0.3 mg cm-2) and the Pt-Ru electrocatalyst (3 mg cm-2) are the same for all anodes. In all images, the aggregation of Pt-Ru/CB was observed by the naked eye. The agglomerate size decreased with high SPIBT content, which was in accord with the trend in Figure 4. The small agglomerates seem to narrow the space between the (35) Lin, H. L.; Yu, T. L.; Huang, C. H.; Lin, T. L. J. Polym. Sci., Part B: Polym. Phys. 2005, 43, 3044. (36) Jiang, S.; Xia, K. Q.; Xu, G. Macromolecules 2001, 34, 7783. (37) Watanabe, M.; Tomikawa, M.; Motoo, S. J. Electroanal. Chem. 1985, 195, 81.

Langmuir 2009, 25(14), 8217–8225

Article

agglomerates and reduce the secondary pore volume on the anode surface. In contrast, the dispersion of the catalyst binder and the Pt-Ru catalyst are not easily seen in the SEM surface images. The energy dispersive spectrometer (EDS) images in Figure 6 confirm that the white sulfur (S) element in SPI-BT and ruthenium (Ru) element in Pt-Ru/CB increased with SPI-BT content from 50 to 100 wt %. Their distribution patterns in the same anode surface were almost the same. Notice that S and Ru elements were uniformly distributed in the anode surface containing SPI-BT (100 wt %), while the elements in the anode containing only PFSI (0 wt %) formed aggregated domains. Though the SEM images provide visual information on the influence of aqueous SPI-BT on the secondary pore size at the anode surfaces, it was difficult to bring insight into the pore structure within the anode. Moreover, the morphological images were not adequate to represent the overall pore characteristics including the primary pore in the anodes. To characterize the secondary pore structure, the bubble point test method was used. For this, each catalyst ink was coated with the same loading content on GDL (porous carbon paper, blank) instead of the SPI-SiO2 membrane, where its catalyst microstructure was believed not to be very different from that coated on the membrane. Figure 7 shows the changes of the secondary pore characteristics with SPI-BT content. After coating with catalyst ink containing PFSI, large pores in the blank became small up to around 1.2 μm with a broad pore size distribution. Even in a relatively small amount of SPI-BT, the secondary pore size was dramatically reduced to the level of 7% of PFSI. The increase in SPI-BT content in the blend induced the uniform distribution of the nanopores of small size. In the anode containing the aqueous catalyst binder composed of 75 wt % SPI-BT and 25 wt % PFSI blend, the most of the nanopores had an average pore size of about 81 nm. The pores were further filled at the anode containing only SPI-BT, resulting in a great reduction in pore volume. The primary pore characteristics were investigated by measuring the Brunauer-Emmett-Teller (BET) surface area for the anodes containing different aqueous binders.38 The features observed in the primary pore (Figure 8) were not greatly different from those of the secondary pore: the primary pore diameter and the pore volume decreased with SPI-BT content. The reduced pore volume was attributed to an increase in the filling level of the primary pores deposited with SPI-BT. It can be observed that SPI-BT contributed to improving catalyst utilization in the electrocatalytic reaction by effectively coating Pt-Ru catalysts mostly located on the wall of the primary pore (> ∼85%), in addition to the secondary pore (< ∼15%). The TEM images in Figure 9 display the difference in the PtRu/CB distribution patterns within the catalyst binders. Considering the much higher electron density of the Pt-Ru catalyst than other elements and the average size of each component in the catalyst inks,29,39 the dark spots and the surrounding gray cloudy regions are likely to have been derived from Pt-Ru catalyst and catalyst binder, respectively. In all images, Pt-Ru/CB particles were located within catalyst binders, indicating the formation of a proton transport network. Figure 9a clearly shows that PFSI induces typical agglomerate structures composed of the primary pore (40 nm), and most of the Pt-Ru/CB particles were dispersed on the wall of the primary pores. In contrast, the introduction of SPI-BT eliminated most of (38) Ilinitch, O. M.; Fenelonov, V. B.; Lapkin, A. A.; Okkel, L. G.; Terskikh, V. V.; Zamaraev, K. I. Microporous Mesoporous Mater. 1999, 31, 97. (39) Roth, C.; Papworth, A. J.; Hussain, I.; Nochols, R. J.; Schiffrin, D. J. J. Electroanal. Chem. 2005, 581, 79.

DOI: 10.1021/la900406d

8221

Article

Lee et al.

Figure 5. FE-SEM images of the anode surfaces coated on SPI-SiO2 membranes at a magnification of 5000 and 500.

Figure 6. S- and Ru-mapping EDS images of anode surfaces coated on SPI-SiO2 membranes.

Figure 7. Secondary pore size and pore size distribution of anodes containing different SPI-BT contents based on the bubble point test method. 8222 DOI: 10.1021/la900406d

Figure 8. Primary pore volumes of the anodes containing different SPI-BT contents as a function of pore diameter obtained via the nitrogen adsorption-desorption method (BET measurement). Langmuir 2009, 25(14), 8217–8225

Lee et al.

Article

Figure 9. High resolution TEM images of Pt-Ru/CB coated with (a) 0 wt % SPI-BT (100 wt % PFSI), (b) 50 wt % SPI-BT, and (c) 100 wt % SPI-BT content.

Figure 10. Voltammograms of anodes in the MEA state. Here, current density was based on the electrochemical surface area evaluated from cyclic voltammetery.

Figure 11. HFR values of the HC-MEAs as a function of SPI-BT content.

the large secondary pores, even at the anode containing a small amount of SPI-BT. Note that each Pt-Ru particle became interconnected with each other (Figure 9b and c). However, it does not mean that the transition into a peculiar pore structure was made up of only the primary pore since the microscopic limit at high magnification (80 000) should be considered. It is clear that SPI-BT created a compact pore packing structure Langmuir 2009, 25(14), 8217–8225

with a high pore filling level and, simultaneously, disturbed the self-aggregation of the Pt-Ru/CB particles. Consequently, SPI-BT built a physical network of Pt-Ru/CB to promote catalyst utilization. Nevertheless, the improved physical connectivity does not always promise a high electrochemical performance because some residual chemicals and elements in the catalyst inks and/or anode layer may negatively affect the catalyst activity.40 Thus, the catalyst activity must be examined by comparing ECSA values of the Pt-Ru catalysts in anodes containing different catalyst binders. Here, the ECSA was measured by using the hydrogen desorption peaks in the low potential region. Figure 10 and Table 1 show that the anode containing only SPI-BT binder has a higher ECSA (21.1 m2 g-1 Pt-Ru) than that containing only PFSI (15.1 m2 g-1 Pt-Ru). This indicates that Et3N salt-form SPI-BT was electrochemically harmless to Pt-Ru catalyst activity. It has been reported that the nitrogen atom in amines does not cause catalyst poisoning.40,41 Also, it was electrochemically verified that the increased filling level of SPI-BT into the primary pores and the improved physical interconnection of Pt-Ru catalyst induced high catalyst utilization. In other words, a relatively high amount of Pt-Ru catalyst became more accessible to protons/electrons in the anode containing SPI-BT owing to the improved catalyst/binder interface. The electric resistance measurement of the HC-MEAs with different SPI-BT contents can provide valuable information on high frequency resistance (HFR, Ω cm2 in Figure 11), which is the sum of all resistances in the pathway from the potentiometer to the MEAs. The measurement was carried out by using the same test equipment and interconnection method. Therefore, the relative differences in the HFR values are derived from electrode resistance and interfacial resistance between the membrane and the electrodes. Unfortunately, it is not easy to isolate each resistance in the HFR values. The electrode resistance is determined mainly by anode catalyst binders, since their resistance values are much higher than those of Pt-Ru catalyst agglomerates. SPI-BT has higher bulk resistance (or low proton conductivity) than PFSI. Thus, it can be easily estimated that the electrode resistance increases with SPI-BT content. On the other hand, the chemical similarity of SPI-BT with the polymer matrix in the SPI-SiO2 membrane can cause a low interfacial resistance. (40) Roy, S. C.; Harding, A. W.; Russell, A. E.; Thomas, K. M. J. Electrochem. Soc. 1997, 144, 2323. (41) Yang, L.; Chen, J.; Wei, X.; Liu, B.; Kuang, Y. Electrochim. Acta 2007, 53, 777.

DOI: 10.1021/la900406d

8223

Article

Lee et al.

Figure 12. Electrochemical single cell performances of the HC-MEAs (circle, b/O) containing anode catalyst binders with different SPI-BT contents: 0 wt % (black), 50 wt % (red), 75 wt % (blue), and 100 wt % (cyan) at (a) 60 °C and (b) 90 °C. For comparison, a reference MEA (black square, 9/0) based on Nafion 117 membrane and PFSI binder for both electrodes was used.

In practice, interfacial resistance dominates electrode resistance in the overall resistance. HFR values of the HC-MEAs with high SPI-BT content became lowered up to half of that of the HC-MEA using PFSI binder only. It means that the anode containing SPI-BT formed a strong interface with SPI-SiO2 membrane. The electrochemical single cell data can be used to comprehensively evaluate the availability of SPI-BT as a catalyst binder when all components except the catalyst binders are identical and the corresponding HC-MEAs are assembled in the same manner. Figure 12 presents the current-voltage polarization curves of the HC-MEAs with different SPI-BT contents. PFSI in a reference MEA based on Nafion 117 membrane (methanol permeability (PMeOH): 1.4  10-6 cm2 s-1 under 10 M methanol at 25 °C)19 brought about excellent single cell performances. In spite of the much reduced methanol permeability (PMeOH: 1.5  10-7 cm2 s-1) of the SPI-SiO2 membrane under the same measurement condition,19 the HC-MEA using the PFSI binder as an anode catalyst binder exhibited poor electrochemical performances, along with severe delamination. The single cell performances became worse at elevated temperatures. The big electrochemical difference between these two MEAs with the same electrode integrity indicates that a stable interface between a membrane and electrodes played a more important role in determining MEA performances than the methanol permeability of membrane materials does. The difference in the single cell performances became smaller as SPI-BT binder content increased. The HC-MEA performances reached levels similar to or somewhat higher than that of the reference MEA at 90 °C. It is believed that this electrochemical enhancement resulted from the synergistic effect of improved catalyst utilization and lowered interfacial resistance with SPI-SiO2 membrane after adding SPIBT. In addition, the cell voltage of HC-MEAs with higher SPI-BT content decreased with slower slope at the highest current density region (concentration polarization region).19 It means that the mass transfer of methanol to Pt-Ru catalyst was enhanced depending on the SPI-BT, in spite of the formation of compact pore packing structure. The relatively improved fuel supply in the HC-MEAs combining with the positive effect of catalyst utilization and stable interface formation leads to electrochemical enhancement in DMFC performance. At present, our work has been focusing on the evaluation of long-term DMFC performance based on the HC-MEA using SPI-BT binder for the anode. The application of SPI-BT to the 8224 DOI: 10.1021/la900406d

cathode and the development of other high conductive hydrocarbon binder materials with their related results will soon be reported.

Conclusions The following conclusions can be drawn from this study. (1) SPI-BT with a chemical structure similar to a membrane matrix in acid form was successfully fabricated in Et3N salt form via thermal solution imidization and blended with PFSI from 50 to 100 wt % to apply as an alternative anode catalyst binder material to the conventional PFSI. SPI-BT exhibited peculiar solubility behaviors to specific solvents. SPI-BT was soluble in common polar aprotic solvents for the synthesis but insoluble in 1 M methanol for DMFC operation. In particular, in the catalyst ink formula used in this study (IPA/ water = 3:1, weight ratio), SPI-BT existed in the colloidal state, where its particle size was much smaller than that of PFSI owing to the shielding effect of Et3N on the intermolecular interaction of SPI-BT chains. However, the salt form SPI-BT led to the reduction of proton conductivities in the blends with PFSI. (2) The use of SPI-BT as an aqueous catalyst binder significantly influenced the size of the catalyst agglomerates and the characteristics of the pores derived from the agglomerate. The agglomerate size became small after adding the high content of SPIBT into the ink containing Pt-Ru/CB particles. Also, SPI-BT with small colloidal size penetrated the agglomerates more easily than PFSI. The pore filling level increased with high SPI-BT contents, particularly in the primary pore. Thus, a physical network was formed for high utilization of Pt-Ru catalysts mainly located on the wall of the primary pore. (3) HC-MEA containing only SPI-BT exhibited higher electrochemical active surface area of the anode than the reference MEA containing only PFSI did. It verified that SPI-BT contributed to improved catalyst utilization. In addition, high compatibility of the anodes containing SPI-BT with a membrane Langmuir 2009, 25(14), 8217–8225

Lee et al.

Article

material and enhanced mass transfer of methanol to Pt-Ru catalyst suppressed the reduced proton conductivity of SPI-BT and, consequently, led to electrochemical performances of single cells similar to or higher than that of HC-MEA using PFSI.

Langmuir 2009, 25(14), 8217–8225

Acknowledgment. This work was supported by the Korea Foundation for International Cooperation of Science & Technology (KICOS) through a grant from the Korean Ministry of Education, Science and Technology in K20701010356-07B010010610. We are thankful to Mr. Y. H. Cho at Seoul Nation University for his help with CV measurements.

DOI: 10.1021/la900406d

8225