Self-assembling multi-layer Pd nanoparticles onto Nafion™ membrane to reduce methanol crossover

Self-assembling multi-layer Pd nanoparticles onto Nafion™ membrane to reduce methanol crossover

Colloids and Surfaces A: Physicochem. Eng. Aspects 262 (2005) 65–70 Self-assembling multi-layer Pd nanoparticles onto NafionTM membrane to reduce met...

324KB Sizes 0 Downloads 8 Views

Colloids and Surfaces A: Physicochem. Eng. Aspects 262 (2005) 65–70

Self-assembling multi-layer Pd nanoparticles onto NafionTM membrane to reduce methanol crossover Haolin Tang a , Mu Pan a,∗ , Sanping Jiang b , Zhaohui Wan a , Runzhang Yuan a a

b

State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan, PR China Fuel Cells Strategic Research Program, School of Mechanical and Production Engineering, Nanyang Technological University, 50 Nanyang Ave. 639798, Singapore Received 4 January 2005; received in revised form 3 April 2005; accepted 3 April 2005 Available online 9 June 2005

Abstract Methanol crossover through the proton exchange membranes (PEMs, e.g., NafionTM membrane) is one of the major obstacles that currently prevent the widespread commercial applications of direct methanol fuel cell (DMFC). In this paper, multi-layer self-assembly NafionTM membranes (MLSA NafionTM membranes) were prepared by alternately assembling charged Pd particles and Nafion ionmers onto NafionTM membranes. The Pd particles, size of about 1.8 nm in average, are charged by PDDA ionomers with zeta potential of 30 mV (pH value of 8.5). The Pd loading of the first-layer MLSA NafionTM membranes was 0.63 ␮g cm−2 , and the surface coverage of the Pd nanoparticles on the NafionTM membrane was estimated as 22%. After 5-double-layer Pd particles/Nafion ionomers assembling, the Pd loading reached to 2.86 ␮g cm−2 . The methanol crossover current of the original NafionTM membranes and 1-double-layer, 2-double-layer, 3-double-layer, 4-double-layer, 5-double-layer MLSA NafionTM membranes were 0.0495, 3.87E−3, 1.38E−3, 7.32E−4, 5.16E−4 and 4.25E−4 A cm−2 , respectively, corresponding conductivities of 0.112, 0.110, 0.105, 0.094, 0.087 and 0.081 S cm−2 . This satisfactory performance has given the MLSA NafionTM membranes a promised prospect of using as proton exchange membrane in direct methanol fuel cells. © 2005 Elsevier B.V. All rights reserved. Keywords: DMFC; Methanol crossover; Self-assembly; Nanoparticle

1. Introduction Fuel cells utilizing perflourosulfonate ionomers membranes, such as NafionTM membrane, have received much attention because they provide high power density at relatively low operating temperature. Compared to fuel cell systems using reformed H2 from methanol, direct methanol fuel cells (DMFC) have advantages of simple system design and cell operation [1,2]. Two major obstacles that currently prevent the widespread commercial applications of DMFCs are low activity of reported electro-oxidation catalysts and crossover of methanol through the proton exchange membranes (PEMs) [3,4]. It has been realized that methanol ∗

Corresponding author. Fax: +86 27 8787 9468. E-mail address: [email protected] (M. Pan).

0927-7757/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfa.2005.04.011

transported through the PEM would be oxidized at the cathode. Such an oxidation reaction lowers the cathode reactant. If a reaction intermediate, such as carbon monoxide, adsorbs onto the catalyst surface, the cathode will be poisoned too, which further lowers its performance [5]. Suppression methanol crossover in NafionTM has attracted intensive attention worldwide. Jia et al. [6] impregnated NafionTM membrane with poly(1-methylpyrrole) by in situ polymerization. The impregnation reduced the methanol crossover but also decreased the proton conductivity. Composite membranes such as sol–gel derived Nafion/silica [7,8], Nafion/zirconium phosphate [9] and Nafion/cesium ions [10] have also been investigated. To achieve significant reduction in the methanol permeability, oxide content has to be high (e.g., 20 wt% silica in the case of Nafion/silica composite [7]). This in turn affects the proton conductivity and other properties such as mechan-

66

H. Tang et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 262 (2005) 65–70

ical stability. Doping NafionTM membrane with cesium ions reduced the methanol permeability but its conductivity also decreased [10]. Using Pd thin films sandwiched between NafionTM membranes [11] and deposition of Pd nanoparticles through ion exchange and chemical reduction [12] were shown to reduce the methanol crossover. Unfortunately, the Pd film increased the overall cell resistance. The dispersed Pd particles through ion exchange and reduction affect the microstructure of the NafionTM membrane, resulting in the reduced cell performance and stability. We have shown recently that charged Pt nanoparticles [13–15], as well as charged Pd nanoparticles [16], can be selfassembled onto the surface of NafionTM membrane, forming a monolayer structure. The self-assembly of these nanoparticles is most likely due to the Coulomb interaction between the positively charged nanoparticles and the sulfonic acid function sites, SO3 − , at the membrane surface. Once the sulfonic acid function sites assembled, namely “sealed” by charged particles, the crossover of methanol through NafionTM membrane has a drastic reduction. In this present paper, we probe the multi-layer self-assembling behavior of charged Pd particles on NafionTM surface and the methanol crossover of the as-resulted multi-layer self-assembly NafionTM membranes (MLSA NafionTM membranes).

2. Experimental 2.1. Chemicals Polydiallyldimethylammonium chloride (PDDA; Fig. 1a, 20 wt%) with average molecular weight of 5000 and PdCl2 (purity, 99.9%) were received from Aldrich Co. NafionTM 112 membranes and Nafion ionomers (Fig. 1b, EW 1100)

were purchased from Du Pont Co., EtOH (purity, 99.9%) and other reagents were obtained from Shanghai Medicines Co. All reagents employed were of analytical grade without further purification. Distilled water (18.0 M cm) was used in this work. 2.2. Preparing of charged Pd nanoparticles and MLSA NafionTM membrane Charged Pd nanoparticles were prepared by reducing the metallic ions with an alcohol in the presence of poly(diallymethylammonium chloride) ionic polymers, PDDA. PDDA solution (0.002 mol L−1 , 80 mL) was put in a three-neck flask under intensive stirring for 10 min and then mixed with appropriate amount of PdCl solution (0.02 mol L−1 , 8 mL) for another 10 min. Sixty millilitres of EtOH was added to the solution under continuous stirring. The pH of the solution was adjusted to 7.5 by adding NaOH. The solution was then refluxed at ∼84 ◦ C in temperaturecontrolled water bath. NafionTM 112 membrane was treated according to the standard procedure [17] of 30 min in 5 wt% H2 O2 solution at 80 ◦ C, 30 min in distilled water at 80 ◦ C, 30 min in 8 wt% H2 SO4 solution at 80 ◦ C and, finally, 30 min in distilled water at 80 ◦ C again. The self-assembly of Pd-PDDA onto NafionTM membrane was carried out by immersing the pretreated membrane in H2 O dispersed Pd nanoparticles at room temperature for 30 min. The pH of the solution was adjusted to 7.5 by adding NaOH. Then, the membrane was rinsed in distilled water followed by immersed in 0.5 mol L−1 Nafion aqueous solution. Then, a Nafion ionomers/charged Pd particles double-layer was assembled onto NafionTM membrane. This procedure was repeated for five times to fabricate multilayer self-assembly NafionTM membranes (MLSA NafionTM membranes). After self-assembly, the membrane was treated in 8 wt% H2 SO4 solution at 80 ◦ C for 30 min, followed by rinsing in distilled water at 80 ◦ C for 30 min to recover it to H+ -form. 2.3. Methods

Fig. 1. Molecule structure of (a) PDDA ionomer and (b) NafionTM membrane and Nafion ionomers.

Morphology of the Pd particles was examined by HRTEM (JEM-2010FEF). Specimens were prepared by first diluting water dispersed Pd particles and then placing a drop of the solution on a thin carbon film supported by a copper grid. Zeta potential of the resulting Pd nanoparticles was examined by zeta potential analysis set (ZetaPALS, BIC). In situ UV–vis spectrometer (UV-2550, Shimadzu, Japan) was carried to supervise the self-assembly procedure, pure NafionTM membrane were used as reference sample. Tapping-mode atomic force microscopy (TP-AFM) was conducted with a SPI3800N microscope (Seiko Instruments Inc.). Atomic adsorption analysis was employed to analyze the Pd loading of the MLSA NafionTM membranes by soaking it in a 2/3HCl + 1/3HNO3 to dissolve Pd. Proton

H. Tang et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 262 (2005) 65–70

conductivities of MLSA NafionTM membranes were measured by using an impedance analyzer. The MLSA NafionTM membranes were recovered to H+ -form, and then fixed in a four-point probe cell consisted of two platinum wire outer current-carrying electrodes (distance 3 cm) and two platinum wire inner potential-sensing electrodes (distance 1 cm). Methanol permeabilities of the MLSA NafionTM membranes were measured using a diffusion cell consisted of two compartments that were separated by vertical membrane. One compartment of the cell had 2 M MeOH solutions and the other compartment was filled with 1 M sulfuric acid supporting electrolyte. Solutions in both cell compartments were vigorously stirred during the experiments. Polished glassy carbon electrode was used as working electrode. Pt foil was used as counter electrode and saturated calomel electrode (SCE, 0.241 V, versus SHE) was used as reference electrode and placed close to the glassy carbon electrode through a Luggin capillary. All three electrodes were placed in the 1 M sulfuric acid compartment. By applying a dynamic potential from 0.1 to 0.5 V versus SCE at 1 mV s−1 on the working electrode and the limiting methanol oxidation current or crossover current measured voltammetrically was used as an indication of the methanol crossover rate [18–20].

3. Result and discussion 3.1. Synthesis of the charged Pd nanoparticles Fig. 2 shows the UV–vis spectrum of Pd nanoparticles synthesis procedure as a function of refluxing time. Before the reflux, the spectra were characterized by two absorbance bands at 209 and 236 nm. With the increase in the reflux time, the intensities at 209 and 236 nm decrease and two new absorbance bands appear at 227 and 286 nm. This indicates that the absorbance bands at 209 and 236 nm are characteristic of Pd2+ ions in PDDA and the appearance of the bands at

Fig. 2. UV–vis spectrum of Pd nanoparticles synthesis procedure as a function of refluxing time.

67

227 and 286 nm indicates the formation of Pd0 -PDDA. The intensities of the bands at 236 and 286 nm grow as the reflux time increases. At t = 8 min, absorbance bands at 209 and 236 nm disappeared, while there was no further increase in the intensities at 227 and 286 nm, indicating the completion of the reduction reactions. Shown in Fig. 3 is the high-resolution TEM micrograph of resulting Pd nanoparticles (refluxed for 13 min). The TEM image indicated the good dispersion of Pd nanoparticles and the average particle size was estimated as 1.8 nm. The zeta potential analysis of the charged particles with pH value of 8.5 has shown that the potential was 30 mV, indicating Pd particles were indeed modified and charged by PDDA ionomers. The specific advantage of the positive charged nanoparticles is that they could be anchor to the sulfonic acid function group, –SO3 − , on the membrane surface. Especially, the Pd particles of 1.8 nm are small than SO3 − cluster (about 4 nm) but larger than SO3 − cluster-bridge channels (about 1 nm) [21,22]. It means that these particles can anchor into SO3 − clusters without entering into the NafionTM membrane. 3.2. Multi-layer self-assembly behavior of Pd nanopatricles The possible self-assembly mechanism of Pd nanoparticles and NafionTM membrane (or Nafion ionomers) is shown in Fig. 4. The attraction is due to the selective force between –SO3 − clusters on the surface of NafionTM membrane and PDDA inomers modified on the Pd particles. The Pd-assembled-onto-membrane procedure in each doublelayer can be looked as the same. Therefore, it is important to determine the first-layer assembling loading of Pd nanoparticles onto NafionTM membrane. After immersing the pretreated membrane in H2 O dispersed Pd nanoparticles at room temperature for 30 min, the membrane was recovered to H+ -

Fig. 3. High resolution TEM micrograph of as-prepared Pd nanoparticles.

68

H. Tang et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 262 (2005) 65–70

Fig. 4. Possible self-assemble mechanism of Pd nanoparticles and Nafion ionomers on NafionTM membrane. (a) Self-assembly of PDDA stabilized Pd particles to NafionTM membrane surface and (b) self-assembly of Nafion ionomers to Pd particles on NafionTM membranes.

form and then soaked in a 2/3HCl + 1/3HNO3 to dissolve Pd and determine the first-layer assembling Pd content. The value was 0.63 ␮g cm−2 . The distribution or coverage of Pd nanoparticles assembled on the NafionTM membrane can be estimated by assuming the monolayered structure and spherical shape of the Pd nanoparticles [14,15]:  2  3 4 1 d LPd = 2ρ (1) π d+l 3 2  2  2 d 1 SPd = π × 100% (2) d+l 2 where LPd is Pd loading, SPd the surface coverage by Pd nanoparticles, d the size of Pd nanoparticles, l the distance between Pd nanoparticles (edge to edge) and ρ is the density of Pd (12.02 g cm−3 ). The number 2 in the equation represents two sides of the membrane. Using d = 1.8 nm and loading of 0.63 ␮g cm−2 as measured for the selfassembled Pd nanoparticles on NafionTM membrane (half of the self-assembled membrane), the distance, l, between Pd nanoparticles was ∼1.6 nm and the surface coverage of the Pd nanoparticles on the NafionTM membrane was 22%. A probable reason for this unsatisfied surface coverage is the EW value (weight of Nafion resin per mol –SO3 − ) of the NafionTM membrane, which is 1100, corresponding to the –SO3 − content in the NafionTM membrane of 7.27%. After immersing in H2 O dispersed Pd nanoparticles, the membrane was rinsed in distilled water followed by immersed in 0.5 mol L−1 Nafion aqueous solution for another 30 min to assembly Nafion ionomers. This double-layer assembly was repeated five times. During each double-layer assembly, UV–vis spectrometer was carried to observe the Pd loading change. As shown in Fig. 4, the intensities of absorbance bands at 227 and 286 nm grown after each assemble procedure, indicating the increase of Pd loading. The

absorbance bands of NafionTM membrane did not appear in the spectrum, this was because its absorbance was shielded by the reference sample, NafionTM membrane, which was used all along the test (Fig. 5). The NafionTM membrane assembled with 5-double-layer was recovered to H+ -form to measure its irreversible Pd loading. The value was 2.86 ␮g cm−2 , slightly lower than five times of 1-double-layer MLSA membrane (0.63 ␮g cm−2 ). It may be ascribed to that, during 2- to 5-double-layer assemblies, the surface coverage cannot match with 22%, which is the surface coverage of the first double-layer assembly. Shown in Fig. 6 is AFM micrograph of 5-double-layer MLSA membrane, which is also recovered to H+ -form. It is shown that the Pd nanoparticles were dispersed on the Nafion surface compared to pure NafionTM membrane [23]. The diameter of most of the Pd particles was about 20 nm, a value considerably larger than the diameter of Pd nanoparticles observed by

Fig. 5. UV–vis spectra of the multi-layer self-assembly NafionTM membrane as a function of numbers of assembling double-layer.

H. Tang et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 262 (2005) 65–70

Fig. 6. AFM micrograph of 5-double-layer MLSA NafionTM membranes; before test, the sample was recovered to H+ -form to remove the reversible Pd particles.

TEM. If we take into account the thickness of the PDDA layer and convolution effects between the AFM tip and the sample, the difference in the diameter was acceptable [24–26]. However, some of the particles on the MLSA NafionTM membranes surface are larger than 20 nm, it reveals some possible aggregation occurs during the multi-layer self-assembly procedure. 3.3. Performance of the multi-layer self-assembly NafionTM membrane The proton conductivity and methanol permeability of the H+ -form membrane assembled with different double layers are shown in Fig. 7. The conductivities of the 1-double-layer, 2-double-layer, 3-double-layer, 4-double-layer and 5-doublelayer MLSA membranes were 0.112, 0.110, 0.105, 0.094,

69

0.087 and 0.081 S cm−1 , respectively. The results show that if the loading of Pd particles on the MLSA NafionTM membranes is ultra low, in other words, the Pd particle layer are very thin, the assembling of Pd particles has little influence on the proton conductivity. If the thickness reached to a definite value, the conductivity has a slight decrease (e.g., 4-layer self-assembly makes the conductivity decrease to 0.094 S cm−1 ). The methanol crossover current of different double-layer MLSA membranes were 0.0495, 3.87E−3, 1.38E−3, 7.32E−4, 5.16E−4 and 4.25E−4, respectively. The blocking of methanol crossover can be attributed to the peculiar sulfonic-group clusters of the NafionTM membrane and the action of Pd particles sealing the clusters on the Nafion surface by self-assembly. The perfluorosulfonated acid polymer such as Nafion typically has the SO3 − side chains fixed at the C–F backbones. Due to its structure, the phase separation and sulfonic-group clusters occurs between the hydrophilic and hydrophobic regions in hydrated Nafion [12]. Thus, hydrated protons can freely move through the channels produced by the phase separation, leading to high conductivity of the membrane. However, phase separation simultaneously gives channels for methanol and water molecules to pass through under the driving forces of concentration, pressure gradients and electro-osmosis. As shown early, the development of ionomers with low methanol diffusivities without compromising the migration freedom of hydrated protonic clusters ions has proven to be very challenging [27,28]. The drastic reduction in the methanol crossover current demonstrates the effective blocking of the sites for the methanol crossover through the self-assembly of the PdPDDA nanoparticles at the NafionTM membrane surface. Proton conductivity and methanol crossover are very important to the DMFC performance. Considering the change of proton conductivity and methanol crossover, the 3-double-layer self-assembly is preferable. In this condition, the methanol crossover decreased to 0.86%, and the conductivity remained 83.9% comparing to original NafionTM membrane. 4. Conclusions Methanol-blocking proton exchange membrane was prepared by self-assembly multi-layer Pd particles on the NafionTM membrane surface in this paper. The drastic decrease of methanol crossover demonstrated the feasibility of this process especially 3-double-layer of Pt particles and Nafion ionomers was self-assembled. In this condition, there was slightly adverse effect on the proton conductivity of the original NafionTM membrane but the methanol crossover had a considerable decrease of 99.14%. Acknowledgement

Fig. 7. The proton conductivity and methanol permeability of the multilayer self-assembly NafionTM membrane; before the tests, the samples were recovered to H+ -form.

The work was financially supported by the Doctoral Program Foundation of Ministry of Education (20020497001).

70

H. Tang et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 262 (2005) 65–70

References [1] D.J. Kim, M.A. Scibioh, S. Kwak, I. Oh, H.Y. Ha, Electrochem. Commun. 6 (2004) 1069–1074. [2] X.M. Ren, T.E. Springer, T.A. Zawodzinski, S. Gottesfeld, J. Electrochem. Soc. 147 (2000) 466–474. [3] Z.H. Zhou, W.J. Zhou, S.L. Wang, G.X. Wang, L.H. Jiang, H.Q. Li, G.Q. Sun, Q. Xin, Catal. Today 93–95 (2004) 523–528. [4] A. Heinzel, V.M. Barrag´an, J. Power Sources 84 (1999) 70–74. [5] B. Curau, E.S. Smotkin, J. Power Sources 112 (2002) 339–352. [6] N. Jia, M.C. Lefebvre, J. Halfyard, Z. Qi, P.G. Pickup, Electrochem. Solid State Lett. 3 (2000) 529–531. [7] N. Miyake, J.S. Wainright, R.F. Savinell, J. Electrochem. Soc. 148 (2001) A905–A909. [8] P. Staiti, A.S. Aric`o, V. Baglio, F. Lufrano, E. Passalacqua, V. Antonucci, Solid State Ionics 145 (2001) 101–107. [9] C. Yang, S. Srinivasan, A.S. Aric`o, P. Creti, V. Baglio, V. Antonucci, Electrochem. Solid State Lett. 4 (2001) A31–A34. [10] V. Tricoly, J. Electrochem. Soc. 145 (1998) 3798–3800. [11] C. Pu, W. Huang, K.L. Ley, E.S. Smotkin, J. Electrochem. Soc. 142 (1995) L119–L120. [12] Y.J. Kim, W.C. Choi, S.I. Woo, W.H. Hong, Electrochim. Acta 49 (2004) 3227–3234.

[13] H.L. Tang, M. Pan, S.C. Mu, R.Z. Yuan, Acta Chim. Sinica 62 (2004) 127–130. [14] M. Pan, H.L. Tang, S.P. Jiang, Z.C. Liu, J. Electrochem. Soc. 152 (2005) A1081–A1088. [15] M. Pan, H.L. Tang, S.P. Jiang, Z.C. Liu, Electrochem. Commun. 7 (2005) 119–124. [16] H.L. Tang, M. Pan, S.C. Mu, R.Z. Yuan, Chin. Sci. Bull. 50 (2005) 377–379. [17] F.A. Lewis, Int. J. Hydrogen Energy 21 (1996) 461–464. [18] P.S. Kauranen, E. Skou, J. Appl. Electrochem. 26 (1996) 909–917. [19] F. Delime, J.M. L´eger, C. Lamy, J. Appl. Electrochem. 28 (1998) 27–35. [20] Z.G. Shao, X. Wang, I.M. Hsing, J. Membr. Sci. 210 (2002) 147–153. [21] H.L. Yeager, B. Kipling, J. Phys. Chem. 83 (1979) 1836–1839. [22] K.A. Mauritz, R.B. Moore, Chem. Rev. 104 (2004) 4535–4585. [23] W.C. Choi, J.D. Kim, S.I. Woo, J. Power Sources 96 (2001) 411–414. [24] K.C. Grabar, K.R. Brown, C.K. Keating, S.J. Stranick, S. Tang, M.J. Natan, Anal. Chem. 69 (1997) 471–477. [25] Z. Peng, X. Qu, S. Dong, Langmuir 20 (2004) 5–10. [26] H. Yu, D. Jiang, J. Wuhan Univ. Technol. 17 (2002) 38–41. [27] K.D. Kreuer, J. Membr. Sci. 185 (2001) 29–39. [28] J.A. Kerres, J. Membr. Sci. 185 (2001) 3–27.