Applied Energy 106 (2013) Contents lists available at SciVerse ScienceDirect. Applied Energy

Transcription

1 Applied Energy 106 (2013) Contents lists available at SciVerse ScienceDirect Applied Energy journal homepage: A sandwich structured membrane for direct methanol fuel cells operating with neat methanol Q.X. Wu, T.S. Zhao, R. Chen, L. An Department of Mechanical Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong Special Administrative Region, China highlights " A sandwich structured membrane for DMFCs operating with neat methanol is proposed. " The membrane offers better water management for DMFCs operating with neat methanol. " The sandwich structured membrane enables improvements in cell performance. article info abstract Article history: Received 31 October 2012 Received in revised form 4 December 2012 Accepted 3 January 2013 Keywords: Fuel cell Direct methanol fuel cell Neat methanol Sandwich structured membrane Water content PtRu Water starvation at the anode represents a challenging issue in the development of direct methanol fuel cells (DMFCs) operating with neat methanol. To tackle the issue, a multi-layered membrane, consisting of an ultra-thin reaction layer sandwiched between two thin membranes, is proposed and developed. The reaction layer is composed of well-dispersed PtRu catalysts, SiO 2 nanoparticles and Nafion ionomers. During the fuel cell operation, the methanol permeated from the anode catalyst layer and the oxygen permeated from the cathode catalyst layer meet and react in the reaction layer of the sandwich structured membrane to form water and CO 2. The produced water is then maintained at a relatively high level by the hygroscopic SiO 2 nanoparticles in the sandwich structured membrane. As a result, such a created water source at a high concentration level can supply the water required not only for the anode methanol oxidation reaction but also for membrane hydration. The performance characterization demonstrates that the DMFC with the sandwich structured membrane results in much higher performance than that with a single layer Nafion membrane does. Ó 2013 Elsevier Ltd. All rights reserved. 1. Introduction The rapid development of consumable electronic devices in recent years calls for innovation in portable power sources with high specific energy. Among existing alternative technologies, direct methanol fuel cells (DMFCs) have attracted the most attention, mainly due to their unique features including relatively simple and compact system design, low noise and operating cost, high energy-conversion efficiency and specific energy, ease in fuel handling, environmental friendliness and low-temperature operation [1 12]. These advantages make this type of fuel cell become a leading candidate to replace batteries in portable applications including notebooks, mobile phones and personal digital assistances. However, the widespread commercialization of DMFC technology is still obstructed by several challenging technical issues such as sluggish kinetics of methanol oxidation reaction (MOR) [13] in Corresponding author. Tel.: ; fax: address: metzhao@ust.hk (T.S. Zhao). the anode electrode, methanol crossover [14] from the anode to the cathode and the water flooding problem in the cathode electrode [15,16]. To reduce the methanol crossover and the resulting mixed-potential at the cathode catalyst layer (CCL), operating a DMFC with diluted methanol solution (typically lower than 4 M) is the most common way in practice. Such an operation, however, inevitably sacrifices the original advantage of high specific energy (4900 W h L 1 ) [11]. Moreover, as a significant amount of water is introduced to the anode in the diluted methanol operation, the large rate of water crossover from the anode to cathode will aggravate the water flooding problem at the cathode, lowering the cell performance [17]. Recently, in order to increase the specific energy, researchers [18 25] have turned their attention to the development of novel strategies that can suppress methanol crossover under the concentrated methanol operation. These strategies include adding a compact microporous layer (MPL) [18], employing an additional porous layer with a high transfer resistance between the fuel reservoir and the anode flow field [19,20], developing an innovative microfluidic anode flow field [21] and introducing a pervaporation membrane /$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.

2 302 Q.X. Wu et al. / Applied Energy 106 (2013) to control the delivery rate of highly-concentrated methanol solution [22 25]. Unfortunately, although the methanol-crossover rate can be dramatically reduced by the above proposed approaches, the cell performance at the highly-concentrated methanol operation also suffers from an unexpected reduction as compared to the diluted-methanol operation. It is believed that the water shortage problem in the anode electrode is the main reason [11,26]. In particular, when feeding neat methanol to the anode, one of the reactants for the anode MOR, water, is no longer contained in the fed fuel. Under such a circumstance, the water shortage problem at the anode becomes the worst. To complete the anode MOR at the neat-methanol operation, one of the solutions is that the produced water at the CCL spontaneously diffuses through the membrane to the anode by utilizing the water concentration gradient between the anode and cathode. As a consequence, the water transport flux from the cathode to anode and the water concentration in the anode catalyst layer (ACL) become factors that limit the anode MOR performance at the neat-methanol operation. In addition, it is worth mentioning that the proton transfer resistance of the membrane is also related to water concentration within the membrane electrode assembly (MEA) and the proton conductivity of the membrane increases with the water content. Therefore, elevating the water concentration across the MEA is also critically important for the maximization of the cell performance at the neat-methanol operation by reducing the Ohmic loss. However, only a few efforts [27 34] have been directed to such an important issue of the water management in DMFCs operating with neat methanol. In an attempt to gain understanding of the water transport characteristics through the membrane at the neat-methanol operation, Wu and Zhao [27] experimentally determined the water transport flux from the cathode to anode by changing various structural parameters and found that the water transport flux depended primarily on the designs of the anode gas diffusion layer (GDL), membrane and cathode GDL. To achieve a better water management, Xu et al. [23] introduced a water management layer (WML) between the cathode GDL and the cathode flow field, Masdar et al. [28] proposed adding a hydrophobic air filter next to the cathode flow field, Li and Faghri [29] employed a perforated cover with a low open ratio at the cathode and Park et al. [30] developed a multilayer MEA structure with hydrophilic and hydrophobic layers in the cathode. All these approaches can increase the mass transfer resistance of water from the CCL to the cathode side, which facilitates the water back diffusion from the cathode to anode. Rather than modification of the cathode structure, Wu et al. [31] added a water retention layer, consisting of SiO 2 particles and Nafion ionomer, onto both sides of the membrane to buildup the produced water within the MEA. Inspired by the original work done by Watanabe et al. [35] who proposed adding Pt catalysts into the Nafion membranes for self-humidification in proton exchange membrane fuel cells (PEMFCs), in this work, a sandwich structured membrane, which is made of an ultra-thin reaction layer sandwiched between two thin membranes, is developed to improve the water management. Such a reaction layer comprises well-dispersed PtRu catalysts, nanosized SiO 2 particles and Nafion ionomer. By virtue of highly-active PtRu catalysts, the permeated methanol from the ACL and the permeated oxygen from the CCL can react with each other to generate water and CO 2 in the reaction layer of the sandwich structured membrane. As the transport distance of water is shortened, the water back diffusion can be enhanced. In addition, the hygroscopic SiO 2 nanoparticles in the sandwich structured membrane enable the produced water to be maintained at a relatively high level. As a result, such a created water source at a high concentration level can supply the water required not only for the anode MOR but also for membrane hydration. 2. Basic idea Consider a conventional MEA design shown in the left-hand side of Fig. 1. Under the neat-methanol operation, the water required for the anodic MOR: CH 3 OH þ H 2 O! CO 2 þ 6H þ þ 6e is transported from the cathode, where water is produced from the oxygen reduction reaction (ORR): 3 2 O 2 þ 6H þ þ 6e! 3H 2 O ð2þ ð1þ Fig. 1. Design of a sandwich structured membrane.

3 Q.X. Wu et al. / Applied Energy 106 (2013) Fig. 2. Fabrication process of the MEA with a sandwich structured membrane. and from the oxidation reaction of the permeated methanol at the CCL: CH 3 OH þ 3 2 O 2! CO 2 þ 2H 2 O Clearly, the anode performance and membrane hydration depends on the water flux through the membrane from the cathode to anode. To ensure a higher water flux, a thinner membrane is required. This, however, will result in a higher rate of methanol crossover, degrading the cell performance. Hence, with the conventional MEA architecture design, it is rather difficult to achieve a balance between the relatively high water flux from the cathode to anode and a low rate of methanol crossover in the reverse direction. In an attempt to address the above-mentioned issue, we propose a sandwich structured membrane illustrated in the righthand side of Fig. 1. Such a sandwich structured membrane consists of an ultra-thin reaction layer sandwiched between two Nafion 211 membranes (25 lm). Since the reaction layer is rather thin, the thickness of the sandwich structured membrane is nearly the same as the membrane (50 lm). The reaction layer is made of well-dispersed PtRu catalysts, SiO 2 nanoparticles and Nafion ionomers. The function of the PtRu catalyst is to facilitate the chemical reaction of the permeated methanol from the ACL and the permeated oxygen from the CCL as indicated in Fig. 1. Obviously, with this layer, a new water source is created in the reaction layer of the membrane and the generated water can be transported through a shorter path to the ACL as compared to the conventional design, which is helpful for increasing the water concentration in the ACL. Moreover, owing to the hygroscopic nature of SiO 2, the water produced by Eq. (3) in the reaction layer of the sandwich structured membrane can be maintained at a high concentration level, which assists to decrease the internal resistance and thus improve the cell performance [31,36,37]. 3. Experimental 3.1. MEAs preparation Conventional MEAs The conventional MEA in this work consisted of a membrane with a thickness of 50 lm and two commercially available electrodes from Johnson Matthey Ò. A SGL Ò carbon paper with a 5 wt.% Polytetrafluoroethylene (PTFE) treatment and a MPL was used as GDLs for both electrodes. Carbon supported PtRu (50% Pt 25% Ru) with a loading of 4.0 mg cm 2 and carbon supported Pt (60% Pt) with a loading of 2.0 mg cm 2 were used to form the ACL and CCL, respectively. To fabricate the MEA, the electrodes were hot pressed onto each side of the membrane at 135 C and 4.0 MPa for 3 min. More details about the MEA fabrication technique can be found elsewhere [38]. ð3þ MEAs with a sandwich structured membrane The sandwich structured membranes were fabricated by the decal method [39]. The solution of SiO 2 (5 15 nm, Aldrich Ò ), PtRu (black, Johnson Matthey Ò ) and Nafion ionomer was sprayed onto a Teflon Ò blank. This thin reaction layer was then transferred onto one side of the Nafion 211 membrane by hot pressing at 135 C and 4.0 MPa for 1 min. This Nafion 211 membrane with a thin reaction layer was hot pressed again with another Nafion 211 membrane to form a multi-layered membrane as shown in Fig. 2. One advantage of this fabrication method is that it can prevent the short circuit problem, which may happen in composite membranes formed by casting methods [35,40,41]. The Nafion ionomer and the SiO 2 contents in the reaction layer of the sandwich structured membrane were maintained to be 60 wt.% and 5 wt.%, respectively, while two different loadings of PtRu in this reaction layer were used in the present work: 0.1 mg cm 2 (designated with sandwich membrane I) and 0.3 mg cm 2 (designated with sandwich membrane II). Finally, the sandwich structured membrane was hot pressed between the commercial electrodes at 135 C and 4.0 MPa for 3 min to form the MEA Single cell fixture with a reference electrode Fig. 3 shows the single cell fixture employed in the present work. The MEA was sandwiched between an anode and a cathode flow field. A perforated 316L stainless steel plates with an open ratio of 47.8% and a thickness of 1.0 mm, and a serpentine flow field, having 1.4 mm channel width, 0.9 mm rib width and 1.4 mm depth, were used in the anode and cathode, respectively. To restrict the methanol delivery rate, a gas gap with a thickness of 10 mm, a perforated plate with an open ratio of 15% and a 28 lm thick anion-exchange pervaporation membrane were added to the anode. An organic glass anode fixture with a 5.0-mL built-in fuel reservoir and an aluminum cathode fixture were used to combine the entire cell setup together. To control the cell temperature, an electrical heating rod was inserted into the cathode fixture. A reversible hydrogen electrode with an active area of 1 cm 2 was installed in the single cell fixture as reference electrode to enable discrimination of the anode and cathode potential so that the effect of the sandwich structured membrane on the anode MOR can be studied. The detailed construction of the reference electrode can be found elsewhere [26,27] Determination of the methanol-crossover flux The crossover flux of methanol was determined by the voltammetric method described elsewhere [42]. To create an inert environment, liquid water with a flow rate of 2.0 ml min 1 was fed into the cathode in the measurement. A positive voltage of 0.85 V was then applied between the cathode and the anode to fully oxidize the methanol permeated through the membrane.

4 304 Q.X. Wu et al. / Applied Energy 106 (2013) Fig. 3. Schematic of a DMFC operating with neat methanol and the arrangement of a reference electrode. Fig. 4. Performances of the MEAs with various membranes Electrochemical instrumentation and test conditions An Arbin BT2000 electrical load test station was employed to control the discharging current, record the transient voltage and measure the internal resistance of the cell. The measured internal resistance is a sum of the electronic and ionic resistance of the cell. As the fuel cell fixture is the same for various MEAs, the electronic resistance is identical for each test. Under this situation, the change in the cell internal resistance is caused by the ionic resistance of the membrane % dry oxygen was fed to the cathode and was controlled to maintain a flow rate of 20 sccm by a mass flow meter (Omega FMA-7105E) for all the tests. The operating temperature of the cell was kept at 50 C. 4. Results and discussion The I V curves of the MEA consisting of a membrane, as well as the MEAs consisting of the sandwich membrane I and II are compared in Fig. 4. It is clear that the cell performance improves notably with the employment of the sandwich membrane I; the peak power density increases from about 40 to 46 mw cm 2, nearly 15% improvement. Such an increase in the cell performance arises from several favorable factors that are explained as follows. Firstly, the introduction of the sandwich membrane I can make use of the chemical reaction by the permeated methanol and oxygen to produce water in the reaction layer of the sandwich structured membrane. Since the diffusion length is only one half of the membrane, the generated water can reach the ACL more easily as compared to the conventional design shown in Fig. 1. As a result, both the water concentration in the ACL and the water content within the membrane are increased, which not only increases the anode MOR performance but also decreases the internal resistance. In addition, taking advantage of the hygroscopic feature of the SiO 2 nanoparticles, the produced water in the thin reaction layer of the sandwich structured membrane can be well retained, which helps to increase the overall water content within the MEA and thus further improves the cell performance. Another favorable effect of the sandwich structured membrane is that it can reduce methanol crossover. The methanol-crossover fluxes under the open circuit condition at 30 and 50 C with various membranes are shown in Table 1. It can be seen that at the both temperatures, the methanol-crossover flux decreases with an increase in the PtRu loading in the reaction layer of the sandwich structured membrane, simply because a higher Table 1 Effect of membranes on the methanol-crossover flux. Temperature ( C) Methanol-crossover flux (lmol s 1 cm 2 ) Nafion 212 Sandwich membrane I Sandwich membrane II loading creates a larger mass-transfer resistance of methanol through the membrane. However, it is worth mentioning that the decrease in the methanol-crossover flux with increasing the PtRu loading is fairly small; the methanol-crossover flux at 50 C decreases from to lmol s 1 cm 2 when replacing the membrane with the sandwich membrane II, only about 5% drop. Thus, it is believed that the reduction in methanol crossover with the application of the sandwich structured membranes is not the major reason accounting for the improved cell performance. Moreover, it is interesting to observe in Fig. 4 that increasing the PtRu loading from 0.1 (sandwich membrane I) to 0.3 mg cm 2 (sandwich membrane II), the cell performance decreases significantly; the peak power density drops from about 46 to 37 mw cm 2, nearly 20% decrease. On one hand, an increase in the PtRu loading leads to an increase in the thickness of the PtRu layer, thereby increasing the cell internal resistance. On the other hand, a thicker reaction layer in the sandwich structured membrane can create a large mass-transfer resistance of water from the CCL to ACL, which in turn lowers the water concentration in the ACL. Hence, the cell performance becomes poorer when a too high PtRu loading is used in the sandwich structured membrane. To further investigate the effect of the sandwich structured membrane, transient cell voltages at the discharge current density of 125 ma cm 2 are measured for MEAs with various membranes including the, as well as sandwich membrane I and II; the results are shown in Fig. 5. It can be seen that all cell voltages are rather stable and more importantly, the cell voltage improved significantly with the application of the sandwich membrane I; the cell voltage increases from about 230 to 275 mv, nearly 20% improvement, when replacing the membrane with the sandwich membrane I. The increased voltage with the sandwich membrane I can be attributed to the enhancement in the anode MOR and reduction in the internal resistance. These conclusions are supported by the respective electrode potential shown in Fig. 6 and the transient internal resistance shown in

5 Q.X. Wu et al. / Applied Energy 106 (2013) Voltage (V) Internal resistance (mohm) Fig. 5. Transient voltages of the MEAs with various membranes at the discharge current density of 125 ma cm 2. Fig. 7. Transient internal resistances of the MEAs with various membranes at the discharge current density of 125 ma cm 2. (a) Anode potential (V) (b) Cathode potential (V) Fig. 7. It is seen from Fig. 6 that the anode potential with the sandwich membrane I is much lower than that with the membrane, whereas the change in the cathode potential with these two membranes is rather small. These results suggest that the improved performance arises from the anode MOR not the cathode ORR. Since the catalyst material and the methanol delivery rate remain nearly unchanged, the reduction in the anode potential should be attributed to the increased water concentration level in the ACL as a result of employing the sandwich membrane I. Meanwhile, the internal resistance shown in Fig. 7 also proves the fact that the sandwich membrane I increases the water content within the MEA. Generally, the existence of the reaction layer in the sandwich membrane is supposed to increase the proton transport distance, thus resulting in a higher internal resistance. However, Fig. 7 shows that the measured internal resistance of sandwich membrane I is lower than that of the conventional membrane. This phenomenon can only be attributed to an increase in the water content with the introduction of sandwich membrane I. Furthermore, it is important to note that the cell internal resistance of the sandwich membrane I is 12.8 mx lower than that of the conventional membrane. This small difference in the internal resistance only contributes to 6.4 mv ( = V) increase in the cell voltage. This value is substantially lower than the reduction in the anode potential (about 50 mv) when replacing the membrane with the sandwich membrane I. Hence, the enhanced anode MOR is the main factor leading to the improved cell performance. In addition, it is worth noting in Fig. 5 that an increase in the PtRu loading from 0.1 to 0.3 mg cm 2 in the reaction layer of the sandwich structured membrane leads to a large drop in the cell voltage; the cell voltage decreases from about 275 to 180 mv, nearly 35% decrease. This is because a too thick reaction layer not only increases the water transfer resistance through the membrane, which in turn increases the anode potential as shown in Fig. 6a, but also hinders the proton transfer through the membrane such that a higher internal resistance is observed in Fig. 7 when using the sandwich membrane II. 0.6 Fig. 6. Transient electrode potentials of the MEAs with various membranes at the discharge current density of 125 ma cm 2 (a) anode potential and (b) cathode potential. 5. Concluding remarks Feeding neat methanol directly to a DMFC it is of paramount significance for maximizing the specify energy of a fuel cell system. The key to increase the cell performance under the neat-methanol operation is maintaining a sufficiently high water concentration in

6 306 Q.X. Wu et al. / Applied Energy 106 (2013) the ACL and the membrane such that both the anode MOR performance and the proton conductivity of the membrane can be simultaneously improved. In the present work, we propose a novel sandwich structured membrane, comprising an ultra-thin reaction layer sandwiched by two Nafion 211 membranes, to elevate the water concentration level within the MEA. The core of the sandwich structured membrane is a reaction layer which is formed with highly-active PtRu catalysts, SiO 2 nanoparticles and Nafion ionomers. With the presence of PtRu catalysts, the methanol permeated from the ACL and the oxygen permeated from the CCL meet and react in the reaction layer of the sandwich structured membrane to form water and CO 2. Thanks to the hygroscopic characteristic of SiO 2 nanoparticles, the water produced in the reaction layer can be maintained at a high concentration level. Thus, this created water source with high water content can provide the water required not only for the anode MOR but also for membrane hydration. We tested and compared the MEAs with a sandwich structured membrane and a single layer membrane, and concluded that the novel membrane with a PtRu loading of 0.1 mg cm 2 can accelerate the MOR and reduce the internal resistance, thus upgrading the cell performance. Acknowledgements The work described in this paper was fully supported by grants from the Research Grants Council (Project No. HKUST9/CRF/11G) and from the Innovation and Technology Fund (Project No. ITS/ 549/09FP) of the Hong Kong Special Administrative Region, China. References [1] Park GG, Yang TH, Yoon YG, Lee WY, Kim CS. Int J Hydrogen Energy 2003;28:645. [2] Shimizu T, Momma T, Mohamedi M, Osaka T, Sarangapani S. J Power Sources 2004;137:277. [3] Liu JG, Zhao TS, Chen R, Wong CW. Electrochem Commun 2005;7:288. [4] Guo Z, Faghri A. J Power Sources 2006;160:1183. [5] Chen R, Zhao TS. Electrochem Commun 2007;9:718. [6] Kamarudin SK, Daud WRW, Ho SL, Hasran UA. J Power Sources 2007;163:743. [7] Baglio V, Stassi A, Matera FV, Blasi AD, Antonucci V, Aricò AS. J Power Sources 2008;180:797. [8] Paust N, Krumbholz S, Munt S, Müller C, Koltay P, Zengerle R, et al. J Power Sources 2009;192:442. [9] Zhu YL, Liang JS, Liu C, Ma TL, Wang LD. J Power Sources 2009;193:649. [10] Zhao TS, Chen R, Yang WW, Xu C. J Power Sources 2009;191:185. [11] Zhao TS, Yang WW, Chen R, Wu QX. J Power Sources 2010;195:3451. [12] Yang WW, Zhao TS, Wu QX. Int J Hydrogen Energy 2011;36:6899. [13] Wasmus S, Kuver A. J Electroanal Chem 1999;461:14. [14] Neburchilov V, Martin J, Wang H, Zhang J. J Power Sources 2007;169:221. [15] Song KY, Lee HK, Kim HT. Electrochim Acta 2007;53:637. [16] Lu GQ, Wang CY. J Power Sources 2004;134:33. [17] Xu C, Zhao TS. J Power Sources 2007;168:143. [18] Lu GQ, Wang CY, Yen TJ, Zhang X. Electrochim Acta 2004;49:821. [19] Nakagawa N, Abdelkareem MA, Sekimoto K. J Power Sources 2006;160:105. [20] Abdelkareem MA, Nakagawa N. J Power Sources 2006;162:114. [21] Wu QX, Zhao TS, Chen R, Yang WW. J Micromech Microeng. 2010;20: [22] Kim HK. J Power Sources 2006;162:1232. [23] Xu C, Faghri A, Li XL. J Electrochem Soc 2010;157:B1109. [24] Feng LG, Zhang J, Cai WW, Liang L, Xing W, Liu CP. J Power Sources 2011;196:2750. [25] Eccarius S, Krause F, Beard K, Agert C. J Power Sources 2008;182:565. [26] Wu QX, Shen SY, He YL, Zhao TS. Int J Hydrogen Energy 2012;37:5958. [27] Wu QX, Zhao TS. Int J Hydrogen Energy 2011;36:5644. [28] Masdar MS, Tsujiguchi T, Nakagawa N. J Power Sources 2010;195:8028. [29] Li XL, Faghri A. J Power Sources 2011;196:6318. [30] Park YC, Kim DH, Lim S, Kim SK, Peck DH, Jung DH. Int J Hydrogen Energy 2012;37:4717. [31] Wu QX, Zhao TS, Chen R, Yang WW. Int J Hydrogen Energy 2010;35: [32] Li XL, Faghri A, Xu C. Int J Hydrogen Energy 2010;35:8690. [33] Wu QX, Zhao TS, Yang WW. Int J Heat Mass Trans 2011;54:1132. [34] Xu C, Faghri A, Li XL. Int J Hydrogen Energy 2011;36:8468. [35] Watanabe M, Uchida H, Seki Y, Emori M. J Electrochem Soc 1996;143:3847. [36] Zhu XB, Zhang HM, Zhang Y, Liang YM, Wang XL, Yi BL. J Phys Chem B 2006;110: [37] Yang HN, Cho SH, Kim WJ. J Membr Sci 2012; :318. [38] Yang H, Zhao TS, Ye Q. J Power Sources 2005;139:79. [39] Song SQ, Liang ZX, Zhou WJ, Sun GQ, Xin Q, Stergiopoulos V, et al. J Power Sources 2005;145:495. [40] Wang L, Yi BL, Zhang HM, Xing DM. Electrochim Acta 2007;52:5479. [41] Joo SH, Pak C, Kim EA, Lee YH, Chang H, Seung D, et al. J Power Sources 2008;180:63. [42] Ren X, Springer TE, Zawodzinski TA, Gottesfeld S. J Electrochem Soc 2000;147:466.