Effect of design of multilayer electrodes in direct methanol fuel cells (DMFCs)

In a direct methanol fuel cell (DMFC), optimized multilayer electrode design is critical to mitigate methanol crossover and improve cell performance. In this paper, we present a onedimensional (1D) twophase model based on the saturation jump theory in order to explore the methanol and water transport characteristics using various multilayer electrode configurations. To experimentally validate the 1D model, two different membrane electrode assemblies (MEAs) with and without an anode microporous layer (MPL) are fabricated and tested under various cell current density and methanol feed concentration conditions. Then, 1D DMFC simulations are performed and the results compared to the experimental data. In general, the numerical predictions are in good agreement with the experimental data; thus, the 1D DMFC simulations successfully model the effects of the anode MPL that were observed experimentally. In addition to the comparison study, additional numerical simulations are carried out to precisely examine the role of the anode and cathode MPLs and the effect of the hydrophobicity of the anode catalyst layer on the water and liquid saturation distributions inside the DMFCs. This paper demonstrates the quantitative accuracy of the saturation jump model for simulating multilayer DMFC MEAs and also provides greater insight into the operational characteristics of DMFCs incorporating multilayer electrodes.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( ) e1 Available online at www.sciencedirect.com journal homepage: www.elsevier.com/locate/he Effect of design of multilayer electrodes in direct methanol fuel cells (DMFCs) Johan Ko a, Kyungmun Kang a, Sunghyun Park a, Whan-Gi Kim b, Soon-Ho Lee b, Hyunchul Ju a,* a b School of Mechanical Engineering, Inha University, 100 Inha-ro, Nam-gu, Incheon 402-751, Republic of Korea Department of Applied Chemistry, Konkuk Uniersity, 268 Chunwon-daero, Chungju 380-701, Republic of Korea article info abstract Article history: In a direct methanol fuel cell (DMFC), optimized multilayer electrode design is critical to Received 31 January 2013 mitigate methanol crossover and improve cell performance In this paper, we present a Received in revised form one-dimensional (1-D) two-phase model based on the saturation jump theory in order to 12 April 2013 explore the methanol and water transport characteristics using various multilayer elec- Accepted 13 April 2013 trode configurations To experimentally validate the 1-D model, two different membrane Available online 11 May 2013 electrode assemblies (MEAs) with and without an anode microporous layer (MPL) are fabricated and tested under various cell current density and methanol feed concentration Keywords: conditions Then, 1-D DMFC simulations are performed and the results compared to the Direct methanol fuel cell experimental data In general, the numerical predictions are in good agreement with the Water transport experimental data; thus, the 1-D DMFC simulations successfully model the effects of the Microporous layer anode MPL that were observed experimentally In addition to the comparison study, Numerical modeling additional numerical simulations are carried out to precisely examine the role of the anode and cathode MPLs and the effect of the hydrophobicity of the anode catalyst layer on the water and liquid saturation distributions inside the DMFCs This paper demonstrates the quantitative accuracy of the saturation jump model for simulating multilayer DMFC MEAs and also provides greater insight into the operational characteristics of DMFCs incorporating multilayer electrodes Copyright ª 2013, Hydrogen Energy Publications, LLC Published by Elsevier Ltd All rights reserved Introduction Direct methanol fuel cells (DMFCs) have a great advantage over hydrogen-powered fuel cells owing to the increased convenience of liquid methanol fuel Therefore, DMFCs are recognized as an excellent choice for portable and military applications and also well suited for back-up power devices such as uninterruptible power supplies (UPS) and auxiliary power units (APU) However, their commercial success in these areas can only be assured by overcoming several technical barriers; one of these is methanol crossover from the anode to the cathode through the membrane Although a conventional approach, i.e., the use of a dilute methanol solution, does mitigate methanol crossover, this approach requires a large fuel reservoir, which considerably lowers the energy density of the DMFC system Therefore, innovative DMFC technologies must be pursued that target efficient operation of a DMFC with highly concentrated methanol fuel In the literature, considerable efforts have been made to mitigate methanol crossover Recent research focuses on * Corresponding author Tel.: ỵ82 32 860 7312; fax: ỵ82 32 868 1716 E-mail addresses: hxj122@gmail.com, hcju@inha.ac.kr (H Ju) 0360-3199/$ e see front matter Copyright ª 2013, Hydrogen Energy Publications, LLC Published by Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.ijhydene.2013.04.069 1572 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( ) e1 multilayer electrode design to enable proper control of methanol and water transport inside a DMFC [1e6] Liu and Wang [1] and Park et al [2] experimentally explored the role of an anode microporous layer (MPL) on DMFCs operating under highly concentrated methanol Abdelkareem and Nakagawa [3] placed a hydrophobic porous carbon plate on the anode side to retard the rate of methanol transport from the methanol reservoir to the anode They reported a maximum power density of 24 mW cmÀ2 at a 16 M methanol feed solution and room temperature Zhang and Hsing [4] proposed a new integrated anode fuel delivery structure that is made of flexible graphite material and functions as a current collector, flow field, and methanol diffusion layer Using the new design, they achieved improved cell performance at high methanol feed concentrations (12 M or above), as compared to that of a conventional DMFC Li et al [5] designed a porous polytetrafluoroethylene (PTFE) plate as a methanol barrier layer and employed it in a DMFC using high concentration methanol Their data exhibit that the combination of a thick PTFE plate (6.4 mm) and thin NafionÒ 212 membrane optimized the DMFC performance with a 20 M methanol solution Wu et al [6] introduced a microfluidic structured inflow region comprising plural micro flow passages They reported good DMFC performance at a methanol concentration as high as 18 M by incorporating it into a conventional anode flow field design On the other hand, parallel efforts have been underway in DMFC model development Shaffer and Wang [7] theoretically studied methanol and water transport phenomena in multilayer MEAs, using a two-phase model based on the saturation jump theory, which was proposed by Nam and Kaviany [8], Pasaogullari and Wang [9], Weber and Newman [10], and Kang and Ju [11] They carried out one-dimensional (1-D) two-phase DMFC simulations with hydrophobic MPLs in both the anode and cathode sides and numerically demonstrated a principle that shows how a hydrophobic anode MPL reduces the net water flux from the anode to the cathode Later, Shaffer and Wang [12] additionally considered an anode transport barrier (aTB) between the anode diffusion layer (DL) and MPL, numerically showing that a hydrophilic TB is more effective for suppressing methanol transport from the anode channel to the catalyst layer (CL) As reported in the literature, saturation jump models theoretically explained methanol and water transport mechanisms as a function of the different porous layer arrangements in an MEA However, these models have not yet been experimentally validated with DMFC performance data obtained under various methanol feed concentrations In this paper, we first prepare two different MEAs (i.e., either with or without an anode MPL) to explore the impact of the anode MPL on DMFC performance under various methanol feed concentration and current density conditions Then, the experimental data are compared with simulation results calculated using a 1-D two-phase DMFC model After experimentally validating the DMFC model, we perform DMFC simulations to investigate the effects of multilayer electrode design on detailed species-concentration and liquid saturation profiles inside DMFCs as well as the resulting overall cell performance A primary focus of this study is to elucidate the role of MPLs and the effect of the hydrophobicity of the anode CL This paper demonstrates the quantitative accuracy of the saturation jump model for simulating multilayer DMFC MEAs and also provides greater insight into the operational characteristics of DMFCs that incorporate multilayer electrodes Experimental details 2.1 MEA fabrication In this study, two different MEAs were designed and fabricated by combining different anode and cathode DLs For the MEA without an anode MPL, wt% wet-proofed carbon paper (190 mm thick, TGPH-060, Toray Inc.) and Sigracet 25BC (235 mm thick, 40 wt% PTFE) coated with an MPL were employed as the anode and cathode DLs, respectively In the other MEA, a highly hydrophobic MPL was coated on the surface of the anode DL The anode MPL was fabricated via the following procedure: Carbon black (Vulcan XC-72R, Cabot Co.) and a 60 wt% PTFE (Alfa Aesar) suspension were mixed to form a slurry, which was stirred for 20 at room temperature and then coated onto the anode DL substrate using an automatic film coater (EQ-AFA-III-220, MTI) The loading of the anode MPL was mg cmÀ2, which results in an MPL thickness of w10 mm The MPL-coated anode DL was then sintered at 350  C for 20 A schematic of the detailed procedure for MPL fabrication and coating is shown in Fig 2.2 Single-cell assembly and testing To form the single-cell DMFC, the MEA (either with or without the anode MPL coating) with an active area of cm2 was prepared and sandwiched between two identical graphite plates with two path serpentine flow channels Both types of MEA are based on a Nafion 115 membrane with carbonsupported Pt-Ru (Hispec 12100, Johnson MattheyÒ) as the anode catalyst and carbon-supported Pt (Hispec 13100, Johnson MattheyÒ) as the cathode catalyst The assembled single cell was connected to a fuel cell test station (Scitech, Korea Inc.) and operated in the galvanostatic mode at a scan rate of 18 mA sÀ1 for each step For detailed information regarding Fig e Schematic diagram of the MPL fabrication and coating process 1573 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( ) e1 MEA fabrication and single-cell test procedures, refer to papers by Kang et al [13] and Ko et al [14] 3.1 Numerical model Model description and assumptions For this study, a 1-D two-phase DMFC model that was previously presented by Ko et al [14,15] is incorporated with a saturation jump model developed in our previous study [11] The combined 1-D DMFC model is then applied to a 1-D computational domain of a DMFC MEA comprising a DL, MPL, and CL on both the anode and cathode sides and a membrane, as shown in Fig along with the detailed species transfer processes The main assumptions invoked in the present model are as follows: (1) The gas phase obeys the ideal gas law, which is valid because all the gases in the DMFC are maintained at a low pressure relative to their respective critical pressures (2) A temperature gradient along the cell thickness is neglected in the 1-D model due to the thin MEA used in the DMFC (3) An isotropic and homogeneous porous diffusion layer is assumed and is characterized by effective porosity and permeability (4) The effect of CO2 blockage on cell performance is negligible (5) Complete consumption of methanol at the cathode CL after it crosses over the membrane from the anode to the cathode is assumed This assumption is reasonable because the methanol oxidation reaction (MOR) is more facile than the oxygen reduction reaction (ORR) and hence the methanol crossing over the membrane is expected to be completely oxidized at the cathode In addition, Wang et al [16] experimentally found the complete oxidation of crossed-over methanol at the cathode (6) The effects of O2 and CO2 crossover through the membrane are neglected 3.2 and gas phases can be described by using MaxwelleStefan’s multicomponent diffusion equation The final form of the methanol equation in the anode and membrane regions can be written as follows: Species balance equations    I I g g l l l l ỵ Nmem ct DM VclM D Vc c À c À H2 O t M M M I 6F 6F ỵ Nmem ỵ ¼ g M 6F clt À clM kH ct À clM  clM (1) In the above equation, the methanol flux that passes , can be driven by the electrothrough the membrane, Nmem M osmotic drag (EOD) due to proton flux and diffusion that arise from the methanol concentration gradient across the membrane according to the following equation: Nmem M ẳ Nmem;eod M ỵ Nmem;diff M  cl  I mem M CL ẳ nd;M ỵ DM F dmem (2) The diffusion flux of oxygen through the cathode side of the cell can be expressed using Fick’s law By considering the ORR and additional oxygen consumption due to methanol crossover in the cathode CL, we achieve the following equation: I þ Ixover g g;eff g NO2 ¼ DO2 VcO2 ¼ À 4F (3) Liquid saturation equations can be obtained by combining the mass conservation equation for the liquid phase and Darcy’s law to describe liquid flow in porous media The final forms of the liquid-transport equations for the anode and cathode regions are: l clM NlH2 O;a À clt Dl;eff Ka l 2scos q dJ M VcM Vs ¼ ÀMWM kr l l l ds m ct À cM   I ỵ Nmem MWH2 O H2 O 6F and ! rl rl   Kc l 2scos q dJ 3I ỵ 2Ixover Vs ẳ MWH2 O ỵ Nmem kr H2 O l rc ds m 6F (4) (5) respectively In Eq (4), Nmem H2 O denotes the water flux across the membrane due to EOD and diffusion and the hydraulic pressure gradient, which is represented as follows: mem;pl Detailed derivation of the species balance equations was reported by Ko et al [15]; thus, only a brief summary of these equations is provided Methanol diffusion in both the liquid mem;eod ỵ Nmem;diff NH2 O Nmem H2 O ¼ NH2 O H2 O I rmem la À lc Kmem plc pla ẳ nd;H2 O ỵ Dmem H2 O F EW dmem MWH2 O nl dmem Fig e Schematic of the computational domain and species transport phenomena of the 1-D DMFC model (6) 1574 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( ) e1 When Eqs (4) and (5) are solved for liquid saturation, the saturation jump model is applied to the interface of two different porous layers (e.g., an aDL and anode MPL) According to Darcy’s law, the liquid pressure should be continuous at the interface, which renders both the gas and capillary pressures continuous at the interface as well As a result, the liquid saturation across the interface of two different porous layers becomes discontinuous, as represented by the following equation: pc;A ¼ pc;B (7a)  1=2  1=2 Á Á À A B scosqA ị J sint;A ẳ scosqB ị J sint;B KA KB (7b) dmem kmem (8) In Eq (8), and hc represent the anode kinetic loss due to the MOR and the cathode kinetic loss due to the ORR, respectively; these values are calculated using electrochemical kinetics expressions derived from the ButlereVolmer equation, as follows:   aa Fha RT   I ¼ iref 0;a aa Fha cM jaCL ỵ Kreac exp RT cM jaCL exp (9) and I ỵ Ixover ẳ iref 0;c  !   cO2 cCL ac Fhc À sc Þ exp ref RT cO2 (10) (13) and sjcBC 0:1; i < 1000 A=m2 > > > >  1:5 < I ¼ 0:1 ; i > 1000 A=m2 > > 1000 A=m2 > > : maximum 0:9 (14) for the anode and cathode, respectively 3.4 where sint,A and sint,B denote the liquid saturations of the porous layers, A and B, respectively; these values are different due to the different porosities (εA vs εB), permeabilities (KA vs KB), and contact angles (qA vs qB) of the two layers Under the given current density, I, the cell voltage is determined from the theoretical thermodynamic potential, anode and cathode over-potentials, membrane ohmic resistance, and contact resistance, as follows: Vcell ¼ U0 À ỵ hc IRcontact I sjaBC ẳ 0:8 Numerical procedures The species equations described in the previous section are implemented using a FORTRAN code and solved via the following procedure: Once the computational domain is initialized, the methanol transport equation is solved followed by calculation of the cell voltage from the known current density and methanol crossover rate Next, the liquid saturation and oxygen-transport equations are solved in succession After the transport equations have been calculated, the related parameters are updated The iterations proceed until the methanol concentration at the anode CL satisfies the convergence criterion (