Experimental studies of the effect of cathode diffusion layer properties on a passive direct methanol fuel cell power output

Experimental studies of the effect of cathode diffusion layer properties on a passive direct methanol fuel cell power output

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3 Available online at www.sciencedirect.co...

1MB Sizes 0 Downloads 1 Views

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Experimental studies of the effect of cathode diffusion layer properties on a passive direct methanol fuel cell power output B.A. Braz, V.B. Oliveira**, A.M.F.R. Pinto* Transport Phenomena Research Center (CEFT), Chemical Engineering Department, Faculty of Engineering of University of Porto, Rua Dr. Roberto Frias S/n, 4200 465 Porto, Portugal

article info

abstract

Article history:

A great challenge in a passive direct methanol fuel cell (pDMFC) is how to reduce both

Received 16 November 2017

methanol and water crossover, from the anode to the cathode side, without significant

Received in revised form

losses on its power output. Different approaches including improving the membrane and

15 September 2018

modifying the cell structure and materials have been proposed in the last years.

Accepted 20 March 2019 Available online 12 April 2019

In this work, an experimental study was carried out to evaluate the effect of the cathode diffusion layer (CDL) properties on the power output of a pDMFC. Towards a cost reduction, lower catalyst loadings were used on both anode and cathode electrodes. Since the main

Keywords:

goal was the optimization of a pDMFC using the materials commercially available, different

Passive direct methanol fuel cell

carbon-fibber materials were employed as CDL. The experimental results were analysed

Cathode diffusion layer

based on the polarization curves and electrochemical impedance spectroscopy measure-

Electrochemical impedance spec-

ments with innovative electric equivalent circuit allowing the identification of the different

troscopy

losses, including the activation resistance of the parasitic cathode methanol oxidation.

Water management Methanol concentration Cost

A maximum power density of 3.0 mW/cm2 was obtained using carbon cloth with a lower thickness as CDL and a methanol concentration of 5 M. © 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction The demand for efficient, renewable and sustainable energy sources increased in the last decade, as well as, the energy request of portable devices due to an increase of its functionalities. Additionally, the consumers demand for these new multifunctional devices stimulated researchers and industry to develop advanced portable fuel cells to overcome systematic limitations of conventional batteries [1]. Passive direct methanol fuel cells (DMFCs) with a passive fuel and

oxidant (oxygen from air) supply, through natural transport mechanisms (such as capillarity, diffusion and convection) were identified as the ideal technology to achieve that [2]. Mostly due to a lack of effective miniaturized hydrogen storage systems, a liquid fuel like methanol, which is easy to handle, storage and refill, is the best option to achieve a high power density with an attractive cost-to-power ratio. Despite these advantages and the fact that passive DMFC systems are simpler and compact, some challenges need to be overcome before its massive utilization and commercialization. These include the sluggish kinetics of both anode and cathode

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (V.B. Oliveira), [email protected] (A.M.F.R. Pinto). https://doi.org/10.1016/j.ijhydene.2019.03.162 0360-3199/© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

reactions, methanol and water crossover, higher cost and lower durability [3]. In addition, it is of equal importance to optimize and improve the mass transport phenomena within the different fuel cell layers, namely the diffusion layers. This is due to the fact that, the passive systems do not rely on external devices to supply the fuel and oxidant (such as pumps or compressors), so the reactants delivery and the products removal is performed by natural mechanism, which lead, inherently, to lower mass transfer rates. Towards that, different studies concerning the use of different membranes, anode diffusion layers and lower catalyst loadings on both anode and cathode electrode have been performed in the last years [3e9]. As known, the methanol crossover is one of the main parameters that negatively affects the performance of a DMFC. In addition, the use of highly methanol concentrations is especially recommended for passive feed systems since will lead to higher mass transfer and methanol oxidation rates. However, these concentrations originate a higher methanol crossover and a lower water concentration on the anode side, which may limit in some extent the oxidation of methanol on this side, since water is needed for this reaction. In the last years, the approach used to avoid this was to feed the cell with diluted methanol solutions. Nevertheless, this lead to a higher water concentration gradient between the anode and the cathode side and to water crossover towards the cathode. This excess of water on the cathode side conducts to cathode flooding, which will hinder the oxygen transport on this side. Moreover, higher water crossover rates, also, increase the methanol crossover rate [10e13]. Therefore, in an ideal operation, the anode of a passive DMFC will be feed with neat (pure) methanol, to achieve a higher fuel utilization, and the water needed for the anode oxidation reaction will come from the cathode side by back diffusion. This water is produced at the cathode by the oxygen reduction reaction. Under this operation, the anode performance depends not only on the methanol oxidation reaction, but also, on the water concentration at the cathode side and its transport rate towards the anode. Additionally, some of the water produced at the cathode is consumed by the parasitic methanol oxidation. Hence, the water management is another critical parameter that should be optimized towards an increase of the passive DMFC performance, especially when the cell is operated with pure methanol. It has been found that the physical and chemical properties of the diffusion layers (DLs), such as morphology, thickness, PTFE content and the existence of a microporous layer (MPL) determine these layer structure characteristics and play an important role in the management of the different mass transport phenomena [15e25]. A good DL should have suitable porosity to allow the flow of both reactants and products, hydrophobicity or hydrophilicity to a proper water management, an adequate thickness to decrease the mass transport resistances and thermal conductivity to allow the heat removal. It should be sufficiently rigid to support the membrane, but have some flexibility to maintain a good electrical contact. All these requirements are best met with carbon based materials, such as carbon paper, carbon cloth, mesoporous carbons and carbon nanofibers and nanotubes [18,19,21,23,24]. Li et al. [20] and Xue et al. [25] studied the use of a stainless steel (SS) fiber felt as, simultaneously, cathode

19335

diffusion layer and cathode current collector in a micro passive DMFC. The results showed that this new cathode structure enhanced the water back diffusion towards the anode, preventing the cathode flooding and increasing the oxygen mass transport and promoted lower methanol crossover rates. However, caution should be taken before using this material to replace the carbon based ones, since as was verified when using SS as current collectors, after a long-term operation they suffer from corrosion, which increases the contact resistances and lead to the presence and accumulation of corrosion products on the different fuel cell layers, poisoning them [3]. Towards an ideal water balance and optimized cell performance and based on a previous work performed by the authors regarding the effect of the anode DL on the performance of a passive DMFC [5], the aim of the current work is to evaluate the effect of using different cathode DLs on a passive DMFC. The main goal was to use materials that are commercially available and based on the current state-of-the-art of the materials used as cathode DLs for passive DMFC systems. In order to reduce the cell overall costs and since the catalyst loadings represent a major portion of them, lower catalyst loadings 3 mg/cm2 Pt/Ru and 1.3 mg/cm2 of Pt were used on, respectively, the anode and cathode electrodes. The effect of the cathode DL on the performance of a passive DMFC, with an active area of 25 cm2, was discussed through the polarization curves and EIS measurements, which allowed to identify and quantify the different losses that negatively affect the cell performance [26].

Experimental Fuel cell design The passive DMFC is composed by two acrylic end plates, one open (cathode side) and the other with a fuel reservoir (anode side), two isolating rubber plates, two stainless steel current collector plates (36 perforated holes with a diameter of 6 mm), two diffusion layers, two catalyst layers and a Nafion 117 membrane (Fig. 1). The three-layer membrane (supplied by QuinTech) has an active area of 25 cm2 and a catalyst loading of 3 mg/cm2 of Pt/Ru on the anode side and 1.3 mg/cm2 of Pt on the cathode side. To evaluate the effect of the CDL properties on the pDMFC power output, four different carbon cloths and carbon papers, each one with a different thickness and surface treatment were tested. The two carbon papers used without MPL, CP and CP_T, were 20 ( ±5) wt. % PTFE treated. The characteristics of the different materials tested as cathode diffusion layer are provided in Table 1, all of them are woven and were supplied by QuinTech, with the exception of CC_MPL_E that was acquired to Fuel Cells Etc. The diffusion layers were pressed by a non-bonded mode on the three-layer membrane when assembling the fuel cell.

Material and methods As mentioned the passive DMFC used in this work has an active area of 25 cm2, an open ratio of 41% and was operated at

19336

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

Fig. 2 e Nyquist plot of a pDMFC with the DHE at the cathode side and a methanol concentration of 1 M at the anode for different cell voltages.

Fig. 1 e Schematic representation of the passive DMFC used in this work.

ambient conditions (ambient pressure and temperature). The experimental tests were performed with different methanol concentrations, 1, 2, 3, 5 and 7 M and using an electrochemical test station Zahner Elektrik GmbH &Co. KG. The polarization curves were realized galvanostatically and the impedance measurements potentiostatically. The EIS tests were carried out with an amplitude of 10 mV, a frequency range from 10 mHz to 100 kHz and at different fuel cell voltages (0.45 V, 0.4 V, 0.35 V, 0.3 V and 0.2 V). The EIS measurements were performed in-situ to allow the measurement of the impedance of the fuel cell as a whole. This is the common procedure used to analyse a single fuel cell, since the information provided by this approach helps on the optimization of the design parameters and operating conditions. As the use of a reference electrode in most of the fuel cell systems is difficult and unpractical, in this work a dynamic hydrogen electrode (DHE) was used, by replacing the methanol or air supply by hydrogen. With this approach, it is assumed that the losses associated with the DHE are irrelevant and consequently all the measurements are attributed to the other electrode, the electrode under study. Therefore, when the DHE is used at the cathode, it is possible to study the anode behaviour and obtain the anode spectrum (Fig. 2) and when the DHE is used at the

anode, it is possible to study the cathode behaviour and obtain the cathode spectrum (Fig. 3). Then, performing the impedance measurements with the pDMFC working with methanol and air, it is possible to obtain the overall pDMFC spectrum (Fig. 4) and compare it with the single-cell ones to distinguish the different electrode losses (anode and cathode) on the overall cell response [25]. The impedance spectra was, then, fitted to an electric equivalent circuit (EEC), using the Thales software from Zahner, to quantify the different losses that affect the cell performance.

Results and discussion In this section, the experimental results obtained in this work are presented and explained under the light of the EIS data. Therefore, this section starts with the interpretation of the EIS data using a selected set of values, which present the same trends and patterns of the other results. Then, the effect of using different carbon based materials as cathode DL on the performance of a passive DMFC is discussed. Finally, the effect of the methanol concentration on the fuel cell performance for the selected cathode DL, the one that lead to the best power output, is presented.

Table 1 e Properties of the materials used as cathode diffusion layer (supplied by Quintech and Fuel Cells Etc). Diffusion layer

Type

Thickness (mm)

Porosity

MPL

CC CC_T CC_MPL CC_MPL_E CP CP_T CP_MPL CP_MPL_T

Cloth Cloth Cloth Cloth Paper Paper Paper Paper

0.400 0.425 0.410 0.454 0.110 0.190 0.240 0.340

0.83 0.83 0.80 0.63 0.78 0.78 0.80 0.80

No No Yes Yes No No Yes Yes

Fig. 3 e Nyquist plot of a pDMFC with the DHE at the anode side and air at the cathode, for different cell voltages.

19337

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

Table 2 e Values for the different resistances obtained by fitting the Nyquist plots of the pDMFC (Fig. 5) with the EEC proposed in this work (Fig. 6) for different fuel cell voltages. Voltage (V) 0.45 0.40 0.35 0.30 0.20

Fig. 4 e Nyquist plot of a pDMFC fed with methanol (1 M) at the anode and air at the cathode, for different cell voltages.

EIS measurements and EEC fitting As mentioned, the EIS measurements can provide detailed information about the fuel cell system, allowing to identify and quantify the different voltage losses that negatively affect the fuel cell performance: ohmic, activation and mass transport losses. The EIS data are usually represented by a Nyquist plot, such as the one shown in Figs. 2e4, where the real impedance (Zre) and the imaginary (Zim) one are plotted at the X and Y-axis, respectively. The characteristic and shape of each spectrum, allow identifying the different losses that affect the cell performance. Therefore, different combinations of several elements have to be considered on the equivalent circuit model that was used to model the impedance spectrum. A typical Nyquist plot of a fuel cell can be divided into different regions according to its frequency range. At the high frequencies region, the imaginary impedance axis intercepts the real impedance one at a value called the high-frequency resistance (HFR), which represents the ohmic losses (as shown in Figs. 2e4). As well known, the electrical resistance is time independent and is described through the Ohm's law. In electrical circuits, it is represented by a resistor (R) (Rohmic in the EEC shown in Fig. 5) and due to its frequency independency, its impedance has only a real part, identified by a single point at the real axis (X-axis) of the Nyquist plot, with a value equal to R. The arc presented at the medium frequency region represents the activation losses and is characterized by a decrease of the resistance with a decrease of the voltage (as

Rohmic (U)

RA (U)

RC (U)

RCrossover (U)

0.18 0.18 0.18 0.18 0.18

0.15 0.15 0.15 0.15 0.15

0.47 0.47 0.47 0.32 0.17

7.2 3.1 1.5 1.2 0.7

shown in Figs. 2e4 and Table 2) [26]. The arc at the lower frequency region, usually, represents the mass transport losses and is characterized by an increase of the resistance with a decrease of voltage (as shown in Fig. 2) [26]. However, in some systems, it is not possible to reach the mass transport losses region and therefore the arcs, at the medium and lower frequency ranges, are due to the activation losses, characterized by a decrease of the resistance with a decrease of the voltage (as shown in Fig. 3). Additionally, in some cases, as the one of the current work, it is possible to see more than two arcs, showing that the system under study has an additional resistance that negatively affects its performance as shown in Fig. 4. The plots in this Figure represent the spectrum of the pDMFC operating with methanol and air, and Table 2 displays the values for the four resistances obtained by adjusting the EEC shown in Fig. 5 to the Nyquist plots. It is possible to notice that none of the resistances increase with a decrease of the cell voltage. This allows to perceive that under the operating conditions studied, the pDMFC does not suffer in a significant extent from concentration or mass transport losses and the activation ones are dominant for all the voltages studied. After this result, the next step is to link the different activation losses to the different phenomena/processes that occur in a working pDMFC. It is known that the activation losses are due to the electrochemical reactions, usually the fuel oxidation (anode side) and the oxygen reduction (cathode side), which justify two of the arcs presented at the Nyquist plot. The other arc, the third one, is characteristic of the DMFCs and its due to an additional electrochemical reaction that occurs at the cathode side of these cells, the methanol oxidation due to methanol crossover. The crossover is one of the major drawbacks of these systems owing to its strong negative impact on the cell power output, explained by the formation of a mixed potential at the cathode side, a loss of fuel at the anode side and cathode poisoning by the intermediate compounds

Fig. 5 e Electric equivalent circuit (EEC) used to describe the EIS experimental data of a pDMFC feed with methanol and air.

19338

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

formed due to the incomplete methanol oxidation reaction on the Pt catalyst (cathode catalyst). The resistance due to methanol crossover, which affects mainly the cathode side, can be clearly identified on the cathode spectrum, shown in Fig. 3, where both resistances decrease with a decrease of the cell voltage, a normal behaviour of the activation losses. Therefore, both resistances presented on the cathode spectrum are due to the two reactions that occur at the cathode catalyst. Additionally, it is well known that the crossover decreases with a decrease of voltage [27], and accordingly, the crossover effect, here estimated by the RCrossover, decreases with a decrease of the voltage, as shown in Figs. 3 and 4 and Table 2. Based on the EIS measurements performed for the pDMFC working with a DHE at the anode or cathode side with methanol and air and on the shape of each spectrum, the EEC used in this work, and presented in Fig. 5, consists on a resistance, Rohm (ohmic resistance) connected in series with other three independent circuits. Each one comprising a resistance, RA (activation losses due to the methanol oxidation at the anode), RC (activation losses due to the oxygen reduction at the cathode) and RCrossover (activation losses due to the methanol oxidation at the cathode) in parallel with a constant phase element, CPEA, CPEC and CPECrossover, associated with the capacitance properties of the double-layer interfaces. The plots depicted in Fig. 4 for the pDMFC working with methanol and air show three semicircles for all the voltages tested, that decrease with a decrease of the cell voltage, indicating a reduction of the losses that negatively affect the cell performance with voltage. This can be also verified through the values of the different resistances presented in Table 2, which were obtained by fitting the EEC (Fig. 5) to the EIS data presented in Fig. 4. There is a good agreement between the EEC fitting and the experimental results, revealing that the EEC proposed reproduces with accuracy the system under study. The approach used in this work is new, since as the authors are aware no other work, regarding a pDMFC used this methodology and the EEC here proposed. Therefore, these results are ground-breaking, identifying the different phenomena that affect the pDMFC power output. It should be here highlighted, that the impedance spectra obtained for the other conditions studied followed the same pattern as the ones presented in this sub-section, with similar results for the ECC fitting and resistance values. The EIS data for a voltage of 0.2 V, near the voltage corresponding to the optimum power density for each condition, were selected to be presented in the following sub-sections and will be used to explain the polarization results and the pDMFC behaviour.

higher thickness than the CC_MPL (0.540 mm and 0.410 mm). Fig. 6 shows the polarization curves for the different carbon cloths tested on the cathode side, feeding the cell with three different methanol concentration, 1 M, 2 M and 3 M. A carbon cloth with microporous layer (CC_MPL) was used as anode DL for all the experiments. The values for the ohmic and activation resistances at 0.2 V, as well as, the maximum power density achieved for each configuration are presented in Table 3. As the methanol crossover is a major drawback in DMFCs optimization, it is expected that the values of RCrossover should represent the major negative contribution to decrease the cell performance, even when tailoring the cathode. This happens because in fact, the parasitic methanol oxidation reaction occurs at the cathode catalyst layer using a non-adequate catalyst. The results presented in Fig. 6 and Table 3 confirm these expectations. Generally, for all the concentrations

Carbon cloth as cathode DL To evaluate the effect of using carbon cloth as cathode DL four different types of carbon cloth with different thickness (CC and CC_T) and surface treatment (CC_MPL and CC_MPL_E) were selected. As can be seen in Table 1, CC and CC_T have the same porosity (0.83) but CC_T has a slightly higher thickness (0.425 mm instead of 0.400 mm). The carbon cloths with a microporous layer present lower porosities than the untreated ones (0.80 and 0.63) and the CC_MPL_E a

Fig. 6 e Effect of different carbon cloths as cathode DL on the cell performance for different methanol concentrations: (a) 1 M, (b) 2 M and (c) 3 M.

19339

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

Table 3 e Values for the different resistances of the EEC for the different carbon cloths tested as cathode DL and its maximum power density. Diffusion layer

Methanol concentration (M)

CC

CC_T

CC_MPL

CC_MPL_E

1 2 3 1 2 3 1 2 3 1 2 3

M M M M M M M M M M M M

Rohmic (U)

RA (U)

RC (U)

RCrossover (U)

Maximum Power density (mW/cm2)

0.47 0.46 0.42 0.45 0.68 0.40 0.55 0.56 0.60 0.55 0.63 0.76

0.030 0.020 0.030 0.010 0.020 0.020 0.010 0.010 0.020 0.010 0.015 0.010

0.39 0.30 0.28 0.34 0.29 0.18 0.44 0.27 0.19 0.45 0.29 0.19

0.63 0.47 0.30 0.59 0.55 0.48 0.87 0.62 0.56 1.03 0.57 0.52

1.34 1.43 1.66 1.26 1.19 1.04 1.06 0.95 1.30 0.96 1.03 1.12

tested, better performances are achieved using carbon cloth generating the lower values of RCrossover representing by far the largest resistance for all the conditions tested. The best performances were obtained with the CC for all the methanol concentrations tested and with the CC-T type carbon cloth for the lower concentration. As found in previous work [12,13], thicker cathodes usually conduct to better results due to a low methanol crossover generated with the higher MEA thickness. Accordingly, in this work care was taken to select carbon cloths with a relatively high thickness as displayed in Table 1. The main question was to check if when using a relatively thick cathode DL, it is worthwhile the use of a MPL which usually contributes to an increase in the MEA cost. As can be concluded from the values displayed at Table 3, the cell configurations using a CC with a MPL (CC_MPL and CC_MPL_E) do not generate, in most of the conditions tested, higher performances, when compared to the CC cathodes without this extra-layer and operating at the same methanol concentration level. It would be expected that using a MPL at the cathode DL would be beneficial for the fuel cell performance due to an enhanced water management and a more uniform reactant distribution at the electrode surface. However, the use of a MPL has a negative impact on the cathode porosity (0.80 for the CC_MPL and 0.63 for the CC_MPL _T. The decrease of the porosity of the DL, negatively affects the oxygen and water diffusion rates to and from the catalyst layer, due to a blocking of the diffusion layer pores [21]. This leads to a lower oxygen concentration at the reaction zone and more active sites available for the methanol oxidation with the available water for this reaction. A relatively high methanol gradient is maintained, contributing for a high level of methanol crossover which is responsible for the reasonably high levels of fuel crossover. Fig. 6a)ec) put in evidence the impact of this relatively lower oxygen concentration in a lower limiting current obtained in particular for the two lower methanol concentrations, when using CC with MPL. The impact of the methanol concentration on the cell performance will be properly analysed ahead in section Effect of methanol concentration.

papers PTFE treated (porosity of 0.78) with different thicknesses, CP and CP_T (0.110 mm and 0.190 mm) and two carbon papers with MPL, CP_MPL and CP_MPL_T, which have a slightly higher porosity (0.80), and thickness of 0.240 mm and

Carbon paper as cathode DL A similar study was performed using carbon paper as cathode DL, with four different materials considered: two carbon

Fig. 7 e Effect of using different carbon papers as cathode DL on the cell performance for different methanol concentrations: (a) 1 M, (b) 2 M and (c) 3 M.

19340

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

0.340 mm, respectively. The polarization curves are depicted in Fig. 7 and the resistance values and maximum power output displayed in Table 4. Likewise, in the study concerning the use of carbon cloth, three different methanol concentrations (1 M, 2 M and 3 M) were tested and a carbon cloth with a microporous layer (CC_MPL) was used as the anode DL. As can be seen in Fig. 7 and Table 4, for all the concentration values tested, better performances were achieved using CP_MPL as cathode DL. These results are expected for this relatively thick CP with a microporous layer with a slightly higher porosity than the two thinner CP materials (Table 1). Carbon paper is less porous than carbon cloth. The introduction of the MPL layer tends to generate an increase in its average porosity which is favorable for the oxygen diffusion and water removal. Therefore, more oxygen reaches the catalyst layer which represents less active sites available on the platinum catalyst for the methanol oxidation. In fact, at the cathode catalyst both methanol and oxygen compete for available sites for the oxidation and reduction reactions, respectively. A relatively low methanol gradient is maintained which tends to generate lower levels of methanol crossover. The enhanced water removal promoted by the MPL layer, helps to alleviate the cathode from the excess water formed avoiding cathode flooding. Despite its slightly lower porosity, the CP_T, surprisingly, showed similar power outputs when compared to those obtained with the CP with a MPL. The two PTFE treated carbon papers CP and CP_T differ mainly in their thickness (0.110 and 0.190 cm, respectively). The presence of the PTFE helps for both materials in the water removal. The greater thickness of CP_T ensures a higher water concentration at the cathode catalyst layer which contributes, as expected, to lower levels of methanol crossover when comparing the performance of these two materials.

Carbon cloth vs carbon paper In the previous sub-sections, two carbon based materials, carbon paper and carbon cloth, with different characteristics where employed as cathode DL and their effects on the fuel cell power output were evaluated. The results showed that, in general, higher performances were obtained using carbon cloth as DL due to an enhanced oxygen diffusion rate, leading to lower cathode activation losses. Despite its higher

Table 5 e Price of the different materials tested as cathode DL (values obtained from QuinTech and Fuel Cells Etc) and total DL cost (25 cm2). Material

Price (V) Size (cm2) Price (V)/cm2 DL cost (V) (25 cm2)

CC CC_T CC_MPL CC_MPL_E CP CP_T CP_MPL CP_MPL_T

95 108 45 12 127 99 45 45

900 900 400 25 361 361 400 400

0.120 0.106 0.113 0.480 0.352 0.274 0.113 0.113

3.00 2.65 2.83 12.0 8.80 6.85 2.83 2.83

thickness, carbon cloth has, in most cases, a porosity higher than carbon paper enhancing the oxygen diffusion towards the catalyst layer, which will improve the oxygen reduction reaction rate and consequently the cathode performance. The difficulty in getting an optimized water and oxygen transport at the cathode side when carbon paper is used as DL leads to higher methanol crossover rates through the membrane and cathode overpotentials. This can be seen by the lower power outputs and higher activation resistances for the experiments where carbon paper was used as DL (Table 4). Concerning costs, as can be seen in Table 5, CP has a price per cm2 usually higher than the carbon cloth one [28,29], however the two best materials in the present study (CC and CP_MPL) have similar prices.

Effect of methanol concentration In the previous sub-sections, better performances were obtained when CC was used as cathode DL. Therefore, in this section this material was used as cathode DL to study the effect of the methanol concentration on the pDMFC power output towards a further increase of its performance. The corresponding polarization curves are shown in Fig. 8 and the different resistances, as well as, the maximum power density achieved for each concentration tested, 1 M, 2 M, 3 M, 5 M and 7 M, are depicted in Table 6. As can be seen in Fig. 8, the open circuit voltage is much lower that the ideal voltage due to methanol crossover towards the cathode side, and decreases with an increase of the

Table 4 e Values for the different resistances in the EEC for the different carbon papers tested as cathode DL and its maximum power density. Diffusion layer CP

CP_T

CP_MPL

CP_MPL_T

Methanol concentration (M) 1 2 3 1 2 3 1 2 3 1 2 3

M M M M M M M M M M M M

Rohmic (U)

RA (U)

RC (U)

RCrossover (U)

Maximum Power density (mW/cm2)

0.55 0.63 0.76 0.54 0.54 0.66 0.42 0.52 0.72 0.50 0.37 0.45

0.010 0.020 0.010 0.020 0.020 0.020 0.010 0.020 0.020 0.020 0.015 0.010

0.38 0.35 0.30 0.36 0.22 0.21 0.45 0.28 0.21 0.42 0.32 0.28

1.01 0.76 0.91 0.82 0.61 0.58 0.84 0.61 0.53 0.92 0.65 0.44

0.68 0.84 0.80 0.95 1.23 1.28 1.09 1.33 1.38 1.01 1.10 1.25

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

19341

cathode side, increasing the oxygen diffusion towards the catalyst sites, with a consequent increase on the oxygen reduction rate and a decrease of the cathode activation losses (RC). However, the results showed that when the methanol concentration is too high, 7 M, the beneficial effect of a higher methanol concentration at the anode and cathode reactions (RA and RC) is lower than the negative effect of an increase of the methanol crossover rate with an increase of the activation losses of the parasitic oxidation reaction (RCrossover). Therefore, in these conditions the cell loses performance (Fig. 8 and Table 6). Fig. 8 e Effect of methanol concentration on the performance of a passive DMFC using the best material as cathode DL: CC.

Table 6 e Values for the different resistances in the EEC for different methanol concentrations and its maximum power density. Methanol Rohmic RA RC RCrossover Maximum concentration (U) (U) (U) (U) Power density (M) (mW/cm2) 1 2 3 5 7

M M M M M

0.47 0.46 0.42 0.41 0.57

0.030 0.020 0.030 0.020 0.030

0.39 0.30 0.28 0.13 0.12

0.63 0.47 0.30 0.24 0.38

1.34 1.43 1.66 3.00 1.87

methanol concentration. This is explained by an increase of the concentration gradient between the anode and the cathode side with an increase of the methanol concentration, which lead to a higher crossover rate through the membrane [26]. Even though, as can be seen by the plots presented in Fig. 8 and the power outputs achieved for each concentration (Table 6), the best performance was obtained with a methanol concentration of 5 M. This is due to the fact that despite an increase on the methanol concentration generates an increase of the methanol crossover rate it also increases the methanol diffusion through the anode layers, due to a higher concentration gradient between them, increasing the amount of methanol that reaches the reaction zone and therefore increasing the methanol oxidation rate. A higher oxidation rate will lead to a lower activation loss on the anode side (RA) and a lower methanol crossover, since the availability of methanol at the catalyst layer is lower, and therefore its concentration gradient, which is the driven force for its crossover, is also lower. A decrease of the methanol crossover rate will, also, conduct to higher cathode performances and therefore lower activation losses for both cathode electrochemical reactions: oxygen reduction (RC) and methanol oxidation (RCrossover). Additionally, a higher methanol concentration at the anode side leads to a lower water concentration at this side, increasing the water back diffusion towards the anode [14,27]. This will avoid cathode flooding and will allow a more efficient water management on the

Conclusions Different carbon-fiber materials with different characteristics were tested as cathode diffusion layer on a passive DMFC, towards an optimization of the cathode performance and consequently the cell performance. The cell performance was evaluated through the polarization curves and the results were explained under the light of the EIS data by modeling it with an innovative electric equivalent circuit, consisting of four resistances and three constant phase elements. This procedure allowed the identification of the different losses that negatively affect the cell performance and quantify them through the estimation of its corresponding resistances: ohmic and anode and cathode activation resistances. The cathode resistance comprises the cathode activation losses due to oxygen reduction reaction and the methanol oxidation reaction, due to methanol crossover. These two contributions were quantified. The EEC employed in this study showed a good agreement with the EIS data, revealing that the EEC proposed reproduces with accuracy the system under study. Among the two different carbon based materials studied (carbon paper and carbon cloth), better results were achieved using carbon cloth as cathode DL mainly due to its higher porosity which conducted to higher oxygen and water diffusion rates and therefore lower cathode activation losses. Regarding the addition of a MPL to the cathode DL structure, it was verified that this is only advantageous when carbon paper is used as backing layer due to an enhancement of its porosity, which is beneficial for the oxygen diffusion and water management. For this specific cell design and conditions tested, the maximum power density, 3.00 mW/cm2, was achieved using carbon cloth without MPL and a lower thickness (CC) as cathode diffusion layer and a methanol concentration of 5 M. Despite this power output is not very attractive, comparing to those already presented in literature, it can be high enough for some specific applications. This value can be justified by the lower loadings used in the current work (3 mg/cm2 Pt/Ru and 1.3 mg/cm2 Pt), which allowed a reduction of 15% on the fuel cell costs, when compared to the common loadings used for this type of fuel cells (4 mg/ cm2 Pt/Ru and 4 mg/cm2 Pt). Additionally, the present work shows that changes in the fuel cell structure and configuration are effective ways to improve the fuel cell performance and costs and achieve the power outputs and the costs needed for real applications.

19342

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

Acknowledgements B. A. Braz acknowledges the Ph.D. fellowship (BEX 12997/13-7) ~ o de Aperfeic¸oamento de supported by CAPES, “Coordenac¸a Pessoal de Nı´vel Superior” - Brazil. V.B. Oliveira acknowledges the post-doctoral fellowship (SFRH/BDP/91993/2012) sup~ o para a Cie ^ ncia e Tecnoported by the Portuguese “Fundac¸a logia” (FCT), POPH/QREN and European Social Fund (ESF). This work was financially supported by: Project PTDC/ NewPortCell-POCI-01-0145-FEDER-032116 - funded by FEDER funds through COMPETE2020 - Programa Operacional Com~ o (POCI) and by national petitividade e Internacionalizac¸a funds (PIDDAC) through FCT/MCTES. POCI (FEDER) also supported this work via CEFT.

references

[1] Kamarudin SK, Achmad F, Daud WRW. Overview on the application of direct methanol fuel cell (DMFC) for portable electronic devices. Int J Hydrogen Energy 2009;34:6902e16. https://doi.org/10.1016/j.ijhydene.2009.06.013. [2] Zhao TS, Chen R, Yang WW, Xu C. Small direct methanol fuel cells with passive supply of reactants. J Power Sources 2009;191:185e202. https://doi.org/10.1016/j.jpowsour.2009. 02.033. [3] Munjewar SS, Thombre SB, Mallick RK. Approaches to overcome the barrier issues of passive direct methanol fuel cell - Review. Renew Sustain Energy Rev 2017;67:1087e104. https://doi.org/10.1016/j.rser.2016.09.002. [4] Ong BC, Kamarudin SK, Masdar MS, Hasran UA. Applications of graphene nano-sheets as anode diffusion layers in passive direct methanol fuel cells (DMFC). Int J Hydrogen Energy 2017;42:9252e61. https://doi.org/10.1016/j.ijhydene.2016.03. 094. [5] Oliveira VB, Pereira JP, Pinto AMFR. Effect of anode diffusion layer (GDL) on the performance of a passive direct methanol fuel cell (DMFC). Int J Hydrogen Energy 2016;41:19455e62. https://doi.org/10.1016/j.ijhydene.2016.06.116. [6] Wu QX, An L, Yan XH, Zhao TS. Effects of design parameters on the performance of passive direct methanol fuel cells fed with concentrated fuel. Electrochim Acta 2014;133:8e15. https://doi.org/10.1016/j.electacta.2014. 03.183. [7] Yuan W, Tang Y, Yang X, Liu B, Wan Z. Structural diversity and orientation dependence of a liquid-fed passive air-breathing direct methanol fuel cell. Int J Hydrogen Energy 2012;37:9298e313. https://doi.org/10.1016/j.ijhydene. 2012.03.01. [8] Yuan W, Tang Y, Wan Z, Pan M. Operational characteristics of a passive air-breathing direct methanol fuel cell under various structural conditions. Int J Hydrogen Energy 2011;36:2237e49. https://doi.org/10.1016/j.ijhydene.2010.11.067.

[9] Yuan W, Tang Y, Wang Q, Wan Z. Dominance evaluation of structural factors in a passive air-breathing direct methanol fuel cell based on orthogonal array analysis. Appl Energy 2011;88:1671e80. https://doi.org/10.1016/j. apenergy.2010.11.009. [10] Song KY, Lee HK, Kim HT. MEA design for low water crossover in air-breathing DMFC. Electrochim Acta 2007;53:637e43. https://doi.org/10.1016/j.electacta.2007.07.044. [11] Jewett G, Guo Z, Faghri A. Water and air management systems for a passive direct methanol fuel cell. J Power Sources 2007;168:434e46. https://doi.org/10.1016/ j.jpowsour.2007.03.052. [12] Xu C, Faghri A, Li X, Ward T. Methanol and water crossover in a passive liquid-feed direct methanol fuel cell. Int J Hydrogen Energy 2010;35:1769e77. https://doi.org/10.1016/ j.ijhydene.2009.12.055. ~ o DS, Silva RA, Rangel CM, Pinto AMFR. Performance of [13] Falca an active micro direct methanol fuel cell using reduced catalyst loading MEAs. Energies 2018;11(2):369. https://doi. org/10.3390/en11020369. ~ o DS, Rangel CM, Pinto AMFR. Water [14] Oliveira VB, Falca management in a passive direct methanol fuel cell. Int J Energy Res 2013;37:991e1001. https://doi.org/10.1002/er.2902. [15] Cao J, Chen M, Chen J, Wang S, Zou Z, Li Z, Akins DL, Yang H. Double microporous layer cathode for membrane electrode assembly of passive direct methanol fuel cells. Int J Hydrogen Energy 2010;35:4622e9. https://doi.org/10.1016/ j.ijhydene.2010.02.012. [16] Wu QX, Zhao TS, Yang WW. Effect of the cathode gas diffusion layer on the water transport behavior and the performance of passive direct methanol fuel cells operating with neat methanol. Int J Heat Mass Transf 2011;54:1132e43. https://doi.org/10.1016/j.ijheatmasstransfer.2010.11.009. [17] Chen M, Chen J, Li Y, Huang Q, Zhang H, Xue X, Zou Z, Yang H. Cathode catalyst layer with stepwise hydrophobicity distribution for a passive direct methanol fuel cell. Energy Fuels 2012;26:1178e84. https://doi.org/ 10.1021/ef201834e. [18] Zhang H, Liu Z, Gao S, Wang C. A new cathode structure for air-breathing DMFCs operated with pure methanol. Int J Hydrogen Energy 2014;39:13751e6. https://doi.org/10.1016/ j.ijhydene.2014.02.145. [19] Cao J, Wang L, Song L, Xu J, Wang H, Chen Z, Huang Q, Yang H. Novel cathodal diffusion layer with mesoporous carbon for the passive direct methanol fuel cell. Electrochim Acta 2014;118:163e8. https://doi.org/10.1016/j.electacta.2013.11.140. [20] Li Y, Zhang X, Nie L, Zhang Y, Liu X. Stainless steel fiber felt as cathode diffusion backing and current collector for a micro direct methanol fuel cell with low methanol crossover. J Power Sources 2014;245:520e8. https://doi.org/10.1016/ j.jpowsour.2013.06.122. [21] Abdelkareem MA, Kasem ET, Nakagawa N, Abdelghani EAM, Elzatahry AA, Khalil KA, Barakat NAM. Enhancement of the passive direct methanol fuel cells performance by modification of the cathode microporous layer using carbon nanofibers. Fuel Cells 2014;14:607e13. https://doi.org/ 10.1002/fuce.201300279. [22] Bresciani F, Rabissi C, Zago M, Marchesi R, Casalegno A. On the effect of gas diffusion layers hydrophobicity on direct methanol fuel cell performance and degradation. J Power Sources 2015;273:680e7. https://doi.org/10.1016/ j.jpowsour.2014.09.149. [23] Deng H, Zhang Y, Zheng X, Li Y, Zhang X, Liu X. A CNT (carbon nanotube) paper as cathode gas diffusion electrode for water management of passive m-DMFC (micro-direct methanol fuel cell) with highly concentrated methanol. Energy 2015;82:236e41. https://doi.org/10.1016/j.energy.2015. 01.034.

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 4 4 ( 2 0 1 9 ) 1 9 3 3 4 e1 9 3 4 3

[24] Wang G, Lei L, Jiang J, Zhou Y, Huang Q, Zou Z, Jiang SP, Yang H. An ordered structured cathode based on vertically aligned Pt nanotubes for ultra-low Pt loading passive direct methanol fuel cells. Electrochim Acta 2017;252:541e8. https://doi.org/10.1016/j.electacta.2017.09.038. [25] Xue R, Zhang Y, Liu X. A novel cathode gas diffusion layer for water management of passive m-DMFC. Energy 2017;139:535e41. https://doi.org/10.1016/j.energy.2017.08.016. [26] Piela P, Fields R, Zelenay P. Electrochemical impedance spectroscopy for direct methanol fuel cell diagnostics. J

19343

Electrochem Soc 2006;153:A1902e13. https://doi.org/10.1149/ 1.2266623. ~ o DS, Rangel CM, Pinto AMFR. One [27] Oliveira VB, Falca dimensional and non-isothermal model for a passive DMFC. J Power Sources 2011;196:8973e82. https://doi.org/10.1016/ j.jpowsour.2011.01.094. [28] http://www.quintech.de/englisch/products/research/GDLUebersicht.php. accessed in July 2018. [29] http://www.fuelcellsetc.com/store/Layers. accessed in July 2018.