A Direct Methanol Alkaline Fuel Cell Based on Poly(phenylene) Anion Exchange Membranes

Rajeswari Janarthanan; Satyananda Kishore Pilli; James L. Horan; Daniel A. Gamarra; Michael R. Hibbs; Andrew M. Herring

doi:10.1149/2.1041409jes

Direct methanol fuel cells based on proton exchange membranes (PEMs) have received considerable attention, as compared to hydrogen fuel cells, they have potential system simplicity, high system energy density, improved volumetric fuel storage and fuel supply.^1,2 Direct methanol alkaline fuel cells (DMAFCs) which employ an anion exchange membrane (AEM) electrolyte are more attractive than devices based on PEMs as they should provide improved electrode kinetics, simple water management, lower methanol permeability, and use of non- platinum metal catalysts.^3–6 Solid-state AEM electrolytes are more attractive than the conventional liquid KOH electrolytes as the carbonates and bicarbonates formed during operation do not precipitate as the cation is the AEM, and although adding an additional ionic resistance to the electrode, do diffuse through the cationic ionomer.^7,8 In spite of these advantages, it is important to note that except for a few reports, high performance in DMAFCs are shown only in the presence of the aqueous basic electrolyte, even for commercially available AEMs.^9,10 In general, alkaline DMFC studies employ commercially available AEMs, such as Tokuyama A-006, A-010, Fumatech FAA-2, and Morgane ADP from Solvay.^9–16 Radiation grafted membranes have been demonstrated to be promising AEMs for DMAFCs with a KOH free methanol fuel.^17,18 Still a great amount of research is being focused on improving the performance of available commercial AEMs as well as on development of new AEMs to achieve stable, highly conducting, and mechanically robust membranes for fuel cells with high performance and long durability.

There are very few alkaline direct methanol fuel cell reports where neat methanol is used and aqueous methanol is still the fuel of choice, in fact high power densities have only been achieved for these systems when the aqueous methanol is mixed with KOH. With a poly(ethylene-co-tetrafluoroethylene) based AEM of 51 ± 2 μm thickness, Varcoe et al., achieved a peak power density of 2.8 mW cm⁻² and an OCV of 0.48V at 50°C in aqueous 2M methanol.¹⁷ Their study also demonstrated an increasing OCV and decreasing peak power density when the thickness of the membrane was increased. At 50°C, with a 153 ± 4 μm AAEM, an OCV of 0.7 V and a peak power density of 1.2 mW cm⁻² were obtained. In another report by Varcoe et al., 1.17 mW cm⁻² power density was yielded with a cross linked AEM at 50°C.¹⁹ The fuel cell performance can be improved in the presence of KOH in the fuel as the hydroxide ions can increase the conductivity of the electrode by flushing out the carbonate/bicarbonate formed by the reaction of the product CO₂ with the OH⁻ electrolyte.

It is very desirable to fabricate stable AEMs to achieve long-term stability in the hydroxide environment.²⁰ The most commonly reported cationic functional group used in AEMs, the benzyl trimethylammonium (BTMA) cation, has been shown to have adequate stability in a fuel cell operating at a relatively low temperature of 50°C.²¹ It has, however, been demonstrated that improved stability can be achieved when the quaternary ammonium group is attached to the polymer backbone via an alkylene spacer with at least four carbon atoms.^22–25 In this work, we aim to study the DMAFC performances of two AEMs developed at Sandia National Laboratories (SNL). Both membranes are based on a poly(phenylene) backbone, the first one developed, aminated tetramethyl poly(phenylene) (ATMPP), contains BTMA cations and the second one, trimethylamine hexamethylene poly(phenylene) (TMAC6PP), has hexamethylene spacers separating the quaternary ammonium groups from the poly(phenylene) backbone (Figure 1). One of us has described the preparation and properties of ATMPP and TMAC6PP membranes in previous reports.^26,27 Both AEMs are stable, showing little or no decrease in their conductivity and IEC values even after 14 days under highly alkaline conditions at elevated temperature.^26,27 Our previous work was focused on understanding the transport properties of ATMPP.²⁸ Low water uptake, water content, and high ion conductivities up to 86 mS/cm at 90°C for the ATMPP membrane demonstrated its potential for promising application in fuel cells.

**Figure 1.** Structure of the polymers used in membranes and as ionomers in electrodes.

In addition to the AEM, a cationic polymer is also required to act as an ionomer in the electrodes to facilitate transport and immobilize the catalyst. Often the ionomer used in the electrodes is the same polymer that comprises the membrane. This is not always the best solution however, since the transport requirements in these two components can be quite different. In a previous study, a fluorinated polymer with guanidinium cations attached, was found to be a very effective ionomer when paired with hydrocarbon ATMPP membranes.²⁹ For the present study, ATMPP has been dissolved and used as an ionomer with ATMPP membranes and a new ionomer, TMAC6PPC6 (Figure 1), has been prepared to be compatible with the TMAC6PP membranes. TMAC6PPC6 is a poly(phenylene alkylene) with quarternary ammonium cations attached by hexamethylene sidechains, and has a hexamethylene spacer built into the backbone to achieve a lower T_g. As with TMAC6PP, the IEC of TMAC6PPC6 can be varied by changing the number of sidechains per repeat unit. The synthesis of TMAC6PPC6 has been described elsewhere.³⁰ It is designed to be chemically compatible with TMAC6PP while allowing for greater transport of water and methanol within the electrodes. Because it is designed to transport methanol easily, TMAC6PPC6 is obviously not used as a membrane to avoid problems with methanol crossover. Like ATMPP, TMAC6PPC6 is soluble in neat methanol and this helps to incorporate the ionomer in the catalyst ink.

A summary of the properties of the membranes used in this study is shown in Table I.^26,30 The ionomers used in this study in membrane form, ATMPP and TMAC6PP have the same IEC value, 2.4 meq/g. The two ionomers used in this study, in solution form for electrode preparation, TMAC6PPC6i and TMAC6PPC6h, are made to achieve two different IECs, Table I. The lower IEC ionomer, TMAC6PPC6i (IEC: 2.13 meq/g) with a low water uptake 121% was synthesized for use in the cathode catalyst ink, whereas the high IEC ionomer, TMAC6PPC6h (IEC: 2.60 meq/g) with a relatively high water uptake, 190% was prepared for use in the anode catalyst ink. The chloride ion conductivities were measured in liquid water at 25°C and the values of all the four ionomers are very close. Hydroxide ion conductivity of ATMPP membranes in carbon dioxide free pure nitrogen as a function of temperature showed an increasing trend with a maximum of 86 mS cm⁻¹ at 90°C and 95% RH.²⁸ Under similar conditions, the bromide and carbonate forms of the membrane exhibited conductivities of 13 and 16.6 mS cm⁻¹ respectively. The hydroxide, bromide, and carbonate ion conductivities of TMAC6PP are 17, 29 and 34 mS cm⁻¹ respectively at the lower temperature of 80°C. Interestingly TMAC6PP exhibits a very high carbonate conductivity. In this paper we now report the fuel cell performance of the characterized AEMs for an alkaline direct methanol fuel cell.²⁸ The current work is focused on benchmarking the AEM direct methanol fuel cell performance of the ATMPP and TMAC6PP membranes with commercial platinum catalysts in both electrodes, where the only instability of the system should be the ionomer and/or carbon supports.

Table I. Properties of AEMs and Ionomers used in this study.

	Measured	Cl⁻	Water	Br⁻	OH⁻	CO₃²⁻
	IEC (meq/g)	Conductivity	Uptake	Conductivity	Conductivity	Conductivity
AEM/Ionomer	¹H NMR	(mS/cm)^a	(%)^b	(mS/cm)^c	(mS/cm)^c	(mS/cm)^c
ATMPP	2.43	18.0	117	13	86	17
TMAC6PP	2.42	17.4	121	29	17	34
TMAC6PPC6l	2.13	16.1	121	NA	NA	NA
TMAC6PPC6h	2.60	18.3	190	NA	NA	NA

^aConductivity of polymer film without catalyst- measured in liquid water at 25°C.^bHydroxide form.^cBr⁻, OH⁻ and CO₃²⁻ conductivities of ATMPP are measured at 90°C and 95% RH, Br⁻, OH⁻ and CO₃²⁻ conductivities of TMAC6PP are measured at 80°C and 95% RH ^a and ^b Data reproduced from Refs. 26 and 30.NA – not applicable, as material was not used as a membrane.

Experimental

Materials

ATMPP, TMAC6PP, and TMAC6PPC6, ionomers were provided by SNL, Albuquerque, NM. We used platinum catalysts from two commercial sources, 20% Pt/C (E-TEK), and 50% Pt on high surface area carbon, TEC10E50E (Tanaka). Gas diffusion layers, GDLs from two different sources, GDL LT 1400-W, E-tek carbon cloth and carbon fiber fabric (Carbonized woven fabric, Panex 30, Zoltek) are used.

Membrane electrode assembly

Table II gives the details of MEAs studied in this work. MEA-1 and 2 are based on the TMAC6PP membrane but differ in the anode GDLs between the hydrophilic material from Zoltek and the hydrophobic material from E-TEK, keeping all other components identical. MEA-3 and 4 are based on the ATMPP membrane with identical GDLs on both electrodes but they contain different catalysts. In general, catalyst inks were made by mixing 50% Pt/C (TEC10E50E, Tanaka Kikinzoku Kogyo) or 20% Pt/Vulcan XC72R (E-TEK), Millipore water, isopropanol and ATMPP or TMAC6PPC6 ionomer (3% solution in methanol) with a catalyst to ionomer ratio of 4:1. An ionomer solution was obtained by dissolving the membrane in methanol. Catalyst inks were painted on 5.48 cm² GDLs to obtain a loading of 1.4 mg of Pt cm⁻² with the E-TEK catalyst and 2.5 mg of Pt cm⁻² with the Tanaka catalyst. The electrodes were hot pressed on ATMPP or TMAC6PP membranes using a digital combo multi- purpose press, DC14, Geo Knight & Co. Inc at 66°C and 80 psig for 90s. Prior to any tests, MEAs were soaked in 1M NaOH for 2h to convert the membrane from the Br⁻ form to OH⁻ form. Following this, MEAs were subsequently soaked in water with frequent changes of water to ensure complete removal of NaOH as indicated by a neutral pH of the washings. For better understanding, the components of all four MEAs are listed in Table II.

Table II. Details of the components used in the fabrication of the MEAs.

MEA	Anode Catalyst	Anode GDL	Anode Ionomer	Cathode Catalyst	Cathode GDL	Cathode Ionomer	AEM
MEA-1	Pt/C, Tanaka	Zoltek	TMAC6PPC6h	Pt/C, Tanaka	E-TEK	TMAC6PPC6l	TMAC6PP
MEA-2	Pt/C, Tanaka	E-TEK	TMAC6PPC6h	Pt/C, Tanaka	E-TEK	TMAC6PPC6l	TMAC6PP
MEA-3	Pt/C, Tanaka	E-TEK	ATMPP	Pt/C, Tanaka	E-TEK	ATMPP	ATMPP
MEA-4	Pt/C, E-TEK	E-TEK	ATMPP	Pt/C, E-TEK	E-TEK	ATMPP	ATMPP

Fuel cell testing

Single cell hardware from fuel cell technologies, Inc with a working area of 5.48 cm² and single serpentine flow fields was used. The supplied anodized aluminum endplates were replaced with stainless steel end plates to withstand the alkaline conditions. The effluent from the fuel cell reaches the back pressure regulators, followed by a water trap which separates out any condensed liquid in the exit lines. An isocratic HPLC pump (Chromtech) was used to pump the methanol solution or methanol in KOH solution to the anode. Oxygen is fed to the cathode through a modular gas handling/gas metering system (Lynntech Industry, Inc.) through a humidity bottle maintained at 80°C. The purge gas N₂ and flow rate of O₂ was controlled using FC power software (Lynntech Industry, Inc.). The fuel and oxygen flow rates were set to 0.5 ml min⁻¹ and 0.2 L min⁻¹ respectively in this study.

A MSTAT4+ multi-potentiostat (Arbin Instruments) was used to conduct polarization experiments. The polarization curves were obtained by stepping down the potential from open circuit potential until the limiting current was achieved. In situ impedance spectra were measured potentiostatically at different potentials on the polarization curve using a Gamry Instruments potentiostat. The frequencies were varied between 100,000 and 0.1 Hz with the data points being 10 per decade. The ac voltage was 10 mV root mean squared.

NMR product analysis

Quantitative ¹³C-{¹H}-NMR spectroscopy was performed on fuel cell effluent samples using a JEOL ECA 500 MHz spectrometer. A 150 s relaxation delay with 150 scans was used and the results were referenced to tetramethylsilane (TMS). A known concentration of 1,4 Dioxane, used as an internal standard, Chromium (III) acetylacetonate, used as a relaxation agent for quantitative results, and deuterated dimethyl sulfoxide (DMSOd6), used as a lock compound, were added to the sample.

Results and Discussion

In order to benchmark the two AEMs, we made four MEAs (MEAs: 1-4), two with the TMAC6PP membrane and two with the ATMPP membrane and investigated their methanol fuel cell performances. The fuel cell test was repeated with three sets of the same MEA^1–4 to understand the reproducibility of the measurement. Details of the MEA components are given in Table II. The high IEC ionomer, TMAC6PPC6h and the low IEC ionomer, TMAC6PPC6l were used on the anode and cathode respectively. TMAC6PPC6h with a higher water uptake was used to increase the hydrophilicity of the anode thereby enhancing the diffusion of aqueous electrolyte and methanol; whereas the low water uptake TMAC6PPC6l was used to help reduce the flooding on the cathode.^14,31 The performance did not show significant change in magnitude when the methanol flow rate was varied from 0.5 and 1.2 ml min⁻¹. Usually, AEMs are operated at low temperatures, below 60°C to avoid any decomposition of the cationic ammonium group. For the materials studied here it has been demonstrated that the materials have adequate stability for our three day fuel cell testing protocol. The TMAC6PP did not degrade and the ATMPP polymer degraded slowly at 90°C in 4M KOH for 14 days showing 8.3% and 33% loss in conductivity after 1 and 14 days respectively.²⁶ However, the membranes remained flexible and tough throughout the test. Also, when operated at elevated temperatures, electrode kinetics can be improved with reduced thermodynamic voltage losses.⁷ For this reason all experiments in this study were performed at 80°C and 100% RH.

The polarization curves (I-V characteristics) and the power density performance in 1M methanol are shown in Figure 2. To overcome mass transport limitations and to enhance a facile transport of fuel and hydroxide ions, a hydrophilic gas diffusion layer can be used at the anode side. The hydrophilic nature of the gas diffusion layer would certainly improve the easy transport of hydroxide ions through the electrode making sufficient OH⁻ ions for the methanol oxidation. We used a hydrophilic Zoltek carbon cloth in MEA-1. As expected, OCV of MEA-1 improved by 20 mV, with a 1.1 fold increase in the current and power densities for the hydrophilic Zoltek anode GDL containing MEA-1 as compared to MEA-2 with a hydrophobic anode GDL (Figure 2 and Table III). The maximum current and peak power densities were 29.5 mA cm⁻² and 3.97 mW cm⁻² for MEA-1 and 26.7 mA cm⁻² and 3.6 mW cm⁻² for MEA-2. From the curves, it can also be realized that the ATMPP based MEA-3 & 4 exhibit OCVs of 0.5 and 0.46 V, which is lower than 0.56 and 0.54 V obtained for MEA-1 and 2 respectively. To see if this is an effect of the membrane, we examined MEA-2 and MEA-3 which are identical except for the polymer used (Table II). The results demonstrated a slightly higher OCV (40 mV) and better performance for the TMAC6PP containing MEA-2 than for the ATMPP MEA-3 (Table III). The peak power densities were 3.6 and 3.1 for MEA-2 and MEA-3 respectively indicating better performance for TMAC6PP over ATMPP. This result agrees with the observation from Hibbs where stable conductivity was observed for TMAC6PP with only 5% drop in conductivity over 1 week but a gradual decrease of about 33% was observed for ATMPP.²⁶

Table III. ADMFC performances in 1M Methanol at 80°C, 100% RH.

		Current density at	Current density at	Peak Power
MEA	OCV (V)	0.1 V (mA cm⁻²)	0.1 V (mA cm⁻² mgPt⁻¹)	density (mW cm⁻²)
MEA-1	0.56	29.5	11.8	3.97
MEA-2	0.54	26.7	10.68	3.6
MEA-3	0.5	26.0	10.4	3.1
MEA-4	0.46	7.2	5.14	0.85

**Figure 2.** DMAFC performance curves for all four MEAs in 1M CH₃OH at 80°C, 100%RH (♦: MEA-1;▲: MEA-2; ●: MEA-3;■: MEA-4, open symbols represent power density curves).

To examine the effect of catalyst, identical MEAs (MEA-3 and MEA-4) that differ only by the catalysts were tested. It can be observed that the OCV dropped to 0.46 V; giving rise to a decreased Pt normalized current density of 5.14 mA cm⁻² mg⁻¹for MEA-4. The Tanaka catalyst (MEA-3) achieved 10.4 mA cm⁻² mg⁻¹ which is two folds better than the E-TEK catalyst (MEA-4). In neat methanol with no hydroxide ions in the fuel, the performances appear to be unaffected by the anode gas diffusion layers. Our values are superior to the values (0.15 to 0.3 mW cm⁻²) obtained with a commercial Fumatech FAA-2 in different concentrations of methanol.¹² Our MEAs also perform better than a QPVA/SiO₂/GA membrane which obtained 0.18 or 0.27 mW cm⁻² in 2M methanol.³² The maximum power data (3.97 mW cm⁻²) for MEA-1 is still lower, however, than the performance (∼ 8.5 mW cm⁻²) observed for an ETFE based AEM by Varcoe and Slade at 80°C with 2 and 2.5 bar back pressures at the anode and cathode respectively.¹⁸ The best data, 3.97 mW cm⁻² and 29.5 mA cm⁻², with our current AEMs in a KOH free fuel appears better than the data obtained from a commercial Morgane, ADP-Solvay membrane which yielded a performance of 5.54 mW cm⁻² and 32.6 mA cm⁻² in 2M methanol that contained 1M NaOH.¹⁵ In 2M Methanol, with a platinum catalyst containing NCNT of 4 mg cm⁻², a peak power density of 5 mW cm⁻² and 0.217 V OCV was obtained. Our current data is certainly better for a KOH free fuel compared to this result obtained in the presence of 2M NaOH using a Nafion 117 membrane as a fuel cell separator.³³

In general, AEM fuel cells show improved performance in the presence of hydroxide ions in the fuel as methanol oxidation requires a stoichiometric amount of hydroxide ions (6). In addition, the presence of hydroxide ions certainly improves the conductivity and flushes out any carbonate/bicarbonate formed in the electrode from the reaction of the product carbon dioxide with the hydroxide. The aim of the paper is to baseline the performance of AEM fuel cells, and the addition of KOH to these systems is technologically important using commercially available membranes.^{9,11,13–16,31,34,35} We further studied the high performing MEA-1 & 2 for fuel cell performance studies in the presence of KOH. Figure 3 demonstrates the fuel cell performances of MEA-2 in the presence of an alkaline electrolyte as a function of methanol and KOH concentrations. We observed a significant difference in the performance when KOH is added to the fuel supply. Comparing the performance of 1M MeOH-1M KOH (Figure 3) to 1M MeOH (Figure 2) anode feed for MEA-2, the OCV increased from 0.54 V to 0.75 V with a four fold increase in the current density in the presence of 1M KOH. With a constant methanol concentration (1M) and different KOH concentrations (0.5M to 2M), the OCV remained unaltered, showing differences in the current and power densities. From 0.5 to 1 M KOH/1 M methanol, there was no difference in the performances up to 20 mA cm⁻², but significant improvement was evident thereafter in the medium to high current density regions. The peak power densities were 16.1 and 25.0 mW cm⁻² for 0.5 and 1 M KOH with 1M methanol respectively with the maximum current densities being 80 and 125 mA cm⁻² for 0.5 and 1M KOH respectively. This observation agrees with the fact that increased electrolyte concentration would increase ionic conductivity in the electrode. A further raise to 2M KOH in the fuel did not yield any better performance, but the values were close to 1M KOH in 1 M methanol. The peak power densities were 16.1, 25.0, and 24.2 mW cm⁻² for 0.5, 1, and 2M KOH respectively. The current densities were 80, 125, and 117 mA cm⁻² for 0.5, 1, and 2M KOH respectively with 1 M methanol. The effect of electrolyte concentration has been demonstrated to attain a peak performance at a specific OH⁻ concentration by different researchers between 3 M and 8 M.^36–38 The reason for low performance at a high OH⁻ concentration has been attributed to electrolyte viscosity, electrolyte resistance, and/or pore blockage due to K₂CO₃ precipitation. In the current study, this limitation due to OH⁻ concentration is observed at with 1 M methanol at 2 M KOH.³⁶ With 1M MeOH-1M KOH as the baseline, increasing the fuel concentration to 2M methanol resulted in lower performance. Loss in performance and reduced OCVs (0.68 and 0.62 for the 2M-1M and 2M-2M respectively) could be related to the relatively higher methanol crossover at higher concentrations of methanol.

**Figure 3.** DMAFC performances of MEA-2 as a function of the concentrations of methanol and KOH (Δ: 1M-1M; ▲: 1M-0.5M; ▲: 2M-1M; ■:2M-2M; ●:1M-2M).

In their work, Scott et al. reported an increase in performance when the methanol concentration was increased from 1 to 2 M but again a drop in the performance when raised further up to 4 M using an alkaline ADP membrane.³⁹ In similar work, Prakash et al. observed a decrease in performance when the methanol concentration was increased to 2 M using an alkaline Tokuyama membrane.¹⁴ When compared to the control 1M MeOH-1M KOH, the current and power densities of 2M MeOH-1M KOH were 2 times reduced while the 2M MeOH-2M KOH showed 1.8 times less current and power density. The OCV, maximum current density and peak power density for 2M-1M were 0.68 V, 60.0 mA cm⁻² and 12.0 mW cm⁻², while for 2M-2M these values were 0.62 V, 68.1 mA cm⁻² and 13.6 mW cm⁻². Based on the above observation, the study on the effect of fuel and electrolyte concentration demonstrated the highest peak power and maximum current densities at 1M MeOH-1M KOH with 25 mW cm⁻² and 125 mA cm⁻² respectively (Table IV).

Table IV. ADMFC results as a function of concentrations of methanol and KOH at 80°C, 100% RH.

	MEA-1			MEA-2
		Current density	Peak Power		Current Density	Peak Power
MeOH – KOH	OCV (V)	(mAcm⁻²)	Density (mWcm⁻²)	OCV (V)	(mAcm⁻²)	Density (mWcm⁻²)
1M – 0.5 M	0.81	190.3	45.5	0.75	80.8	16.1
1M – 1M	0.84	226	53.8	0.75	125.0	25.0
1M – 2M	0.76	212.9	50.4	0.75	117.8	24.2
2M – 1M	0.84	166.4	43.4	0.68	60.0	12.0
2M – 2M	0.74	100.3	20.0	0.62	68.1	13.6

We also investigated the effect of fuel and electrolyte concentrations on the performance for MEA-1 with a hydrophilic anode GDL (Figure 4). While there was only slight improvement in the use of aqueous methanol as the fuel for MEA-1 over MEA-2, in the presence of KOH, considerably higher performance was observed for MEA-1 compared to MEA-2 with a hydrophobic GDL (Table IV). While there is no significant increase in OCV when the KOH concentration was raised from 0.5 M to 1M, relatively better current and power densities were observed. The maximum current and peak power densities obtained in 1 M KOH are 226 mA cm⁻² and 53.8 mW cm⁻², compared to 190.3 mA cm⁻² and 45.5 mA cm⁻² for a 0.5 M KOH. A further rise to 2M resulted in a performance decline, with an 80 mV loss in OCV, a maximum current density of 212.9 mAcm⁻² and a peak power density of 50.4 mWcm⁻² were achieved. As observed in MEA-2, increasing the methanol concentration to 2M resulted in poor performances with the lowest being 2M MeOH in 2M KOH. The OCVs, current densities and power densities were 0.84V, 166.4 mA cm⁻² and 43.4 mW cm⁻² for 2M MeOH and 1M KOH and 0.74V, 100.3 mA cm⁻² and 20.0 mW cm⁻² for 2M MeOH and 2M KOH. In the presence of KOH, the hydrophilic carbon cloth GDL (Zoltek) at the anode generated a significant improvement in the performance. The best performances of MEA-1 and MEA-2 were observed in 1M MeOH-1M KOH with a two-fold increase in power density for MEA-1.

In the literature, it has been demonstrated that a hydrophobic cathode and hydrophilic anode enhance the fuel cell performance. It was shown by Kim et al., that hydrophobicity decreases the methanol exposure to the catalyst.³¹ Similarly, in their study Suryaprakash et al., have shown that a hydrophilic anode GDL would enhance the diffusion of aqueous electrolyte and methanol and a hydrophobic cathode GDL would prevent flooding at the cathode.¹⁴ Our results agree with this observation, MEA-1 with the hydrophilic Zoltek GDL at the anode and the hydrophobic E-TEK GDL at the cathode performs better than MEA-2 with both electrodes in contact with hydrophobic GDLs. In the presence of hydroxide ions in the methanol, DMAFC performances of the membranes, ATMPP and TMAC6PP are superior when compared to some commercial membranes as well as on other AEMs.^{9,11,13,16,32,40} As expected from the previous studies, the current AEMs, ATMPP and TMAC6PP are certainly promising candidates for alkaline fuel cell application.

Operation of fuel cells using an hydroxide electrolyte in the fuel feed at elevated temperatures is expected to be challenging due to the caustic nature of the hydroxyl anion. Figure 5 presents the durability of MEA-2 in 1M MeOH-1M KOH at an applied current density of 35 mA/cm² at 80°C. It was necessary to pause the experiment at 36 h to refill the water in the oxygen humidifier bottle after which the experiment was restarted. Interestingly, the break in the experiment could be seen in the data as a voltage rise of 30 mV back to 0.4 V. The voltage lost after 67 h corresponded to an overall degradation rate of 400 μV h⁻¹. Considering the harsh chemical conditions of the test, this is still encouraging as the voltage did not degrade rapidly to zero and further refinement of operating conditions is expected to result in further improvements in stability. In comparison, Varcoe et al., demonstrated a degradation rate of 95 ± 10 μV h⁻¹, i.e. ca. ¼ the degradation rate of our cell for a 233h ADMFC test with a cross-linked AEM in a KOH free fuel at 50°C.²¹

**Figure 5.** Durability data for MEA-2 at 35 mA/cm² in 1M Methanol and 1M KOH at 80°C, 100%RH over 70h (the arrow marks the stop and restart of the experiment).

Nyquist plots of the fuel cell as a function of potential were produced from electrochemical impedance spectroscopy (EIS) performed in situ under fuel cell operation (Figure 6). Impedance spectra studied on MEA-1 in a KOH free methanol fuel at 80°C as a function of d.c. potentials at OCV, ca. 0.3 V, and 0.2 V is shown in Figure 6. The data did not appear to be Kramers-Kronig consistent,⁴¹ presumably because even though the voltage degradation rate appears manageable, the baseline of the EIS measurement was changing over the time scale of the experiment, 2-3 hours. We were able to measure the internal cell resistance and charge transfer resistances from the EIS.^12,37 The internal cell resistance that includes ohmic resistances from the membrane, electrode, and their interfaces are measured from the high frequency x-axis intercepts.⁴² The charge transfer resistances (R_ct) have been calculated from the intercept of the arc and the real impedance axis. At all potentials, one large arc was seen, that was shallower than a perfect semicircle indicating that the electrodes are micro porous which would approximate a constant phase element in an equivalent circuit. The potential dependent arc tends to reach the x - axis as the cell potential is decreased down from open circuit voltage (OCV) to 0.2 V. Except the cell potential, all other components and conditions are the same for all the plots. Hence, this can be attributed to the charge transfer resistance on the electrodes as a function of the cell potential. At open circuit voltage, when no current is drawn, as anticipated the charge transfer resistance was at its maximum, 3.0 Ω cm⁻². When the cell potential was stepped down, due to increasing current density and decreasing overpotential, the charge transfer resistances decreased down from 3.0 Ω cm⁻² at OCV to 0.8 Ω cm⁻² at 0.2 V. The second aspect examined from all the spectra is the cell resistance at the high frequency intercept; this is an indication of the ohmic internal resistance of the fuel cell. The ohmic internal resistance is a sum of resistances from various fuel cell components. It was observed to be almost the same at all potentials (Table V). This is to be expected owing to identical components in the fuel cell at all potentials.

Table V. Resistance values from Impedance spectra for MEA-1 in a KOH free methanol fuel at 80°C.

Cell Potential	High Frequency	R_ct
(V)	Intercept (mΩ cm⁻²)	(Ω cm⁻²)
OCV	28.6	3.0
0.4	26.6	1.7
0.3	26.1	1.0
0.2	26.4	0.8

**Figure 6.** Impedance spectra of MEA-1 in 1M Methanol at 80° C, 100% RH as a function of potential (.... at OCV; − ● ● − at 400 mV; − − − at 300 mV; —at 200 mV).

¹³C{¹H}-NMR liquid state spectroscopy on the fuel cell effluent (Figure 7) with an anode solution containing 1M Methanol and 1M KOH shows that CO₃²⁻ is the only product of the methanol oxidation detected in the fuel cell effluent under these conditions. The overall fuel cell reaction (Equation 3) in an alkaline direct methanol fuel cell can be determined by the anode (Equations 1) and cathode (Equation 2) half reactions 3.

**Figure 7.** ¹³C{¹H} NMR spectra of the anode solution of a DMFC after running a galvanostatic experiment at a voltage of 400 mV using 1M MeOH-1M KOH.

The carbonate we observed in the product is a result of reaction in [Equation 4], where CO₂ reacted with OH⁻ resulting in carbonate formation.

A galvanostatic fuel cell test using MEA-2 with 100 mL of 1M MeOH-1M KOH analyte solution was run at a potential of 400 mV until all of the reactant solution had passed through the fuel cell in ca. 1h.⁴³ Quantitative NMR analysis on the effluent using 1,4 dioxane as an internal reference, was used to close the mass balance to 95%, indicating that a total of 13.5% of the methanol was converted to carbonate. The gaseous CO₂ from the effluent is probably not fully accounted for in this NMR study. In the liquid effluent, NMR has detected only 13.5% of CO₂ which was converted to CO₃²⁻. Further diagnostics are required to account for the gaseous products. Future work will include a detailed study on methanol oxidation products from this DMAFC work using online GC and more optimized MEAs with non-precious metal catalysts.

Conclusions

The current work was focused on benchmarking the DMFC performances of two AEMs and a new ionomer developed at Sandia National Laboratories using commercial platinum catalysts. Preliminary studies using these newly developed AEMs suggest that they are potential candidates for ADMFCs. Normalized current density values demonstrated good performance for the poly(phenylene) AEM, with a cation consisting of six methylene spacers attached to a trimethyl ammonium group, with a maximum current density of 11.8 mA cm⁻² mg⁻¹ in a KOH free fuel. However, the use of KOH free fuel and these AEMs in ADMFCs was not sufficient to achieve practical levels of performance. Additions of KOH lead to significant improvement in the performance. Study of the performance of the fuel cell as a function of GDL hydrophobicity was investigated with two different GDLs. The results demonstrated that a hydrophilic anode GDL and hydrophobic cathode GDL give the best performance. We observed that the MEA with a hydrophilic Zoltek gas diffusion layer on the anode showed the best performance (226 mA cm⁻² and 53.8 mW cm⁻²) in the presence of KOH, suggesting that a hydrophilic gas diffusion layer plays a significant role in improving the fuel cell performance. While the fuel cell still showed a measurable rate of degradation under these conditions, an overall degradation rate of 400 μV h⁻¹ during a durability test for a typical MEA in 1M methanol and 1M KOH was observed over 67 h. These results are encouraging for further developments in this technology using these AEM and ionomer chemistries.

Acknowledgments

This work was supported by the Laboratory Directed Research and Development (LDRD) program at Sandia National Laboratories. Sandia National Laboratory is a multi-program laboratory operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Company, for the U. S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.

A Direct Methanol Alkaline Fuel Cell Based on Poly(phenylene) Anion Exchange Membranes

Article metrics

Author affiliations

Author notes

Dates

Abstract