A New Palladium-Based Catalyst for Methanol Steam Reforming in a Miniature Fuel Cell Power Source

By Oleg Ilinich, Ye Liu, Christopher Castellano,

Gerald Koermer, Ahmad Moini and Robert Farrauto

A novel palladium-based catalyst has been developed for use in a miniature fuel cell power source for portable applications, incorporating a polymer electrolyte membrane (PEM) fuel cell. Hydrogen, which is the fuel for the cell, is produced in a ceramic microreactor via the catalytic reaction of methanol steam reforming: CH₃OH + H₂O → 3H₂ + CO₂. The need for a new catalyst in this application is driven by the limitations of traditional particulate catalysts based on copper oxide, zinc oxide and alumina (Cu-Zn-Al catalysts), which have low thermal stability and high sensitivity towards air and condensing steam. These features result in a declining activity and mechanical integrity of Cu-Zn-Al catalysts under the frequent start-stop conditions typical of the operational mode of the miniature power source. The new Pd-based catalyst has activity and selectivity similar to those of Cu-Zn-Al catalysts, but is more durable and stable under the duty cycle conditions of a portable power source. In the microreformer, the catalyst is washcoated directly on the walls of the steam reforming section, providing favorable conditions for efficient heat transfer between the heat-generating catalytic combustion section of the microreformer and its heat-consuming steam reforming section.

1. Introduction

Current trends in energy demand for portable electronics show that the power consumption in devices such as cell phones, personal digital assis-tants (PDAs), notebook computers and digital cameras continues to rise. Consumers demand small size, light weight, but long battery life. The batteries most commonly used in these devices are of the rechargeable (secondary) type. However, in certain applications such as those used by exploratory expeditions, first responders and the military, recharging a battery in the field is often difficult, if possible at all. Therefore a heavy load of disposable (primary) batteries must be transported to remote locations. One solution to this problem is to use a small portable battery charger powered by a fuel cell.

Due to the low energy density of compressed hydrogen, using it as a fuel for a portable fuel cell charger is not a viable option. Liquid fuel, in particular methanol, has a much higher energy density and is easier to transport and handle, which makes it practically attractive for this application. Methanol is used as a fuel in two different types of fuel cells. In the direct methanol fuel cell (DMFC) (Figure 1), methanol is fed directly to the anode where it reacts with water, generating electrons which travel through the external circuit as electric current, Reaction (i):

CH₃OH + H₂O → CO₂  + 6H⁺ + 6e^–   (i)

Protons travel through the proton-conducting polymer electrolyte membrane (for example, Nafion^®) to the cathode where they react with oxygen from the atmosphere, to produce water. The subject catalyst of this article was developed for use in a reformed methanol fuel cell, shown in Figure 2. In essence this is a classic type of fuel cell, invented in 1839 by William Grove. Here the fuel is hydrogen which is fed to the anode, where it splits electrocatalytically into protons and electrons, Reaction (ii):

H₂  → 2H⁺ + 2e^–                             (ii)

In contrast to Grove’s cell, however, in our application the hydrogen is produced by the catalystic steam reforming of methanol, Reaction (iii):

CH₃OH + H₂O → 3H₂ + CO₂                              (iii)

This reaction has received much attention in the last decades as an attractive route to hydrogen supply. An excellent review on methanol steam reforming (MSR) for hydrogen production has recently been published.

Fig. 1 Schematic of a direct methanol fuel cell

Fig. 2 Schematic of a reformed methanol fuel cell

The MSR reaction can be efficiently catalysed by copper-based catalysts, including the Cu-Zn-Al particulates otherwise used in methanol synthesis and the water-gas shift reaction. These catalysts are commercially produced, and have been used successfully in industry for many years. By their nature they are quite sensitive to the process conditions. In particular, they are prone to sintering at temperatures above about 280 to 300ºC, which results in a significant decline in activity, and also deteriorate both mechanically and in performance if steam condenses on them. Besides, a Cu-Zn-Al catalyst can develop dangerously strong exotherms if in its oxidised state it is exposed to a reducing environment, or, in its reduced (active) state, to an oxidising environment, such as ambient air. Therefore in industrial settings Cu-Zn-Al catalysts are operated under carefully controlled conditions. The operating cycle includes a lengthy start-up with slow reduction in syngas (H₂/CO) or hydrogen heavily diluted with nitrogen to minimise the reduction exotherm. The reduced and activated catalysts normally operate under steady-state conditions.

By contrast, the duty cycle anticipated for the MSR catalyst in the miniature fuel cell power source is much more demanding. The miniature power source will be operated with frequent starts and stops, during which liquid (reformate) will condense and may even freeze on the catalyst. Besides, slow reduction in dry gas with low concentrations of a reductant will clearly be unavailable, and the catalyst will have to be activated  upon direct contact with the methanol/water feed mixture. The properties of Cu-Zn-Al catalysts are incompatible with these requirements, and therefore a new application-specific catalyst had to be developed. This catalyst is a further improvement over the family of palladium-zinc-based catalysts which have been developed for fuel cell applications in recent years.

2. Experimental

The novel MSR catalyst has been developed using a combination of rapid catalytic screening and detailed parametric studies simulating the duty cycle of the miniature power source. The catalyst consists of Pd and Zn on an oxide support with proprietary additives. In the present paper this catalyst is designated as ‘Pd-Zn/oxide support’. It is fabricated in a powder form, then slurried and used as a washcoat on the channels of the micro-reformer (Figure 3) to be incorporated into a miniature fuel cell power source.

The catalytic performance of a MSR catalyst is determined by two characteristics: activity and selectivity. A good MSR catalyst should provide high rates of conversion of methanol and water into hydrogen and carbon dioxide, while side reactions should be minimised. In the MSR process, in addition to the target steam reforming Reaction (iii), several undesired side reactions can occur. In particular, carbon monoxide can be formed via methanol decomposition, Reaction (iv):

CH₃OH → CO + 2H₂                   (iv)

and/or reverse water-gas shift, Reaction (v):

CO₂  + H₂  → CO + H₂O          (v)

Fig. 3 (a) The prototype microreformer (with U.S. quarter dollar coin for scale);

(b) cross section of the microreformer with Pd-Zn/oxide support

Carbon  monoxide,  a  known  catalytic  poison, can only be tolerated by the fuel cell catalysts if its concentration in the reformate is low. Even for thefuel cells based on polybenzimidazole(PBI) membranes,  which  operate  at  elevated temperatures (~180ºC) and are more CO-tolerant than the fuel cells  with  Nafion®  membranes  which  operate  at~ 80ºC, the CO level in the reformate must not exceed 1–2%.

Another  possible  side  product  is  methane, which   could   be   generated via methanation. Reactions (vi) and (vii), that consumes considerable amounts  of  hydrogen,  while  generating  large amounts of heat:

CO₂  + 4H₂  → CH₄  + 2H₂O      (vi)

CO + 3H₂  → CH₄  + H₂O          (vii)

In the context of this article, the selectivity is understood as the ratio of the concentration of a given product at certain conversion of methanol to the sum of concentrations of all carbon-containing gas-phase products (in our catalytic tests no coke formation  was  observed),  for  example  Equation(viii):

               (viii)

where S_CO2 % is the selectivity towards CO₂, and [CO₂], [CO] and [CH₄] are the concentrations of the corresponding species.

3. Results

3.1 Performance Testing of the Cu-Zn-Al and ‘Pd-Zn/Oxide Support’ Catalysts in the Simulated Start-Stop Mode

The  anticipated  duty  cycle  of  the  miniature power source will include frequent starts and stops, with variable periods of steady-state operation. The catalysts to be used in the microreformer must be thoroughly tested under conditions simulating such operation, and a special laboratory test procedure has been developed for this purpose. The test consisted  of  the  following  elements  as  illustrated  in Figure 4: (a) initial heat-up of the reactor to the reaction temperature, followed by starting the flow of the methanol/water feed mixture and (b) steady-state operation for approximately sixty hours; (c)multiple start-stop cycles consisting of (d) cooling down to 40ºC with the feed flow stopped (simulated shutdown of the power source); (e) heating up to the reaction temperature; (f) restarting the flow of the methanol/water feed mixture with steady-state operation for approximately two hours.

Fig. 4 Experimental temperature profile in methanol steam reforming reactor during the catalyst performance test

In the catalytic performance test using the above protocol, the commercial Cu-Zn-Al catalyst, initially run under steady-state conditions, showed stable activity for a period of about two weeks. However, soon after the onset of the start-stop temperature cycling the activity began decreasing, which continued for another two weeks of operation, showing no signs of stabilisation. However the CO₂ selectivity remained very high (in excess of 99%) throughout the test.

For the catalyst ‘Pd-Zn/oxide support’, tested using the same protocol, the activity under the steady-state conditions was also stable and close to that of Cu-Zn-Al catalyst. At the beginning of the temperature cycling, a slight drop in the methanol conversion was registered but the activity remained stable thereafter. The CO₂ selectivity was also stable throughout the test at 98 ± 0.6%.

The temperature dependence of the initial CO₂ selectivity (i.e. S_CO2 at low conversions of methanol) was further investigated in the range 230 to 320ºC, and was found to decrease with the temperature from 98.2 to 94.2% (Figure 5). It was also observed that for the given reaction conditions CO₂ selectivity is fairly constant over a broad range of methanol conversions; however it decreases at high conversions (above ~ 95 to 97%).

3.2 Mechanical Strength of the Cu-Zn-Al and ‘Pd-Zn/Oxide Support’ Catalysts

It is known that exposure to a liquid can cause mechanical deterioration of particulate Cu-Zn-Al catalysts. This practically important aspect has not been sufficiently addressed in the open literature. Therefore in addition to catalytic per-formance we also analysed the Cu-Zn-Al catalyst for its mechanical strength before and after sixty temperature cycles, and found that the catalyst pellets lose about 80% of their initial strength as a result of the temperature cycling. The number of starts and stops of the future miniature power sources will certainly be much greater, and there-fore a more significant negative impact on the mechanical strength and finally on the integrity of the catalyst should be anticipated. This is yet another reason why particulate Cu-Zn-Al catalysts cannot be used in the miniature fuel reformer, and why a new and more robust catalyst had to be developed.

The ‘Pd-Zn/oxide support’ catalyst was deposited on the walls of the prototype microchannel reformer, with the hydraulic diameter of a channel measuring a few hundred microns (Figure 3). The catalyst was successfully tested in the microreformer, producing hydrogen-rich reformate via MSR. The same catalyst on different support structure is now used commercially in other applications.

3.3 ‘Pd-ZnO/Oxide support’ Compared to ‘Pd-ZnO/Al₂O₃’, Pd/Al₂O₃ and Cu/CeO₂ Catalysts

High MSR activity with high CO₂ selectivity is well documented for Pd-Zn catalyst, which were first discovered by Iwasa et al, and are now being extensively investigated for hydrogen generation in portable fuel cell power systems. The literature information pertaining to the MSR performance of the Pd-Zn catalystic system, and available to the authors, deals with catalysts composed of Pd supported on zinc oxide and also Pd-ZnO compositions supported on alumina.

Fig. 5 Initial CO₂ selectivity versus the methanol steam reforming reaction temperature for the catalyst ‘Pd-Zn/oxide support’. Test conditions: molar ratio of feed CH₃OH:H₂O=0.88; gas hourly space velocity (GHSV) = 230,000 h^–1 (powder catalyst); CH₃OH conversions within 11%

Driven by a continuing interest in inexpensive non-precious metal catalysts for MSR applications, new copper-based catalystic compositions are also being developed. To better understand the strengths and possible limitation of our Pd-based catalyst, it is important to compare performance of this catalyst with that of other Pd- and Cu-based catalysts. To that end, following a procedure similar to those described in reference, we prepared two Pd-ZnO catalysts on different supports: alumina and an oxide support material used in preparation of the new Pd-based catalyst, both catalysts having equivalent contents of Pd and ZnO. These catalysts are hereinafter referred to as ‘Pd-ZnO/Al2O3’and ‘Pd-ZnO/oxide support’, respectively. For comparison, alumina-supported palladium catalyst with the same amount of palladium as in Pd-ZnO samples was also prepared via the same procedure. In addition, a copper/ceria catalyst with 20wt% CuO was prepared by incipient wetness impregnation of ceria with aqueous copper nitrate solution followed by drying and calcining. All catalysts were compared in term of their activities and selectivities, now at elevated temperatures typical for certain advanced applications. The samples (3g of each) were tested in a laboratory flow reactor as granules 250 to 710um in size. This particle size ensures test conditions free of pore diffusion. The feed mixture consisted of a reagent grade methanol and deionised water in volume ratio MeOH:H₂O=1:1(molar rato 0.88:1). The feed flow rates(from 0.3cm³ min^-1 to 2.0cm³ min^-1) and the catalyst temperatures(375 to 475℃) were programmed and controlled throughout the test as shown in Figures 6 to 9. The automated gas chromatographic anslysis was conducted at twenty mintues intervals intervals by sampling the gas mixture exiting the reactor. Performance of all the catalysts described above is analysed below.

Methanol conversion and CO₂ selectivity for the catalyst ‘Pd-ZnO/oxide support’ are plotted in Figure 6 for temperatures 375℃, 425℃ and 475℃ and a range of flow rates. Ia can be seen that the catalyst has stable activity with CO₂ selectivity ranging form around 70% at full conversion of methanol to around 82% at lower conversions.

Under the same experimental conditions the sample 'Pd-ZnO/Al₂O₃' is less active, and has much lower CO₂ selectivity(Figure 7). This catalyst ages, partially lossing activity in the course by lower methanol conversion in the last segment of the run as compared with the first. The main product with this sample is CO; under the experimental conditions of the test, methane is also produced with selectivity ranging from 1 to 6%. This implies significantly less efficient production of hydrogen with ‘Pd-ZnO/Al₂O₃’, and also different reaction pathways for the two similar catalysts ‘Pd-ZnO/Al₂O₃’,  and ‘Pd-ZnO/oxide support’.

The ‘Pd/Al₂O₃’ sample is even less active than ‘Pd-ZnO/Al₂O₃’, is more prone to ageing and has very poor CO₂ selectivity(Figure 8). CO and methane are the dominant carbon-containing products with the Pd/Al₂O₃ catalyst under the catalyst composition unsuitable for hydrogen generation.

The Pd-free sample with Cu impregnated on ceria has high initial activity and relatively high CO₂ selectivity; however it ages rapidly with a significant loss in activity(Figure 9).

Fig. 6 methanol steam reforming performance test of the catalyst 

‘Pd-ZnO/oxide support’

Fig. 7 methanol steam reforming performance test of the ‘Pd-ZnO/Al₂O₃’ catalyst

Fig. 8 methanol steam reforming performance test of the ‘Pd/Al₂O₃’ catalyst