Catalysis for Low Temperature Fuel Cells PART I: THE CATHODE CHALLENGES

By T. R. Ralph and M. P. Hogarth Johnson Matthey Technology Centre, Blounts Court, Sonning Common, Reading RG4 9NH. U.K.

Much of the pegotmance still to be gained in proton exchange membranefuel cells (PEMFCs) in use today is available from improvements to the cathode, traditionally made from unsupported or carbon-supported platinum. The search for improved cathode electrocatalysts has resulted in the development ofplatinum alloys which iftailored to the desired stack operating conditions can double the activiry for oxygen reduction. Recently advances have been made in cathode design which have raised perjormance levels in PEMFCs. The new electrocatalysts and cathode designs have increased electrical eficienc.y and power densities to the PEMFC stack, needed for commercial use. Improvements have also been achieved at the anode, by developments in platinum-ruthenium anodes for carbon monoxide and cell reversal tolerance. In this first paper; new cathode materials and designs are discussed; a second paper to be published in the April issue will look at anode advances.

The fuel cell of choice for a wide range of appli- cations spanning portable, stationary and transportation markets is the proton exchange membrane fuel cell (PEMFC) (1). This is princi- pally because of the high power density and the relatively low temperature of operation. Today the PEMFC typically operates at close to 80°C although there is a desire to move to &her tem- peratures close to 150°C to mitigate the effects of carbon monoxide (CO) poisoning at the anode. The basic unit cell in the PEMFC stack is shown in Figure 1. The membrane electrode assembly @A) is the key component where hydrogen and air react

electrochemically to generate elecmcal power. The MEA is typically located between a pair of flow field plates to give a single cell. The flow field plates are designed to distribute the reactant gases across the face of the MEA and also to collect the electrical current from the MEA. Sufficient unit cells are connected electrically to generate the desired power output (1). Depending on the appli- cation, a PEMFC system may contain from tens to a few thousand MEAs to produce from a few watts to several hundred kilowatts of power.

The MEA is a five layer structure containing at the centre the proton exchange membrane

Fig. I The bnsic unit cell of a proton exchange membrane fuel cell is assembled in the required numbers tv deliver the necessay power output. The component lavers of the membrane electrode assemblv (two catalvsed substrates nnd the PEM electrolyte) are laminated together and located between the flow field plates to form the unit cell

Phtinrm Metah Rm, 2002,46, (l), S 1 4 3

electrolyte which separates the electrode structures to prevent reactant gas mixing and the formation of an electrical short. Each electrode consists of a gas diffusion substrate with the platinum-based (Pt) electrocatalyst layers located between the membrane and the substrate. The electrocatalyst can be deposited either on the gas diffusion sub- strate or on the proton conducting membrane electrolyte using techniques such as screen print- ing, flexographic printing, gravure printing, spraying or rolling and calendering.

Electrocatalyst layers are typically from 5 to 20 p thick with the complete MEA being around 400 to 500 p thick. The MEA layers are normal- ly bonded together by hot pressing catalysed substrates to the membrane or, in the case of catal- ysed membranes, by compressing the gas diffusion substrate to the membrane during stack assembly.

Electrocatalyst Layers In the MEA the electrocatalyst layers employ

platinum group metal electrocatalysts to generate the electrochemical power by the reduction of oxy- gen at the cathode and the oxidation of hydrogen at the anode. Pt-based electrocatalysts are required to provide stability in the corrosive environment of the PEMFC. These are also the most active electrocatalysts for oxygen reduction and are among the most active for hydrogen oxidation. In pre-commercial PEMFC systems, carbon-support- ed Pt is employed at the cathode and carbon- supported platinum-ruthenium at the anode.

Carbon Black Supports

To achieve economical Pt loadings in the MEA (< 1 mg Pt cm-2) the electrocatalysts are supported on high surface area carbon blacks with a high mesoporous area (> 75 m2 g-' C) and a degree of graphitic character. Common supports are avail- able from Cabot Corporation (Vulcan XC72R, Black Pearls BP 2000), Ketjen Black International, Chevron (Shawinigan) and Denka. The support material must provide a high electrical conductivi- ty, give good reactant gas access to the electro- catalyst, have adequate water handling capability, particularly at the cathode where water is generat- ed, and also show good corrosion resistance,

Table I Relationship between the Platinum Loading of a Ketjen Carbon Black Supported Electrocatalyst and the Catalyst Metal Area

wt.% Pt on carbon

40 50 60 70

XRD Pt crystallite

size, nm

2.2 2.5 3.2 4.5

CO chemisorption metal area, m2 g-' Pt

120 105 88 62

especially under the highly oxidising conditions which occur at the cathode.

An important design criterion is that the elec- trocatalyst layers should be reasonably thin. This minimises the cell potential losses due to the rate of proton diffusion and reactant gas permeability in the depth of the electrocatalyst layer, which can become limiting as the current density increases during PEMFC operation (2). In contrast to auto- catalysts and process catalysts which have metal loadings of less than 5 wt.%, PEMFC electrocata- lysts typically have very high metal loadings of 2 40 wt.%, to minimise the electrocatalyst layer thick- ness. This challenges the catalyst manufacturer. High metal loading electrocatalysts must be pro- duced in volume with high metal dispersions.

This can be achieved by precipitating the Pt group metals using chemical reduction and an aqueous slurry of the carbon black support (3). The preparation provides the ease of manufacture, high yield and reproducibility necessary for the production of electrocatalyst at batch sizes from 1 to 25 kg and above. For example, Table I shows typical Pt dispersion properties for 40 to 70 wt.% Pt on Ketjen carbon black supported electrocata- lysts prepared at Johnson Matthey. As expected, as the metal loading increases on the carbon support, the Pt crystallite size increases. This is directly reflected by a reduction in the metal area available for gas phase CO chemisorption. Even at 70 wt.% Pt, however, the Pt crystallite size is still less than 5 nm and the metal area is still above 60 m2 g-' Pt. This is significantly better than the Pt crystallite

Pkzfinzm Metah Rm., 2002,46, (1) 4

size of 5.5 to 6 nm and CO metal area of 20 to 25 m2 g-' Pt for an unsupported Pt black electrocata- lyst that was used by Ballard Power Systems in the early years of MEA development (4).

Figure 2 shows the improved Pt dispersion on the microstructure of a 50 wt.% Pt supported on Ketjen electrocatalyst. The individual Pt crystal- lites (2-3 nm) are visible on the primary particles of the carbon black that fuse together to form the carbon aggregates which are typical of fuel cell car- bon supports. It is the mesoporosity within these aggregates that gives the carbon blacks their gas diffusion and water handling capability.

Polymer Solutions In addition to the electrocatalyst, a second key

component in the electrocatalyst layer is the pro- ton conducting polymer electrolyte. If the electrocatalyst is to be utilised in the PEMFC elec- trode reactions, there must be contact between the Pt-based electrocatalyst and the protons present in the membrane electrolyte. It has been shown (5) that the utilisation is low unless the electrocatalyst layer contains a soluble form of the proton con- ductmg membrane electrolyte.

A number of groups employ organic perfluoro- sulfonic acid polymer solutions (6) but at Johnson Matthey the corresponding aqueous solutions are usually employed (7). This reduces the risk of sin- tering the electrocatalyst during the preparation of the electrocatalyst layer - due to the possibility of reaction between the Pt-based electrocatalyst and the organic solvents in the polymer solution. There is also less concern that ttace levels of organic material will be left in the electrocatalyst layer after MEA manufacture. It is difficult to completely remove organic materials from the car- bon black supports used in the PEMFC due to the good adsorption properties of their highly porous structure. Any ttace contaminant might react with the electrocatalyst and poison the Pt or change the water handling capability of the cathode, which would eventually produce a water-flooded MEA. Both effects significantly reduce the h4EA perfor- mance, although the drop in performance may not be seen until the PEMFC has been operating for several thousands of hours.

Fig. 2 Trunsmission rlecrron microgruph showing flir P t ctytollire distriburiori for n 50 wr.8 Pr supported on Krfjen cnrlmi black fuel cell curulyst

Aqueous polymer solutions are also much more friendly in terms of the volume manufacture of MEAS, as using them avoids the risks and costs associated with solvent h a n h through an MEA plant. Aqueous technology does, however, present a challenge in terms of producing a suitable ink from the electrocatalyst and the aqueous polymer solution for preparing the electrocatalyst layer. Johnson Matthey have devoted much effort to ensure that the electrocatalyst is completely wetted by the polymer solution and that an ink with a sub- micron particle size distribution and the correct rheology is produced.

Cathode Electrocatalysts A typical performance from an MEA is shown

schematically in Figure 3. This illustrates the dra- matic impact that the cathode has on MEA performance when operating the anode on pure hydrogen. At all practical current densities, tens of mV are lost at the anode. At the cathode, howev- er, even on pure Pt the most active oxygen reduction material, in excess of 300 mV are lost from the thermodynamic potential for oxygen reduction at low current densities, due to the com- parably sluggish electrochemical kinetics. This is reflected in exchange current densities of around

PIatinwm Metuls Rev., 2002, 46, (1) 5

CURRENT DENSITY. rnA ern-'

lo-'' to lo-'' A cm-' for oxygen reduction on Pt at ambient temperature.

The mechanism of oxygen reduction is extrem- ely complex and a number of factors contribute to a lowering of the catalytic activity on Pt. The requirement to break a strong 0-0 bond early in the ditect 4-electron reduction (Reaction (i)), which is the desired pathway to maximise the elec- trical efficiency of the PEMFC, is notable. In addition the open circuit voltage (OCV) is lowered from the thermodynamic potential for oxygen reduction on Pt due to the production of some peroxide (Reaction (2)) and the formation of a range of possible platinum oxides (Reaction (iii)) at hlgh cell potentials.

0 2 + 4H' + 4- = 2H20 E"25-c = +1.23 V (VS. NHE) (i)

0 2 + W' + 2e = H202 E02y~ = +Oh8 V (VS. NHE) (ii)

Pt + H20 = Pt-0 + 2H' + 2e- Eo2yc = +0.88 V (VS. NHE) (i)

Platinum Alloys The development of a more active oxygen

reduction electrocatalyst than Pt has been the sub- ject of extensive research for a number of decades (8). In the 1980s UTC developed Pt-based metal alloy electrocatalysts supported on carbon black using carbothermal reduction. In the phosphoric

Fig. 3 Schematic showing the typical ceN potential versus current density out- put from an MEA operating on pure hydrogen. The major factors that control the MEA performance in the various regions of the cell potential versus current density curve are ident8ed along with their relative contribution to the electrical efficiency losses

acid fuel cell (PAFC) Pt alloys developed by UTC showed a higher kinetic performance than the pure Pt analogues. The gain in performance was of the order of 28 mV in the linear region of the Tafel plot. Since the Tafel slope was 90 mV decade-' at the operating temperature of 180°C, the gain cor- responds to a 2-fold increase in the catalytic activity for oxygen reduction and a 2% increase in the electrical efficiency of the fuel cell. The most stable system selected for commercialisation in the PAFC was found to be a ternary PtCr alloy sup- ported on a graphitised furnace carbon black.

Kinetics Based on the improved performance in the

PAFC, there has been much investigation of the 'intrinsic' kinetic performance in the PEMFC of a range of Pt-based metal alloy electrocatalysts (9, 10). We prepared PtFe, PtMn, PtNi, PtCr, PtZr and PtTi alloys, containing 20 wt.% Pt, supported on Vulcan XC72R (9). X-ray diffraction confirmed from a contraction (Fe, Mn, Ni, Cr, Ti) or expan- sion (Zr) of the face centred cubic Pt lattice parameter, that Pt alloys were formed, at an atom- ic ratio of Pt to base metal of 75:25. The kinetic performance in the cathode of MEAs, prepared by hot pressing catalysed substrates to Dow XUS13204.10 membrane, was evaluated in a small single cell. The single cell operating conditions were selected to minimise the cell potential losses at the anode and to reduce the mass transport losses

PLafinrrm Metah Re#., 2002, 46, (1) 6

Fig. 4 7ufel plots .showing tlie 'intrinsic ' kinetic benejit fbr o q g e n reduction of PtFe, PtMn and PtCr alloys conipured to pure PI supported on Vulcan XC72R (it 20 wr. O/c PI. The MEAs (< I /rig PI em-') Lire based on catalysed substrates bonded to Dow XUSl3203.10 mernhrane electrolyte. A srnall single cell (25 cni') is operuting at 75°C. in hvdrogen/o.rygen, 30W377 kPa. 300/300 cin-' niin 75"C/85"C bottle liumidification

950

5 E W. 900 W

LL

u A- 5 850 I-

W + 0 a _I

v BOO

750 100 1000 10000

SPECIFIC ACTIVITY, P A crn-'l P t )

at the cathode. This was achieved by running the single cell on pure hydrogen and oxygen, and by using high reactant gas flow rates. This procedure ensured that the kinetic performance of the cathode was controlling cell performance.

The 'intrinsic' kinetic performance of the Pt alloy and the pure Pt electrocatalysts was com- pared on a specific activity basis to take account of any differences in the Pt surface area of the mate- rials operating in the fuel cell. Specific activities were calculated from the current density, i, and the electrode Pt surface area (EPSA):

EPSA is a measure of the maximum Pt surface area available for reaction in the cathode. The cath- ode EPSA was measured by employing cyclic voltammetry i n - s h in the single cell to measure the protonic contact between the proton conducting electrolyte and the Pt electrocatalyst. This involved measuring the charge for the electrooxidation of CO to COZ from the cyclic voltammogram and cal- culating the EPSA based on the established ratio of charge to Pt surface area for the catalyst (4).

Tafel plots were constructed by obtaining the MEA area resistance, R, using a non-linear least squares analysis of Equation [l]:

i/EPSA (A cm-z/cmz Pt cm-3

Ecd = E,- b logi -iR "I where E,a is the cell potential, E, is a constant which is dependent on the cell operating conditions

and the cathode electrocatalyst, b is the Tafel slope for oxygen reduction and i is the current density.

Representative Tafel data for PtFe, PtMn and PtCr alloys compared to pure Pt are shown in Figure 4 (9). The kinetic performance of a range of Pt alloy cathodes was found to be of the order of 25 mV higher than the corresponding pure Pt cathode in the linear region of the Tafel plot. The Tafel slope was 60 mV decade-' at the operating temperature of 80°C. In contrast to planar bulk Pt studies, where a doubling in Tafel slope corre- sponding to a reduction in the Pt oxide surface at less positive cathode potentials is often reported (for example (ll)), there was no indication of a change in Tafel slope. This increased performance confirmed that the improved activity found in the PAFC also translated to the PEMFC environment with a 2.5-fold increase in the 'intrinsic' kinetic activity for oxygen reduction with the Pt alloys. Agam this resulted in a 2% increase in the electr- cal efficiency of the PEMFC.

Stability The stability of the Pt alloys was also examined

by monitoring the membrane and the anode of the MEA for base metal content both before and after operation in the single cell using electron probe microanalysis (EPMA). This showed that only the more stable PtCr, PtZr and PtTi base metal doys were not leached into the acidic membrane

Phtimm Metah b., 2002,46, (1) 7

>

CURRENT DENSITY, mA c 6 ’

Fig. 5 For air operation, in Ballard Mark 5E hardware, the kinetic benefit of a PtCr alloy cathode is masked by mass transport losses. The comparative performance ofthe PtCr alloy and a pure Pt cathode electrocatalyst is shown using aic helox (21% 02 in helium) and O2 as oxidants and H, as fuel. The MEAs (< I mg Pt substrates bonded to Nafion 115 membrane electrolyte. The cell is operated at 80°C in hydrogedaic helox, oxygen, 30W308 kPa. IS/.?, 2, 10 stoichiometry, full internal membrane humid$cation

are based on catalvsed

electrolyte. The PtFe, PtMn and PtNi alloys deposited Fe, h h and Ni, respectively, in the mem- brane and in the anode of the MEA. In the short time-scale of the measurements (200 hours), this did not lower the performance of the MEA, but with time, sufficient base metal would be leached from the cathode to lower the kinetic benefit from the Pt alloy electrocatalyst. In addition, unlike the PAFC, the proton conductivity of both the mem- brane and the proton conducting polymer present in the anode electrocatalyst layer will at some point be dramatically reduced. The base metal ions occu- py the sulfonic acid groups in the proton conducting electrolyte normally available for pro- ton conduction. This severely resmcts the type of Pt alloys that can be used in the PEMFC.

Practical Operation

The study then moved to evaluate the more sta- ble PtCr alloy in pie-commercial PEMFC stacks. Here, the aim was to realise the ‘intrinsic’ kinetic benefit of the Pt alloy under practical PEMFC operating conditions. A 40 wt.% PtCr alloy at a Pt to base metal atomic ratio of 75:25 was prepared on Vulcan XC72R. Catalysed substrate-based

MEAs were manufactured by hot pressing the electrodes to Nafion 115 membrane electrolyte. The results of tesang in a Ba.llard Mark 5E smgle cell are shown in Figure 5 which presents the cell potential versus current density plots for operation with air, helox (21% 0 2 in helium) and pure oxy- gen as oxidants, and hydrogen as fuel. In these measurements for each oxidant the gas flow rate was kept constant to prevent the water balance in the MEA being adjusted. This adjustment can lead to changes in the cathode performance resulting from modification of the proton conductivity of the electrocatalyst layer or from a change in the rate of oxygen permeability through the electrocat- alyst layer. Figure 5 shows that when using pure oxygen there is a clear kinetic benefit of 25 mV with the PtCr alloy compared to the pure Pt-based cathode. With more practical operation in air, however, the kinetic benefit was not realised and the PtCr alloy showed comparable performance to the pure Pt cathode.

We make use of air, helox and pure oxygen as oxidants to distinguish between oxygen g a s diffu- sion and oxygen permeability limitations in the cathode. The helox gains (the helox-air perfor-

Phtinwn Metah Rm, 2002,46, (1) 8

Fig. 6 For air operation in Ballard Mark 5 E hardware underfull humidification conditions a modified PtCr alloy recovers the ‘intrinsic’ kinetic benefit from the alloy. The MEAs (< I mg Pt cm ’) are based on catalysed substrates bonded t o Nafion 115 membrane electrolyte. The cell is operated at 80°C in hydrogetdair; 30W308 kPa, I S / . ? .stoichiometp, ,full internal membrane humidification

OPERAT I NG LIFETIME, hours

mance) are associated with oxygen gas diffusion losses and the oxygen gains (02-helox perfor- mance) with oxygen permeability losses. Figure 5 shows that the air and helox performance is com- parable for the PtCr alloy and the pure Pt electrocatalyst. The helox gains are therefore com- parable. There are no additional oxygen gas diffusion limitations present in the cathode with the PtCr alloy. Figure 5 also indicates that the oxy- gen gains are larger for the PtCr alloy. This confirmed poorer rates of oxygen permeability in the electrocatalyst layer with the PtCr alloy, because the PtCr alloy is more hydrophilic than the pure Pt catalyst. This resulted in the PtCr cathode retaining more of the liquid water produced by oxygen reduction. Consequently the oxygen permeability in the cathode electrocatalyst layer was reduced.

Armed with this information the PtCr alloy electrocatalyst was made more hydrophobic by the addition of a very small quantity of a third metal that was both stable in the acidic environment and did not disrupt the PtCr alloy structure. The per- formance of this modified PtCr alloy was comparable on pure oxygen and hydrogen opera- tion to the precursor PtCr alloy. On operation in air, however, the modified PtCr alloy showed the 25 mV improvement in activity predicted by the kinetic performance. Further, as shown in Figure 6, this improved performance was retained for 500 hours of continuous operation. Examination of the MEA by EPMA indicated the Cr had not been transported into the membrane or into the anode electrocatalyst layer.

Thus, by adjusting the MEA water balance - which can probably be achieved by a number of routes - the ‘intrinsic’ kinetic benefit of the Pt alloys can be realised under a variety of practical PEMFC operating conditions. Work optimising the benefit of the Pt alloys over the required oper- ating lifetimes is on-going.

Why Are Pt Alloys More Active for Oxygen Reduction?

There has been much debate regarding the rea- sons for the hgher activity of Pt alloys for oxygen reduction - dating from the early work on PAFCs. Among the reasons put forward (see, for example (8)) are: 0 the measured improvement in the stability to sintering, 0 surface roughening due to removal of some base metal, which increases the Pt surface area, 0 preferential crystal orientation, 0 a more favourable Pt-Pt interatomic distance, 0 electronic effects and 0 oxygen adsorption differences due to modified anion and water adsorption.

While it is not possible to be conclusive, the improved oxygen reduction activity in the PAFC has been linked to a reduction in the Pt particle size effect with all Pt alloys (12). That larger Pt particles are more active for oxygen reduction in the PAFC with pure Pt electrocatalysts has been proven by a number of groups (see, for example, (13)). For Pt alloys, Pt crystallites in the range from 2 to 4 nm, do not show a reduction in specific

Pkztinum Metah Rcv., 2002,46, (1) 9

n

$ 0 50 100 150 200 250 300 350 400 450 500 550 600 0 CL 2 u

EPSA, crn2 P t ~ m - 2

activity thereby restoring some of the lost activity in this particle size range that is evident with pro- gressively smaller pure Pt crystallites. It was proposed that the Pt surface atoms restructure during oxygen reduction resulting in a perfor- mance loss and that the base metal within the Pt crystallite prevents the restructuring (12).

However, the existence of the particle size effect in the PEMFC has not been proven, although, Wilson et a/. (14) have suggested it does exist. At Johnson Matthey the Pt particle size effect in the PEMFC was probed by measurement of the kinetic performance of a series of cathodes prepared from unsupported Pt black and 40 wt.% Pt electrocatalysts on different carbon supports. The plot of kinetic current for oxygen reduction versus the EPSA for the different cathodes is shown in Figure 7. The results show that at a given EPSA there appears to be a clear trend of hlgher oxygen reduction activity from larger Pt crystal- lites. The Pt black (5.7 nm) and the Shawinigan (5.5 nm) electrocatalysts give much higher oxygen reduction activity than the Pt electrocatalysts sup- ported on Vulcan XC72R (3.5 nm) and the BP 2000 (2.1 nm) carbon supports. The slopes of the lines in Figure 7 reflect the specific activities of the different Pt electrocatalysts. The specific activ- ities for oxygen reduction at 900 mV, iR free (vs. NHE) are 1.85 (Pt black), 0.94 (Shawinigan), 0.48 (Vulcan XC72R) and 0.34 (BP 2000) mA cml’ Pt.

While these results do seem to support the exis- tence of the Pt particle size effect in the PEMFC,

Fig. 7 The relationship between the kinetic performance for owgen reduction and the cathode EPSA jbr a runge of cathode electrocatulysts with different electrochemical ureas (ECA.0. The MEAs are based on catal.ysed substrates bonded to Nafion 115 membrane electrolyte. The Ballard Mark 5E single cell is operated at R O T , in hydrogen/ovgen, 30W30X kPu. I .5/10 stoichiometty full internal membrane humidification

care must be exercised. In the PEMFC the size and structure of the proton conducting electrolyte in the electrocatalyst layer mean that the utilisation of the Pt is a complicating factor that cannot be separated from the particle size effect in Figure 7. For example, the Pt black electrocatalyst shows a much higher specific activity for oxygen reduction than 40 wt.% Pt supported on Shawinigan, although the Pt crystallite sizes are comparable. It seems that the utilisation under load is higher for the Pt black system, due to a more favourable interaction with the protons in the aqueous per- fluorosulfonic acid solution present in the electrocatalyst layer. This is perhaps not surprising because of the lack of carbon in the unsupported Pt black to mask the Pt surface area from the pro- ton conducting polymer.

The effect of Pt utilisation is, however, likely to be less dramatic when comparing the different carbon supported electrocatalysts. Some indica- tion of this is shown in Figure 4 by the relative ‘intrinsic’ kinetic performance of the heat treated Pt and the pure Pt cathodes.

The effect of heat treating an electrocatalyst to produce larger Pt crystallites should minimise the effect of utilisation since the carbon support is unchanged. For the 20 wt.% Pt on Vulcan XC72R electrocatalyst, the heat treatment sintered the Pt crystallites from 2.2 to 3.5 nm corresponding to a CO metal area reduction from 100 to 60 m2 g-’ Pt. The ‘intrinsic’ kinetic performance from the heat treated Pt cathode is some 10 mV higher. It seems

Phhnnm Metah Rev., 2002,46, (1) 10

Table II Kinetic Current Versus EPSA for Type A Cathodes

EPSA, cm2 Pt c m 2

125 240 365 520

Cathode Pt loading,

0.73 0.97

Current density at 900 rnV (iR free),

mA cm-*

70 120 190 255

Specific activity at 900 mV (iR free),

mA cm-2 Pt

0.56 0.50 0.52 0.49

plausible that the Pt particle size effect is also evident in the PEMFC. This would also explain why the Pt alloys are more active in the PEMFC (a reduction in the Pt particle size effect). Indeed, that the Pt alloys are more active for oxygen reduc- tion does itself provide support for the existence of the particle size effect in the PEMFC. It would also explain why no performance benefit was observed with planar bulk Pt alloy cathodes in acid electrolytes (1 1).

Alternatives to Pt-Based Cathode Electrocatalysts

Pt alloys offer a performance gain of 25 mV, which increases the electrical efficiency of the PEMFC by 2%. This is the limit of the perfor- mance gain that can be expected using the approach of modifymg the Pt face centred cubic lamce structure by alloying with base metals. If much more of the 300 mV available at the cathode is to be recovered, then a fundamentally different approach is required.

The search for alternative oxygen reduction electrocatalyst materials - to Pt - has been the subject of extensive research over a number of years. Appleby and Foulkes (8) have reviewed much of the research. Recent literature has high- lighted ruthenium-based chalcogenides (15, 16), pyrolysed Fe porphyrins (17) and metal carbides (1 8) as offering significant oxygen reduction activ- ity. Indeed Co-based macrocydic compounds have been used commercially in the cathode of alkaline-based metal-air batteries. However, to date, in acid, none of the alternative electrocatalysts

have been as active as Pt; their durability has to be established and gas diffusion electrode structures have to be prepared with the electrocatalysts pre- sent in high surface area.

A possible area for initial focus might be in the cathode of the direct methanol fuel cell (DMFC), since in contrast to Pt the alternative electrocata- lysts are not deactivated for oxygen reduction by the methanol transported from the anode. If the electrocatalysts are not competitive with Pt in the DMFC they will fall far short of the requirements for the PEMFC.

The focus of the continued search for the elu- sive electrocatalyst for oxygen reduction in acid media should be on the development of materials with the required stability, and greater activity than Pt. Additionally, t h i s will require that the electro- catalyst be prepared in a high surface area form to compete effectively with carbon supported Pt electrocatalysts. Electrocatalysts which are less expensive than Pt, but which are less active, will not move PEMFC technology forward. The drive must be to raise system efficiencies for stationary applications, to lower electricity costs, and to raise power densities for transportation applications, so that capital costs are lowered. This requires more active electrocatalysts than Pt to raise the MEA performance.

Improved Utilisation of the Cathode Electrocatalyst

While the development of much more active oxygen reduction electrocatalysts is important, it represents a significant challenge. As a result,

Phtinwm Me& h., 2002,46, (1) 11

mg Pt c r r 2

0.27 0.44

attempts to raise the cathode performance have focused on improving the utilisation of Pt (the most effective oxygen reduction electrocatalyst).

The results of our work, which produced the much higher performance from the Pt black cath- odes shown in Figure 7, pointed to the possibility of significant performance gains available from improving Pt utilisation with carbon supported electrocatalysts. Different approaches to electrode design have been adopted, depending on the target MEA Pt loadings. Type A cathodes were devel- oped to perform at higher electrode Pt loadings (to 1.0 mg Pt cm ) and Type B cathodes were designed for performance at lower electrode Pt loadings (< 0.25 mg Pt cm-'). Such low Pt loadings are necessary to penetrate the large automotive market.

Table I1 shows clearly that for a range of Type A cathodes based on 40 wt.% Pt supported on Vulcan XC72R as the EPSA is increased (when the Pt loading is increased) the kinetic current for oxy- gen reduction (from pure hydrogen and oxygen operation) increases in direct proportion. This is shown by the similar specific activities calculated from the EPSAs. For these cathode structures the EPSA gives a relative measure of the active Pt sur- face area in the electrodes. This performance sc+ is particularly useful for some stationary and portable applications where higher Pt loadings can be tolerated economically to achieve an improved kinetic performance.

09.

> O B '

< 0 7 . : 0 6 . - I-

+ : 0 5 A d 0 4 -

0 3.

0 2,

However, for applications requiring lower MEA Pt loadings higher performing cathodes than Type A can be produced. It is possible to increase the kinetic performance over that achieved with Type A cathodes. Maximising the interaction between the Pt crystallites and the proton con- ducting polymer (which has a complex structure to accommodate the sulfonic acid groups and their associated protons and water of hydration) can raise the utilisation of the Pt electrocatalyst. Type B cathodes have been developed at electrode Pt loadings of less than 0.25 mg Pt cm-' with improved EPSAs and improved performances - compared to Type A structures. With Type B cathodes, prepared from 40 wt.% Pt supported on Vulcan XC72R, specific activities of ca. 1 mA cm-' Pt at 900 mV, iR free (vs. NHE) have been achieved. This represents a doubling in the specif- ic activity for oxygen reduction (see values for Type A cathodes in Table II) corresponding to a doubling in the Pt utilisation under load.

Most important, as shown in Figure 8, the kinetic benefit translated to an improved MEA performance under practical operating conditions with air as oxidant and hydrogen as fuel. A kinetic performance gain of 20 mV was evident at low current densities as projected by a doubling in the Pt utilisation. This was reflected in an increased EPSA (230 cm' Pt cm-' at 0.23 mg Pt cm-').

Figure 8 shows that at higher current densities there were additional performance gains (due to a

-0- Type Acathode, 020rng Pt cni' 4 Type Bcathode, 017 mg Pt cni'

Fig. 8 For low electrode Pt loadings (< 0.25 mg Pt em') Type B cathodes show the higher performance due to improved Pt electrocatalyst utilisation, lower electrode resistance and enhanced mass transport performance. The MEAs (< 0.5 mg Pt em ') are based on catalysed substrates bonded to Nafion 115 membrane electrolyte. The Ballard Mark 5E single cell is operated ut 80°C in hydrogerduir: 30W308 kPa, 1.92 stoichiomern!. full internal membrane humidification

Platinum Metuh Rev., 2002, 46, (1) 12

0 200 400 600 800 1000 1200 1400 1600 CURRENT DENSITY, mAcm-'

reduction in both the electrode resistance and mass transport performance losses) using the Type B cathodes. This is reflected by a lowering in the slope in the pseudo-linear region of the cell poten- tial vs. current density graph (Figure 8). The improved cathode design increased the electrical efficiency at low current densities by 2%, but at higher current densities the electrical efficiency was raised by as much as 7%. As the electrode Pt loadmg was raised above 0.25 mg Pt cm-’, howev- er, the increased EPSA of the Type B cathodes did not translate to a higher performance. This can be attributed to thickness limitations in the electrode due to a reduction in the rate of oxygen perme- ability through the cathode (2).

Conclusions Most of the performance loss from the ther-

modynamic potential of the PEMFC is due to the cathode. Aqueous-based inks, prepared from hgh Pt loading electrocatalysts with high Pt dispersion and perfluorinated sulfonic acid polymer solu- tions, have been used to prepare relatively thin electrocatalyst layers. Employing Pt-based metal alloy electrocatalysts prepared on traditional car- bon black supports in the electrocatalyst layer produces a 25 mV performance gain. This repre- sents a 2% increase in the electrical efficiency of the PEMFC.

Only the more stable Pt-based metal alloys, such as PtCr, P a r , PtTi, can be used in the PEMFC, due to dissolution of the base metal by the perfluorinated sulfonic acid in the electro- catalyst layer and membrane. For improved per- formance from the Pt-based metal alloys, it is necessary to tailor the electrocatalyst layer to achieve the optimum MEA water balance under the selected PEMFC operating conditions. This can be achieved by modifjmg the Pt alloy electro- catalyst with a third metal and probably also by altering the electrocatalyst layer design.

There are strong indications that the particle size effect is present in the PEMFC as well as in the PAFC. The performance gain from the Pt- based metal alloys is probably linked to a reduction in the particle size effect in the Pt crystallite range from 2 to 4 nm. It is tentatively proposed that the

base metal prevents the Pt surface atoms from restructuring during oxygen reduction.

To develop cathode electrocatalysts with high- er performance than the Pt-based metal alloys requires a completely new approach. At present there are no alternative cathode electrocatalysts to Pt. All others are less active and their durability is unproven. The focus of future research needs to be on improved performance and not on lower cost alternatives to Pt with comparable or lower performance.

Different electrocatalyst layers have been developed for high Pt loadings and for Pt loadings below 0.25 mg Pt ern-'. Improved performance can be achieved at the lower electrode Pt loadings by improving the utilisation of the cathode elec- trocatalyst. Secondary benefits are available due to proton diffusion and mass transport performance gains. At high electrode Pt loadings the perfor- mance gains are negated by oxygen permeability limitations.

Acknowledgement This paper is based on part of a keynote lecture given by

T. R Ralph at the 17th North American Chemical Society Meeting held at the Westin Coast Harbour Castle Hotel, Toronto, Ontario, Canada, from the 3rd to 8th June 2001.

References T. R. Ralph and G. A. Hards, Cbem. Ind, 1998, (9), 337 T. E. Springer, M. S. Wilson and S. Gottesfeld, J. Ekrtmcbem. Soc., 1993,140, 3513 L. Keck, J. S. Buchanan and G. A. Hards, U.S. Patent 5,068,161; 1991 T. R. Ralph, G. A. Hards, J. E. Keating, S. A. Campbell, D. P. Willdnson, M. Davis, J. St-Pierre and M. C. Johnson, J. Ekrtmcbem. Soc., 1997, 144, 3845 I. D. Raistrick in “Diaphragms, Separators and Ion Exchange Membranes”, eds. J. W. Van Zee, R E. White, K. Kinoshita and H. S. Burney, The Electrochemical Society Softbound Proc. Series, Pennington, NJ, 1986, PV 86-13, p. 172 S. Gottesfeld and T. A. Zawodzinski, Ah. Ekrtmcbem. Sci Eng., 1997, 5,231-240 J. Denton, J. M. Gascoyne and D. Thompsett, Eumpeun Patent 731,520; 1996 J. Appleby and F. R. Foulkes, “Fuel Cell Handbook”, Krieger Publishing Company, Malabar, Florida, 1993, Chapter 12 T. R. Ralph, J. E. Keating, N. J. Collis and T. I. Hyde, ETSU Contract Report F/02/00038,1997

Pkatinum Metals Rev., 2002,46, (1) 13

2

3

4

6

7 .

8

9

5

1

10

11

12

13 14

15

16

17

S. Mukerjee and S. Srinivasan,]. Ekcfmand Cbem., 1993,357,201 S. Gottesfeld and T. A. Zawodzinski, Adv. Ekcfmcbem. Sci Eng., 1997, 5,20>217 J. S. Buchanan, G. A. Hards, L. Keck and R. J. Potter, in 1992 Fuel Cell Seminar Program & Abstracts, Tuscon, AZ, 1992, p. 536 K. Kinoshita, J. Ekcfmcbem. SOC., 1990,137,845 M . S. Wilson, F. H. &on, K. E. Sickahs and S. Gottesfeld,]. Ekcfmcbem. SOC., 1993,140,2872 N. Alonso-Vante and H. Tributsch, Nature, 1986, 323,431 R W. Reeve, P. A. Christensen, A. Hamnett, S. A. Haydock and S. C. Roy,]. Ekxfmbz. SOL, 1996,145,3463 G. 0. Sun, J. T. Wang and R. F. Savinell,]. AppL Ekcfmcbem., 1998,28,1087

18 R Cote, G. Laande, G. Faubert, D. Gauy, J. P. Dodelet and G. Denes, 1. New Matm Ekcfmcbem. Systems, 1998, 1, 7

The Authors Tom Ralph is Product Development Manager at the Johnson Matthey Technology Centre, responsible for MEA design. He has been working with PEMFCs since 1991 and with developing all aspects of the MEA - electrocatalyst, electrocatalyst layer, gas dif- fusion substrate and solid polymer membrane - and in integrating MEAs into customer hardware.

Martin Hogarth is a Senior Scientist at the Johnson Matthey Technology Centre and has worked in the area of DMFCs since 1992. His main interests are in the development of new catalyst materials and high-performance MEAs for DMFCs. More recently his interests have expanded into novel high-temperature and methanol impermeable membranes for the PEMFC and DMFC, respectively.

The direct methanol fuel cell (DMFC) is a vari- ant of the proton exchange membrane (PEM) fuel cell and uses aqueous methanol directly without prior reforming. In the DMFC methanol is con- verted to carbon dioxide and hydrogen at the anode. The hydrogen then reacts with oxygen, as in a standard PEM fuel cell. Conventional materi- als for DMFCs include platinum-ruthenium (Pt-Ru) for the electrode electrocatalysts and car- bon in various forms as the electrocatalyst support. Electrocatalysts with hgh activity for methanol oxidation are essential for improved performance of DMFCs. Such catalysts are generally prepared as unsupported metal colloids or nanocomposites with the metal nanoparticles supported on an elec- trically conducting carbon of hgh surface area. Mixed metal Pt-containing catalysts are presently used for methanol oxidation.

Now, scientists from the Department of Chemistry at Vanderbilt University, with a col- league from the Corrosion Research Center, University of Minnesota, U.S.A., have developed a Pt-Ru/graphitic carbon nanofibre (GCNF) nano- composite which exhibits hlgh relative performance as a DMFC anode catalyst (E. S. Steigerwalt, G. A. Deluga, D. E. Cliffel and C. M. Lukehart, J. Phy. Chem. B, 2001,105, (34), 8097-8101).

As part of ongoing studies of new synthetic strategies for preparing metal alloy/carbon com- posites, they prepared and characterised a Pt-Ru/GCNF composite, where the GCNF sup-

Platinum-Ruthenium Anode Catalyst for DMFC port has the 'herringbone' atomic structure. The source of both metals was the molecular precursor (q-C2H4) (CI)Pt(p-C1)2Ru(C1) (q3:~3-2,7-dimethyloc- tadienediyl).

Reductive decomposition of the precursor formed widely dispersed Pt-Ru nanocrystals, and a multistep deposition procedure ensured total metal content of - 42 wt.% at bulk Pt/Ru atomic ratio of - 1:l. The metal alloy nanoclusters had average particle size of 6 nm (calculated from XRD peak widths) or 7 nm (measured directly from TEM images). Small amounts of Ru metal and oxidised Ru species were also present.

When used as an anode in a working DMFC, the composite enhances fuel cell performance by - 50% relative to that recorded for an unsupported Pt-Ru colloid anode catalyst. Further work on the metal alloy/GCNF anode catalysts is envisaged.

Fuel Cell Catalysts Brochure Alfa Aesar has just published a new 4-page

brochure highlighting a range of noble metal, HiSPECm fuel cell catalysts. Johnson Matthey manu- factures HiSPEC"?" brand catalysts which provide superior performance in proton exchange membrane and direct methanol fud cells. The HiSPECm range consists of single and bimetallic supported and unsupported catalysts which can be used on both anodes and cathodes.

Copies of the new Fuel Cell Brochure can be obtained from Alfa Aesar: Tel: +800-343-0660; Fax: +800-322-4757; or E-mail: info@alfa.com.

Pkztinum Metals REX, 2002, 46, (1) 14