Topics in Catalysis

, 52:2117

Nanostructured Carbons as Platinum Catalyst Supports for Proton Exchange Membrane Fuel Cell Electrodes

Authors

    • Laboratoire de Génie Chimique (B6a)Université de Liège
  • Sandrine Berthon-Fabry
    • Mines ParisTech, Centre Énergétique et Procédés
  • Marian Chatenet
    • Laboratoire d’Électrochimie et de Physico-chimie des Matériaux et des Interfaces (LEPMI)UMR 5631 CNRS/Grenoble-INP/UJF
  • Julien Marie
    • Mines ParisTech, Centre Énergétique et Procédés
  • Mathilde Brigaudet
    • Mines ParisTech, Centre Énergétique et Procédés
  • Jean-Paul Pirard
    • Laboratoire de Génie Chimique (B6a)Université de Liège
Original Paper

DOI: 10.1007/s11244-009-9384-0

Cite this article as:
Job, N., Berthon-Fabry, S., Chatenet, M. et al. Top Catal (2009) 52: 2117. doi:10.1007/s11244-009-9384-0

Abstract

To improve mass transport in the catalytic layers of proton exchange membrane fuel cells, the usual Pt catalyst support (carbon blacks) can be advantageously replaced by carbon aerogels or xerogels. The pore texture of such materials can indeed be tailored, which enables choosing an adequate pore texture minimizing diffusional limitations within the catalytic layers.

Keywords

Carbon xerogelsCarbon aerogelsPt catalystsPEM fuel cells

1 Introduction

Carbon blacks [1, 2] have been used in numerous catalytic applications, and remain the elected material for fuel cell electrodes [1]. These carbons are composed of microporous near-spherical particles of colloidal sizes (~20–60 nm), coalesced together as aggregates (1–100 μm). The global pore texture of a catalytic layer composed of carbon black supported catalyst strongly depends on the properties of the carbon selected (mainly the grain size and the specific surface area, both being hardly tunable) and on the electrode processing conditions. In addition, mass transport limitations are often observed, thereby limiting the electrode performances. This major drawback calls for the development of new carbon materials with controllable and tunable pore texture. In addition, and especially for electrochemical applications, high purity of the carbon supports is essential [1].

Among the numerous new carbon materials developed, carbon xerogels, aerogels and cryogels were recently envisaged as an alternative to carbon blacks in proton exchange membrane (PEM) fuel cell electrodes [36]. Indeed, their pore texture can be adjusted within a wide range: for instance, internal diffusional limitations in gas phase heterogeneous catalysis can be suppressed by replacing activated carbon supports by carbon gels [7]. The present paper aims at showing the advantages of carbon gels as electrocatalyst supports for PEM fuel cell electrodes.

2 Experimental

2.1 Carbon Gel Preparation and Characterization

Carbon gels, i.e. aerogels, xerogels and cryogels, are obtained by polycondensation of resorcinol (R) with formaldehyde (F) in a solvent, followed by drying and pyrolysis in inert atmosphere [8, 9]. The difference between aerogels, xerogels and cryogels lies in the drying procedure (supercritical drying, evaporation or freeze-drying, respectively). RF aqueous gels lead to very pure carbon materials, and their morphology can be modified via the synthesis conditions. As shown in Fig. 1, carbon gels issued from resorcinol-formaldehyde polymeric systems consist of spherical-like interconnected nodules, the size of which can be adjusted through an adequate choice of the composition of the gel precursor solution: in particular, the pH of the RF solution is the crucial operating variable, although the solvent content and the nature and amount of additives [10, 11] also influence the morphology of the final material. Depending on the pH, the average size of the nodules can be increased from about 5 nm up to 3–4 μm. As a consequence, the size of internodular voids, i.e. the pore size distribution, depends on the gel synthesis conditions.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-009-9384-0/MediaObjects/11244_2009_9384_Fig1_HTML.jpg
Fig. 1

Morphology of carbon gels prepared at various pH a pH 6.50, b pH 6.00, c pH 4.00. The size of the carbon nodules is completely determined by the composition of the RF solution, and does not depend on the drying technique (here: evaporative drying)

The drying step is also determining [12]: supercritical drying and freeze-drying were developed to avoid sample crushing upon drying due to capillary tensions at the liquid–gas interface. However, the evaporative drying of resorcinol-formaldehyde aqueous gels, without any pre-treatment, followed by pyrolysis under inert atmosphere, can produce carbon materials with sufficiently developed pore texture provided the synthesis variables of the gel are correctly chosen [13]. Carbon gels display a bimodal distribution featuring micropores (~0.25 cm3 g−1), located inside the nodules, and meso- or macro-pores of different size (~5 nm–3 μm) and total volume (0.5–5 cm3 g−1) corresponding to internodular voids (Fig. 1). Carbon aerogels and cryogels display a wider pore texture range than carbon xerogels: all pore sizes can be obtained by using evaporative drying, but the pore volume and the pore size are strongly correlated when the samples bare relatively small pores [12]. So, it is not possible to produce carbon xerogels with both small pores (<30 nm) and high pore volumes (>1.5 cm3 g−1), which can be done by supercritical drying. However, heterogeneous catalysis usually requires large pore size (>30 nm) to enable easy mass transfer, and in that extent the evaporative drying leads to sufficient pore volume [7].

2.2 Synthesis of Highly Loaded Pt/Carbon Gel Catalysts for PEM Fuel Cells

Carbon supported metal catalysts with very good dispersion can be obtained by impregnation of carbon gels [7, 14, 15], but the metal weight percentage of these catalysts (<5 wt%) remains too low for use in PEM fuel cells, especially in the case of cathodes of air/H2 devices. Indeed, in order to minimize ohmic and transport losses within the electrode, and to compensate for both the sluggish oxygen reduction reaction rate and the usual lack of contact between Pt and the ionomer (see Sect. 2.3), two antagonistic conditions must be fulfilled: (i) the thickness of the catalytic layer must be low enough, and (ii) the metal loading of the electrode must be high enough. This implies the selection of catalysts with high metal weight percentage: usually, 20–80 wt% [1].

In order to prepare Pt/carbon gel catalysts with high metal weight percentage, Marie et al. [3, 5] developed a method which included an impregnation step with a diluted H2PtCl6 aqueous solution, followed by direct metal reduction in liquid phase with NaBH4. The obtained catalysts displayed high metal content (~30–35 wt%) and were thus used at the cathodes of PEM fuel cells [36]. The Pt particle size distribution was bimodal (Fig. 2a): small Pt particles (2–5 nm) were surrounded by large crystals and aggregates (10–30 nm). The latter constituted about 60% of the Pt [4], which is not compatible with high metal utilization. In fact, the amount of Pt that can be deposited by impregnation onto a support while keeping a high metal dispersion is limited by the interaction between the support and the metal precursor and by steric effects [16].
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-009-9384-0/MediaObjects/11244_2009_9384_Fig2_HTML.jpg
Fig. 2

Pt/carbon xerogel catalysts prepared on the same support (pore size range 50–85 nm) a impregnation with H2PtCl6 followed by direct reduction in liquid phase with NaBH4 (30 wt% Pt [4]); b SEA method, one impregnation (7.5 wt% Pt [17, 18]); c SEA method, three successive impregnations (22.7 wt% Pt [18])

High dispersion is achieved only if the non-reduced precursor itself is well dispersed at the surface of the support, which can be enhanced by maximizing the electrostatic interactions between the support and the precursor. Based on these observations, the so-called Strong Electrostatic Adsorption (SEA) method [16] was recently used to prepare Pt/C catalysts with high metal loading: electrostatic interactions between the metal precursor and the support were maximized by adjusting the pH of the carbon/water/Pt precursor slurry to the adequate value, which depends on both the surface chemistry of the support and the nature of the Pt precursor. In the case of the impregnation of carbon xerogels with H2PtCl6 aqueous solutions, the initial pH leading to the highest Pt uptake was found to be 2.5 [17]; the corresponding maximum Pt loading was ranging from 8 to 10 wt% while keeping very small Pt particles after reduction under H2 flow (~2 nm, homogeneous, see Fig. 2b). It is thus not possible to increase the amount of Pt deposited without altering the excellent metal dispersion obtained: this explains why 30 wt% Pt catalysts prepared in one single impregnation followed by liquid-phase reduction displayed lower metal dispersion.

Successive impregnations via the SEA method can be performed to increase the Pt weight percentage up to acceptable values [18]. It was shown that the adsorption sites of the support are completely recovered after filtration, drying and reduction of the catalyst: consecutive impregnation steps lead to increasing regularly the Pt weight percentage without modifying the Pt particle size (~2 nm, see Fig. 2b, c). The SEA method allows controlling finely the structure and the morphology of the metal nanoparticles (average particle size, size distribution, degree of agglomeration) and thus may be used to minimize the mass of Pt used in PEM fuel cells.

2.3 Pt/Carbon Gel Catalysts in PEM Fuel Cells

Membrane-Electrode Assemblies (MEAs) were prepared with various carbon gels as cathodic catalyst supports and tested in an air/H2 monocell device at 70 °C. The MEAs were prepared by the so-called decal method [4, 5], and a commercial Pt/carbon black catalyst was used at the anode side. The cell test bench and the measurement procedures are completely described in reference [4]. All conditions (thickness, Nafion® content of the catalytic layer, pretreatments, etc.) were kept identical, except for the nature of the carbon support for the Pt nanoparticles. So as to compare the two procedures, both the impregnation with H2PtCl6 followed by reduction in liquid phase with NaBH4 [35] and the SEA method were used. The performances of the carbon gel supported catalysts were finally compared to that of a commercial electrode (Pt/carbon black).

3 Results and Discussion

As shown in Fig. 3a, a catalytic layer of a typical PEM fuel cell electrode is composed of Pt/carbon black catalyst particles, interparticle voids and ionomer, hot-pressed between a proton-exchange membrane and a diffusion layer (carbon felt) [1]. To be active, the Pt particles must be in contact with the carbon support and connected to the membrane via the ionomer (Nafion®). In addition, reactants and products must circulate easily through the catalytic layers: (i) the catalyst must be accessible to the gas reactant (H2 or O2), which percolates through the porous structure of the catalytic layer; (ii) protons and electrons have to be collected by the ionomer (Nafion®) and the catalyst support, respectively, and driven to the membrane (protons) and the current collector (electrons); (iii) water must be transported to the membrane for humidification, and the excess eliminated to avoid electrode flooding. Practically, a large fraction of the Pt can be inactive due to a lack of contact between Pt and Nafion®, and to the presence of liquid water within the pores of the cathode. This is compensated by the use of catalytic layers with very high Pt loadings that increase the electrode cost. As carbon black particles are loosely bonded and packed together in aggregates, the pore structure of the catalytic layer not only depends on the nature of the carbon black, but also on the electrode processing. Thus, in carbon black-based air-fed cathode, where oxygen, proton and water transports are involved in the oxygen reduction reaction, high potential losses due to diffusional limitations (themselves related to the pore texture of the catalytic layer) offset the cell performances [19].
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-009-9384-0/MediaObjects/11244_2009_9384_Fig3_HTML.gif
Fig. 3

Structure of a PEM fuel cell electrode: membrane, catalyst layer and gas diffusion layer (GDL). a Carbon black as catalyst support: the catalytic layer is composed of aggregates of loosely bonded carbon black particles containing Pt, b carbon gel as catalyst support: the catalytic layer is composed of carbon gel micromonoliths made of a rigid 3D structure, insensitive to compression and with a defined pore texture. The Nafion® network and Pt particles are represented in light grey and black, respectively

On the contrary, the meso/macropore texture of catalytic layers prepared from carbon gels is totally independent on the electrode processing and can be regulated by choosing correctly the composition of the pristine wet gel. In the electrode structure, the carbon black particle agglomerates are replaced by micromonoliths of carbon gel made of covalently bonded carbon nodules, which preserves the pores located in-between (Fig. 3b).

The total voltage loss of the cell with regard to the reversible H2/O2 cell voltage can be decomposed into several components [20]: (i) the cathode overpotential, ηORR, due to the slowness of the O2 reduction kinetics; (ii) the ohmic losses, ηOhm, due to the resistance of the proton migration through the membrane, the electronic contact resistances between the flow-fields plates and the diffusion media, and the contact between the carbon grains; (iii) the mass-transport losses, or ‘diffusion overpotential’, ηdiff, induced by slow O2 diffusion through the diffusion layer and the catalytic layer. In such systems, kinetic and mass transport losses of the anode can be neglected [20]. The first two contributions to the voltage loss can be measured: ηORR is obtained from measurements at low current densities, i.e. from data obtained in the near-absence of mass transport limitations and ohmic resistance; ηOhm is obtained by measurement of the ohmic resistance of the cell by impedance spectroscopy, performed in situ. Finally, all the other terms being known, ηdiff can be calculated.

As an example, Fig. 4 shows the contribution of diffusion to the voltage loss of several MEAs prepared with various carbon gels as supports. In all cases, the Pt/C catalysts were prepared by impregnation with H2PtCl6 followed by reduction in liquid phase with NaBH4 [35]. One observes that ηdiff depends on the pore size: in the xerogel series, ηdiff decreases regularly when the pore size increases from 40 to 300 nm. However, the aerogel (sample A), the maximum pore size of which is about 25 nm, display diffusion overpotential equivalent to that of the xerogel with the largest pore size (~250–300 nm). This can be explained by the pore volume difference: the pore volume of the aerogel is more than twice that of xerogel (4.8 vs. 2.2 cm3 g−1). These results show how adjusting the pore size of the catalyst support may influence mass transport limitations in PEM fuel cells.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-009-9384-0/MediaObjects/11244_2009_9384_Fig4_HTML.gif
Fig. 4

Diffusion overpotential versus experimental current density per surface unit of electrode. MEAs prepared with different carbon gel supports at the cathode. (filled triangle) Mesoporous carbon xerogel (pore size ~20–25 nm), (filled square) carbon xerogel with small macropores (pore size ~50–85 nm), (filled diamond) carbon xerogel with large macropores (pore size ~250–300 nm), (open circle) mesoporous carbon aerogel (pore size ~20–25 nm)

Besides mass transport, the cathode performance can be increased by increasing the global utilization ratio of the metal. Indeed up to 90% of the platinum particles of a catalytic layer may be inactive, because of diffusional limitations in the catalytic layer or because the truly active Pt surface is low [21]. The latter problem is due to a lack of contact between the Pt particles and the electrolyte (Nafion®), and to metal agglomeration observed in catalysts with high metal weight percentage. Figure 5 compares the mass activity of catalysts prepared on various supports and with different Pt deposition methods. First, the performance of the aerogel can be reached by using a xerogel with an appropriate pore size. Second, increasing the metal dispersion (SEA method) allows increasing the cathode performances in terms of Pt mass activity with regard to catalysts obtained by impregnation and reduction in liquid phase (Sect. 2.2): in the case of the carbon xerogel support, the current produced per mass unit of Pt, im, is almost doubled when the SEA technique (double impregnation) is used, which is attributed to the elimination of large Pt particles and aggregates obtained by impregnation and reduction in liquid phase. Third, the performances of the carbon gel-supported Pt catalysts can be compared to that of a commercial cathode containing Pt/carbon black catalyst. One observes that both electrodes display similar performances in the so-called ‘kinetic’ region (im < 1 kA gPt−1), where the mass-transport limitations are negligible. This indicates that the intrinsic mass activities of the Pt nanoparticles in these two electrodes are close, exhibiting no dramatic difference of Pt particles morphology. On the contrary, in the mass-transport controlled region (im > 1 kA gPt−1), the Pt utilization is clearly higher in the case of the SEA catalyst supported on carbon xerogel. Such performance enhancement is attributed to the decrease of mass transport limitations and to the improvement of Pt-Nafion® contact, both inherent to the carbon texture.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-009-9384-0/MediaObjects/11244_2009_9384_Fig5_HTML.gif
Fig. 5

Cell voltage versus Pt mass activity. (open circle) carbon aerogel (pore size ~20–25 nm), impregnation + liquid phase reduction [4]; (filled square) carbon xerogel (pore size ~50–85 nm), impregnation + liquid phase reduction [4]; (filled triangle) carbon xerogel (pore size ~50–85 nm), SEA method (double impregnation, reduction under H2 flow at 450 °C [18]), (cross) commercial cathode (Pt/carbon black, Paxitech)

4 Conclusions

Carbon gels issued from drying and pyrolysis of resorcinol-formaldehyde aqueous gels can advantageously replace carbon blacks as catalyst support for PEM fuel cell electrodes. Indeed, their pore texture can be regulated so as to minimize mass transfer limitations in the catalytic layer of the fuel cell cathode. The pore size and void fraction of the catalyst support are determining in the mass transport phenomena, and both parameters can be tuned via the pristine gel composition and drying procedure. Carbon xerogels, which are the easiest to prepare, can lead to performances close to that of carbon aerogels or cryogels provided that the pore size and volume are large enough. Finally, the global Pt utilization can be enhanced by replacing commercial Pt/carbon black catalysts by Pt catalysts supported on carbon xerogels and prepared by the ‘SEA’ method.

Acknowledgements

N. Job thanks the F.R.S.-FNRS (Belgium) for a postdoctoral researcher grant. The Belgian authors thank the Fonds de Recherche Fondamentale Collective, the Ministère de la Région Wallonne and the Interuniversity Attraction Pole (IAP-P6/17) for their financial support, and acknowledge the involvement of their laboratory in the Network of Excellence FAME of the European Union Sixth Framework Program. The French authors thank the Groupement des Écoles des Mines (GEM). N.J. and M.C. thank Egide and the WBI for funding within the Hubert Curien Partnership (Tournesol project #20389PF).

Copyright information

© Springer Science+Business Media, LLC 2009