High Performance PEM Fuel Cell with Low Platinum Loading at the Cathode UsingMagnetron Sputter Deposition

Platinum cluster formations have been investigated as a way to reduce the amount of Pt at the cathode of polymer electrolyte membrane fuel cells. One, two, and three layers of Pt (0.05mg/cm) sputtered directly on microporous layers of gas diffusion layers with and without interfacial carbon-Na�on layers and carbon-polytetra�uoroethylene (CP�FE) layers have been used as a cathode. Comparison with experimental results had showed that the best performance was obtained with three layers of Pt sputtered on carbon-Na�on containing 34.8 wt.% of Na�on and sputtered carbon-polytetra�uoroethylene containing 16.9 wt.% of polytetra�uoroethylene. High limiting current densities (>1.1 A/cm) have been reached with cathode Pt loading as low as 0.05mg/cm. SEM imagery and cyclic voltammetry characterization have been performed to consolidate this study. High Pt utilization can be showed by this method. e factor in�uencing Pt utilisation in the oxygen reduction reaction is intrinsically related to Pt clusters formation and helps in enhancing the PEMFC performance with low Pt loading.


Introduction
Several researches have been done in the latest decade to reduce the amount of Pt in polymer electrolyte membrane fuel cells (PEMFCs) principally in the cathode to reduce its weight, volume, and cost [1][2][3][4][5][6][7][8].e oxygen reduction reaction (ORR) occurring at the cathode of a PEMFC cell is catalyzed by high loadings of Pt to reduce signi�cantly the overpotential loss.e activation overpotential is the principal loss, and its magnitude depends on the reaction kinetic parameters.More precisely, it depends on the size of the exchange current density [9].erefore, improving reaction kinetic performance is done by increasing the exchange current density, which depends on reactant concentration, activation barrier, reaction sites, and temperature.ose factors hinder large-scale commercialization and have motivated intense research to low loading of Pt, more active, inexpensive, and stable.Promising directions include metal alloys, notably Pt-and Pd-based inorganic compounds such as chalcogenides and organic compounds such as transition metal macro-cycles, have been investigated [10][11][12][13].In addition of, the complexity of the catalyst layer has permitted the studies about coating methods, water management, Na�on content, and carbon supported Pt [6,[14][15][16][17][18][19].It is obvious that the approach of Pt alloying with nonnoble transition metals can create some active alloy catalysts for catalytic activity enhancement towards ORR [12].However, there are many concerns about the long-term stability of these Ptalloy catalysts due to leaching issues of the metals under the cathode-operating environment [12].A better utilization of low loading pure Pt must be fully investigated as it has the highest activity of elemental metals in order to improve catalyst layer performance.
Natarajan and Hamelin [20] have shown an increase in membrane electrode assembly (MEA) performance with three sputtered layers of Pt at the anode (total loading of 0.05 mg/cm 2 ) as compared with a standard single layer of Pt MEA.In the present paper, we have investigated the Pt cluster formation to reduce the Pt loading at the cathode of a PEMFC while retaining high performances using a magnetron sputter technique.Different con�gurations of the active layer based on the distribution of the Pt sputtered on carbon-Na�on (CN) and carbon-polytetra�uoroethylene (CP�FE) dry inks are used to fabricate MEAs.e MEA performance is analysed by its current voltage characteristic.Microscopic visualization of the electrodes as well as cyclic voltammetry measurements is carried out to get the structural relationship, the cluster formation, and the distribution of Pt. was used as a proton conductive agent and a binder for the catalyst layer.e ratio of Na�on (PTFE, resp.)solution in carbon at the cathode was controlled to 1 : 4. A ratio 1 : 1 of Na�on solution in carbon was prepared for the anode.e mixture was magnetically stirred for at least 24 h.e ink is then applied on a 5 cm 2 te�onized carbon paper (SGL, Sigracet) using the brushing method and dried for approximately 2 h in order to achieve a total carbon loading of 0.5 mg/cm 2 .Aer drying, multiple layers of sputtered Pt were applied onto the substrate surface to reach a total loading of 0.05 mg/cm 2 .e multilayer sputtering consisted of sputtered Pt on CN or CPTFE (Table 1).e electrodes obtained were used as cathodes.Anode electrodes with a Pt loading of 0.05 mg/cm 2 sputtered directly on the microporous layer (MPL) of carbon papers were used with all samples, except for samples EA2 and PA2 that have a CN ration of 1 : 1. Na�on 212 (DuPont) was used as the membrane.e MEA obtained was placed between two Furon gaskets.A hydraulic hot press was used to stick the MEA and the Furon under a 1000 psi pressure at 130 ○ C during 2 min aer being preheated during 1 min under atmospheric pressure.All assemblies were inserted between two graphite plates with serpentine �ow �elds and then placed in a single cell test �xture from Fuel Cell Technologies.�umidi�ed hydrogen (80%) and air (80%) was fed into the cell at 200 sccm at the anode and 360 sccm at the cathode.e experiments were carried out at a cell temperature of 80 ○ C. A standard MEA with a Pt loading of 0.2 mg/cm 2 for both of the anode and the cathode was also prepared for comparison.

Cluster Formation and Performance Analysis.
In order to permit the formation of Pt clusters, multiple layers of Pt were sputtered on the surface of the substrates using Automatic Sputter Coater model NSC-3000 from Nano-Master [21�.Argon is �rst introduced into the chamber in order to provide an environment in which plasma could be ignited and maintained.When the plasma is ignited, positive ions strike the target cathode, which then released Pt atoms by means of momentum transfer.ese Pt atoms form a vapor that condenses into a thin �lm on the substrate surface (located on the platen).is procedure is repeated one, two, or three times to get one, two, or three layers, respectively separated or not by CPTFE or CN layers until a total amount of Pt (0.05 mg/cm 2 ) is reached for each sample.Sputtering takes place at 10 −4 Torr where a better uniformity is achieved.Scanning electron microscopy (SEM) was performed to analyse the morphology and the elemental compositions of the Pt clusters.Finally, cyclic voltammetry was performed on a 3 mm glassy carbon working electrode for different con�gurations of sputtered Pt in order to compare their active surface area.

Results and Analysis
3.1.Polarization Curves.Figure 1 shows the polarization curves for MEA1, MEA2, and MEA3 with the same low Pt loading.In the low current density region where activation losses are dominant, little difference can be seen between them.When comparing the cell voltages in the ohmic region, a visible difference can be seen between the three MEAs.e ohmic resistance increases with the number of sputtered layers showing that the distribution of Pt has an in�uence on electron and proton accessibility to reaction sites.is can be due to contact resistance between layers.In the concentration region, it is easy to see that MEA3 has a higher limiting current density (1.422 A/cm 2 ) then MEA1 (1.017 A/cm 2 ) and MEA2 (1.079 A/cm 2 ).e maximum power (0.5517 W/cm 2 ) was reached with MEA3 and is almost two times higher than the one with MEA1 (0.3272 W/cm 2 ) with the same Pt loading.is gain of performance obtained with MEA3 needs to be pointed out here for that experimental result.is improvement may be due to an increase of surface reaction reached by the reactants at higher current densities caused by a higher porosity of the active layer.e pore volume has been calculated by BET analysis from nitrogen adsorption curves.Although this technique is not the most accurate for the scale of the average pore width (70 nm), it shows an augmentation of the pore volume for multiple Pt layers.e isolated Pt particles observed in an agglomerate catalyst that do not participate in electrochemical reactions maybe reduced due to the stacking of Pt before linking them with Na�on.icker active layers coupled to uniform dispersions of Pt clusters may also be the cause of the better polarization curve obtained with the three layers con�guration.e presence of MPL, CN layers, layer interfaces, and the low Pt loading may prevent signi�cant accumulation of li�uid water at the reaction sites and does not impose severe mass transport limitations, either by blocking the gas pores or by covering up the reactive surface area.e diffusion of reactant to the reaction sites is noticeably improved with this multilayered electrode con�guration.
Experiments were performed using the performance enhancements of Pt cluster formation shown in Figure 1 to compare the use of CN and CPTFE layers in between the Pt layers (Figures 2 and 3).e cathodes of EA1 and PA1, with a 34.8 wt.% of Na�on and 1�.� wt.% of PTFE, respectively, show best performances (Figures 2 and 3) than those of EA2 and PA2 with a CN ratio of 1 : 1 at the anodes (EA2 and PA2).
It is important to point out that the reduction of the mass of Pt in the active layer results in an increase of its porosity.is increase in porosity enhances mass transport by increasing the effective oxygen diffusion coefficient and increasing the area available for oxygen dissolution.us, the Pt clusters formed at the cathode reaction sites are well supplied in oxygen, resulting in a high power density and high limiting current density (>1.2 A/cm 2 ).e limiting current density achieved by EA1 is higher than that of PA1 due to the higher proton conductivity of Na�on.is ability promotes good proton conductivity in the cathode reaction sites leading to better performances.However, the comparison between the maximum power densities reached with MEA, EA, and PA series with standard MEAs, with a Pt loading of 0.2 mg/cm 2 , reveals that the maximum power reached by standard MEA is not too much different from that of EA and PA series, as shown in Figure 4.

Microstructure Characterization
Using SEM.SEM images of Figure 5 show the difference between the studied con�gurations of MEA1, MEA2, and MEA3.�e can F 5: SEM images of sputtered Pt on carbon paper: (a) MEA1 with one layer of 0.05 mg/cm 2 ; (c) MEA2 with two layers of 0.025 mg/cm 2 ; (e) MEA3 with two layers of 0.015 mg/cm 2 and a third layer of 0.02 mg/cm 2 with their magnitude (b), (d), and (f), respectively.
distinguish the formation of Pt clusters.e Pt clusters are signi�cantly larger in Figures 5(c) and 5(e) than those in Figure 5(a).Remember that the distributions showed in Figures 5(a), 5(c), and 5(e) are formed by sputtering multiple layers of Pt for a total loading of 0.05 mg/cm 2 (one layer of 0.05 mg/cm 2 , two layers of 0.025 mg/cm 2 each, and two layers of 0.015 mg/cm 2 each plus a third layer of 0.02 mg/cm 2 ).e gain in performance of these cluster con�gurations can be explained by the formation of larger Pt clusters with increased active surface area (as will be seen in Section 3.3).In Figure 5(c), the Pt is dispersed in the GDE that resulted from less cluster formation, resulting in isolated sites leading to a low active surface for the same amount of Pt.

Cyclic Voltammetry Measurement.
To get more insight into the experimental results for the different performances observed between one, two, and three catalyst layers, cyclic voltammetry (CV) was performed using glassy carbon working electrode.e same total amount of Pt (0.05 mg/cm 2 ) was sputtered on all glassy carbon electrodes for one, two and three layer con�gurations.e voltammograms are shown in Figures 6, 7, and 8 for one, two, and three layers of Pt. e electrochemical active surface (EAS) was calculated using the following equation: e analysis of the results shows that the three layers of Pt have a greater electrochemical active surface (EAS) that of one and two layers as presented in Table 2. ose results re�ect very well the di�erence in performance observed between MEA1, MEA2, and MEA3 (Figure 1).us, the e�ective areas for the ORR and HOR are in�uenced by the manner in which the Pt is distributed on the electrode.

Conclusion
e electrochemical characterization of MEA with one, two, and three layers of Pt (total loading of 0.05 mg/cm 2 ) at the cathode showed that the best polarization curve performance was achieved with three layers.High limiting current density could be achieved using Pt cluster formation (>1.1 A/cm 2 ) on CN and CPTFE layers between the Pt layers.Higher Pt utilization is responsible for this gain in performance due to the reduction of isolated sites.e mass transport has been enhanced due to the increase of limiting current density.e results also show that that the use of carbon-Na�on is better than the use of carbon-PTFE due to the decrease of ohmic loss, suggesting a better proton conductivity of Na�on than that of the PTFE solution.e maximum power reached by the use of CN or PTFE is similar to that obtained from the standard MEA.

F 1 :F 2 :
Comparison of the polarization and power curves for MEA1, MAE2, and MEA3.Comparison of polarization and power curves for EA1 and EA2.
T 1: Summary of all studies con�gurations.
T 2: EAS and hydrogen adsorption charge results for di�erent con�gurations of sputtered Pt.Cyclic voltammograms of three layers of sputtered Pt (2 × 0.015 mg/cm 2 + 0.02 mg/cm 2 ) on a glassy carbon working electrode of 3 mm in diameter in a solution of 1 M H2SO4 under nitrogen atmosphere.
where  ads (mC/cm 2 ) is the mean of charges exchanged during hydrogen adsorption and desorption,  ref takes the numerical value 0.21 mC/cm 2 and represents the charge required to oxidize a monolayer of hydrogen, and  Pt is the Pt loading (mg/cm 2 ).