Introduction

Proton exchange fuel cell (PEMFC) is considered as a very promising energy supply technology for automotive and stationary applications. The system is based on the oxidation of hydrogen at the anode forming protons and producing electrons and on the reduction of oxygen at the cathode in the presence of electron coming from the anode through the external electric circuit and protons crossing the solid membrane electrolyte. In such systems, the hydrogen oxidation reaction (HOR) is very rapid and therefore not limiting, whereas the oxygen reduction reaction (ORR) occurs with very slow kinetics [1, 2]. In order to overcome this issue, PEMFC cathodes are generally loaded with high amount of platinum-based catalysts. Different approaches were proposed to decrease the amount of precious metals in electrodes, such as synthesis of binary or ternary alloys [3,4,5,6], core-shell structures [7,8,9], shaped nanoparticles [10,11,12], and hollow nanoparticles [13, 14]. An alternative strategy consists in improving active surface sites by increasing the quality and quantity of three-phase boundaries in electrodes [15, 16]. A three-phase boundary (TPB) represents a spatial site were the ion conductive material, the electron conductive material, and the reactant are confined together on the catalytic site, allowing the catalytic reaction [17, 18]. Currently, the TPBs are created by mechanically mixing the catalytic powder, typically platinum nanoparticles supported on a carbon black powder (Pt-NPs/C) with an ionomer (typically Nafion) in a dispersant (typically a water/alcohol mixture). The catalytic ink is then dispensed either on the gas diffusion electrode or on the solid membrane electrolyte, and the Nafion is recast during drying. Although the implementation of this methods for the formation of TPBs reduced the Pt loading in electrodes [19, 20], it has been estimated that only 20 to 40% of the full potential activity was achieved [21,22,23] and that part of this low effectiveness was due to the low Pt utilization generally in the range of 40–60% [22].

To overcome this limitation, we have recently proposed to create the TPBs at the molecular level by grafting a proton conductive polymer directly on the Pt-NPs [24, 25]. Indeed, the possibility of creating iono-covalent bond between a metallic surface and a molecule bearing a thiol function or a disulfide molecule is very well documented [26,27,28,29]. In the present case, the symmetrical molecular structure of the disulfide polymers should lead to the grafting of two chains on one Pt-NP surface (Pt-S-polymer), as we showed in a previous work that the post-grafting of disulfide molecules did not occur on the carbon support, only on Pt surface [24].

Carbon-supported Pt-NPs modified by the polystyrene sulfonic acid (Pt-NPs-(PSSA)/C) displayed higher electrochemically active surface area and activity than the Pt-NPs/C catalyst in both three-electrode electrochemical cell and real fuel cell configurations [30]. In the present paper, we investigated the effects on the electrocatalytic activity and selectivity for the oxygen reduction reaction (ORR) of the nature and structure of the grafted polymers and of the functional groups introduced. For this purpose, different PSSA-based polymers were synthesized and grafted on Pt-NPs: polystyrene sulfonic acid (PSSA), poly(2,3,5,6-tetrafluorostyrene sulfonic acid) (PTFSSA), poly(3-(2,3,5,6-tetrafluoro-4-vinylthiophenol)propane-1-sulfonic acid) (PTFV-S-PSA), and poly(3-(2,3,5,6-tetrafluoro-4-vinylphenoxy)propane-1-sulfonic acid) (PTFV-O-PSA). The introduction of fluorostyrene groups is expected to change the hydrophilic/hydrophobic character of the organic crown, whereas the introduction of a spacer between the tetrafluorostyrene group and the sulfonic acid group is expected to affect the ionic conductivity. The electrocatalytic behavior of the different nanocomposite materials has been evaluated and compared in classical three-electrode cell as well as under real fuel cell working conditions.

Experimental

Synthesis of Pt-NPs/C

The carbon supported platinum nanoparticles were prepared using a microwave-assisted polyol method [30]. A first solution was prepared by dissolving 265.0 mg of hexahydrated hexachloroplatinic acid (H2PtCl6.6 H2O, 99.9% purity from Alfa Aesar) in 100 mL ethylene glycol (puriss. p.a., 99.5% Fluka). A second solution was prepared by dissolving 4.0 g of NaOH (reagent grade ≥ 98%, Sigma-Aldrich) in 100 mL ethylene glycol to reach a concentration of 1.0 mol L−1. The pH of the first solution was adjusted to 11 by adding dropwise the second solution. Vulcan XC 72 carbon powder (150 mg) thermally treated for 4 h under N2 atmosphere was then added, and the mixture was homogenized in an ultrasonic bath. The reaction balloon was put in a microwave oven (MARS 5 from CEM Corp.) for ca. 2 min under continuous microwave irradiation at 1600 W until reaching 100 °C; then, microwave pulses were applied for 5 min to maintain the reactor temperature at 100 °C. After the reduction reaction of Pt salt has occurred, the pH of the solution was decreased down to 2 by addition of HCl (ACS reagent, 37% fuming, Sigma-Aldrich). After dilution with 50 mL of ultrapure water (Millipore, Milli-Q, 18.2 MΩ cm), the mixture was homogenized for 5 min in an ultrasonic bath. The catalytic powder was filtered. The solid recovered was abundantly washed with ultrapure water and dried overnight at 60 °C. A heat treatment for 2 h at 200 °C in air was applied to the catalytic powder in order to remove remaining adsorbed organics from the synthesis.

Syntheses and Characterization of Disulfide Polymers

The syntheses and characterization of the precursors and of the different disulfide polymers used for the grafting of the Pt-NPs [31, 32] are explained in the supporting information (SI 1). Figure 1 shows the structure of each polymer.

Fig. 1
figure 1

Chemical structure of the different disulfide polymers synthesized for further grafting on Pt-NPs. a Polystyrene sulfonic acid (PSSA). b Poly{2,3,5,6-tetrafluorostyrene sulfonic acid} (PTFSSA). c Poly{3-(2,3,5,6-tetrafluoro-4-vinylthiophenol)propane-1-sulfonic acid} (PTFV-S-PSA). d Poly{3-(2,3,5,6-tetrafluoro-4-vinylphenoxy)propane-1-sulfonicaacid} (PTFV-O-PSA)

Full size image

Grafting of the Disulfide Polymers on the Carbon Supported Pt-NPs

For the grafting step, Pt-NP/C and hexylamine were mixed and stirred to obtain a stable suspension. A second solution of disulfide polymer in water/hexylamine (50/50 v/v) was prepared. The second solution was added dropwise to the first one under stirring, and the reaction was allowed to take place for 12–15 h and to rest without stirring for ca. 30 min. The amount of polymer was appropriate to obtain a grafting density of 2.82 × 10−7 molpolym. gPt−1 (The grafting density of polymers on Pt-NPs has been verified by elemental analysis [30]). The mixture was then filtered, and the modified Pt-NP-(polymer)/C powders were washed successively with acetone (3 × 30 mL), ethanol (3 × 30 mL), and Milli-Q water (3 × 30 mL) to remove hexylamine and polymer chains that did not graft on platinum nanoparticles. The modified catalytic powder was then dried overnight at 60 °C.

Physicochemical Characterizations

The metal and polymer loadings in the catalytic powders were determined by thermogravimetric analyses (TGA) using a DTA Instruments Q600 thermobalance. The samples were heated from 25 to 800 °C at 10 °C min−1 under an air flow of 100 mL min−1. The mean particle size and the particle size distribution were determined by transmission electron microscopy (TEM) using a JEOL JEM 2010 HR instrument with a resolution of 0.19 nm, based on the counting of 200 nanoparticles (using the ImageJ free software [33]).

Electrochemical Measurements

A home-made thermostated three-electrode electrochemical glass cell was equipped with a reversible hydrogen electrode (RHE) as reference and a glassy-carbon plate (3 cm2 geometric surface area) as counter electrode. The working electrode consisted in the dispensing of 3 μL of a catalytic ink (after homogenization) on a 0.0707 cm−2 geometric surface area glassy carbon disc in order to reach 100 μgPt cmgeom−2. The ink for the Pt-NPs/C reference catalyst consisted in 17.7 mg of the catalytic powder suspended in 2.646 mL of ultrapure water and 0.354 mL of Nafion solution (5 wt% Nafion perfluorinated resin solution in aliphatic alcohols). The inks for the Pt-NPs-(Polymer)/C catalysts were Nafion-free and consisted in appropriate amounts of catalytic powders (depending on the polymer molecular weight) to obtain 100 μgPt cmgeom−2 dispersed in 3 mL water. The electrochemical experiments were performed at 20 °C in an aqueous (ultrapure water) 0.1 mol L−1 HClO4 (Suprapur, Merck) electrolyte. Cyclic voltammograms (CV) were acquired in N2-saturated electrolyte (Air Liquide, U-Quality) using a Model 362 Scanning Potentiostat from Princeton Applied Research. ORR was studied by the rotating disc electrode (RDE) technique in an O2-saturated (Air Liquide, U-Quality) electrolyte; the low potential scan rate of 1.0 mV s−1 was applied from 1.05 to 0.30 V vs RHE in order to reach quasi-steady-state conditions necessary for determining accurate kinetics data. Rotating ring-disc electrode (Pine Instrument) was also used to determine the ratio of hydrogen peroxide formed over the whole potential range studied. For ORR studies, Voltalab PGZ402 potentiostats (Radiometer analytical) were used.

Fuel Cell Tests

Fuel cell cathodes of 5 cm2 geometric surfacearea were prepared by dispensing a Nafion-free catalytic ink containing the desired amount of Pt-NPs-(polymer)/C nanocomposite on a commercial gas diffusion layer (SIGRACET 24 BC) to reach 0.4 mgPt cm−2. The anodes and the reference cathode consisted in Pt-NP/C + 25 wt% Nafion with a Pt loading of 0.2 mg cm−2 and 0.4 mgPt cm−2, respectively. The reference membrane electrode assembly (MEA) was prepared by hot pressing of the Nafion-containing electrodes on both sides of a Nafion 211 membrane (25 μm thickness) at 115 °C and 3.5 MPa for 2.5 min. The MEA fitted with Pt-NP-(Polymer)/C cathodes were just mechanically pressed during the cell assembly (no hot pressing). Fuel cell measurements were performed using a BIOLOGIC FCT Tester and a PAXITECH 5 cm2 cell at T = Tautothermal (stable temperature reached by the cell under working conditions without external heating) and at T = 60 °C. Oxygen and hydrogen \( \left({\mathrm{P}}_{{\mathrm{O}}_2}={\mathrm{P}}_{{\mathrm{H}}_2}=2\kern0.5em {\mathrm{bars}}_{\mathrm{abs}}\kern0.5em \mathrm{and}\kern0.5em {\uplambda}_{{\mathrm{O}}_2}={\uplambda}_{{\mathrm{H}}_2}=1.5\right) \) were humidified at room temperature.

Results and Discussion

Synthesis of Nanocomposite Catalysts

The synthesis and characterization of PSSNa were previously discussed elsewhere [30] (SI 2). In the case of PTFSSNa and PTFV-S-PSNa, the starting material was the poly(2,3,4,5,6-pentafluorostyrene), the synthesis and the characterization of which are described in SI 3. The sulfuration of the poly(2,3,4,5,6-pentafluorostyrene) consisted first in introducing a sulfur atom in the para position of the aromatic cycle leading to the formation of thiolate species (SI 4). PTFSSNa was obtained by oxidation using hydrogen peroxide of the thiolate functions into SO3Na sulfonate functions (SI 5). PTFV-S-PSNa was obtained from the previous thiolate molecule prepared as explained above. In this case, the functionalization consisted in a nucleophilic substitution reaction between the nucleophilic thiolate function and the electrophilic site of the 1,3-propanesulfone cycle, involving the opening of the cycle and the formation of the pendent three-carbon chain bearing the SO3Na group (SI 6). In the case of PTFV-O-PSNa, the first step consisted in the introduction of hydroxyl functions in para position of the (2,3,4,5,6-pentafluorostyrene) molecules. The second step consisted in a nucleophilic substitution reaction between the nucleophilic hydroxyl function and the electrophilic site of the 1,3-propanesulfone cycle, involving the opening of the cycle and the formation of the pendent three-carbon chain bearing a SO3Na group (SI 7 and SI 8). The polymer characterization data are given in Table 1.

Table 1 Nominal (theor.) and actual (exp.) values of the degree of polymerization (DPn), conversion rate (conversion), molecular mass (Mn and Mw), and polymerization index (Ip) of the different polymers synthesized by ATRP
Full size table

The polymerization reactions involving the functionalized tetrafluorostyrene monomers led always to slightly lower conversion rates than that of styrenesulfonate monomers, which could be explained by the bigger molar weights (almost twice higher) of the polymers prepared with the functionalized tetrafluorostyrene monomers than that prepared from the styrene sulfonate monomers. Moreover, the polymolecularity indexes were higher for polymers prepared with the functionalized tetrafluorostyrene monomers than that prepared from the styrene sulfonate monomers, indicating wider chain length distributions in the former cases compared to PSSNa. However, the chain length distributions remained relatively narrow in all cases.

Ex Situ Physical Characterization of Nanocomposite Catalysts

After synthesis of the nanocomposite catalysts, the physical characterizations were carried out by TEM and TGA. Transmission electron images were taken to evaluate the morphology of the catalytic powders and the Pt-NPs mean size. Figure 2a shows a representative TEM image of the Pt-NPs/C catalytic powder prepared by the microwave-assisted polyol method. Pt nanoparticles (black dots of few nanometers) are well disseminated over the carbon support (gray particles of several tenths nanometer), although a few agglomeration is visible. The histogram in Fig. 2b represents the size distribution obtained from the counting of 200 isolated Pt nanoparticles. The size distributions are presented using a class width of 0.5 nm (i.e., particles with a size from 2.25 nm excluded to 2.75 nm included are attributed to the 2.5 nm class). A narrow distribution has been obtained, and a mean size of ca. 3.0 nm has been evaluated. Because Pt-NPs modified by PSSNa, PTFSSNa, PTFV-S-PSNa, and PTFV-O-PSNa derive from the same reproducible synthesis method, it is not expected that the mean size and size distribution were affected by the subsequent grafting of polymers. Figure 2c, e shows as examples representative TEM images of Pt-NPs-(PSSNa)/C and Pt-NPs-(PTFV-S-PSNa)/C catalytic powders, respectively. The histograms in Fig. 2d, f represent the size distribution obtained from the counting of 200 isolated particles for Pt-NPs-(PSSNa)/C and Pt-NPs-(PTFV-S-PSNa)/C catalysts, respectively. Same distributions as for Pt-NPs/C are obtained, which confirm the previous assumption (2.9 ± 0.9 nm for Pt-NPs/C, 3.0 ± 0.8 nm for Pt-NPs-(PSSNa)/C, and 2.9 ± 0.8 nm for Pt-NPs-(PTFV-S-PSNa)/C).

Fig. 2
figure 2

TEM images and corresponding particle size distribution obtained from the from the counting of 200 isolated Pt nanoparticles of a, b Pt-NPs/C, c, d Pt-NPs-(PSSNa)/C, and e, f Pt-NPs-(PTFV-S-PSNa)/C catalytic powder prepared by the microwave-assisted polyol method

Full size image

TGA measurements were performed to confirm the presence of the polymer crown around the nanoparticles. Figure 3a compares the thermograms (mass loss with respect to the initial mass in %) and the mass loss derivative with respect to the temperature T obtained as a function of the temperature for Pt-NP/C and Pt-NPs-(PSSNa)/C (Pt-NPs functionalized by the sodium polystyrene sulfonate). The black lines are typical of measurements on a dry Pt-NP/C material [34], with a single peak in the curve of the mass derivative vs T centered at ca. 385 °C corresponding to a single mass loss process (carbon support combustion). In the case of Pt-NPs-(PSSNa)/C (gray lines), the curve of the mass loss derivative vs T displays two peaks, the first one centered at ca. 257 °C and the second one at ca. 415 °C, corresponding to two mass loss processes. The peak centered at ca. 257 °C corresponds to the desulfonation of the polymer and that at ca. 415 °C is attributed to the combustion of carbonaceous species. For temperatures higher than 450 °C, Pt is the only remaining material, allowing a direct assessment of the metal loading. Figure 3b compares the thermograms and the curves of the mass loss derivative with respect to the temperature T obtained as a function of the temperature for fluorinated materials, Pt-NPs-(PTFSSNa)/C and Pt-NPs-(PTFV-S-PSNa)/C. Both materials start to decompose between 150 and 300 °C. In the case of the Pt-NPs-(PTFV-S-PSNa)/C, with a three-carbon pending chain, the decomposition corresponds to the degradation of the sodium sulfopropylate group, whereas in the case of Pt-NPs-(PTFSSNa)/C, the decomposition corresponds to the desulfonation of the phenyl group. The combustion of the remaining polymer and of the carbon support occurs for higher temperatures. Crossing these data with those from elemental analysis reveals a good reliability on polymers and platinum loadings as determined by TGA. The curves of the mass loss derivatives with respect to temperature obtained for the three samples (Fig. 4) indicate that the fluorinated polymers are thermally less stable than the PSSNa. The desulfonation occurs at lower temperature for Pt-NPs-(PTFSSNa)/C (the peak being centered at ca. 200 °C) than for Pt-NPs-(PTFV-O-PSNa)/C (ca. 225 °C) and for Pt-NPs-(PSSNa)/C (ca. 257 °C). This could be due to the attractive effect of fluorine which tends to weaken the R-phenyl (R=S or O) bond, as well as the bond between the phenyl group and the polymer chain. Table 2 gives the characterization data of the grafting of polymers on Pt-NPs. The grafting density was always around 2.8 × 10−7 molpolym. gPt−1, as expected, for all polymers. Such a grafting density corresponded to about 0.4 ft of polymer per platinum nanoparticles, meaning that in average, only one Pt-NP among five is decorated by polymer chains (considering that two polymer chains are grafted on one Pt NP, owing to the symmetrical disulfide structure of the initial polymers).

Fig. 3
figure 3

a Thermograms (mass loss with respect to the initial mass in %) and the mass loss derivative with respect to the temperature T obtained as a function of the temperature for non-fluorinated materials Pt-NP/C (black lines) and Pt-NPs-(PSSNa)/C gray lines) materials. b Thermograms and the curves of the mass loss derivative with respect to the temperature T obtained as a function of the temperature for fluorinated materials, Pt-NPs-(PTFSSNa)/C (black lines) and Pt-NPs-(PTFV-S-PSNa)/C (gray lines). TGAs were performed in air with a temperature variation rate of 10 °C min−1

Full size image
Fig. 4
figure 4

Mass loss derivatives with respect to temperature (in the temperature range from 150 to 300 °C) for Pt-NPs-(PSSNa)/C (black lines), Pt-NPs-(PTFSSNa)/C (dashed black line), and Pt-NPs-(PTFV-S-PSNa)/C (gray line). TGAs were performed in air with a temperature variation rate of 10 °C min−1

Full size image
Table 2 Experimental molecular mass Mnexp. of the different polymers synthesized by ATRP, Pt, and polymer loading in catalysts, grafting density, and number of polymer feet per platinum nanoparticle
Full size table

Electrochemical Characterization of Nanocomposite Catalysts

The electrochemical characterization was employed to assess the electrochemically active surface area (ECSA) and the resulting catalytic activity of the nanocomposite catalysts. Cyclic voltammetry was used for the determination and comparison of the ECSA of the different catalysts. Figure 5a compares the CV profiles of Pt-NPs-(Nafion)/C and Pt-NPs-(PSSA)/C, whereas Fig. 5b shows the CV recorded on the different Pt-NPs-(polymer)/C nanocomposites. From these CV, the ECSA was determined by integrating the hydrogen desorption region corrected from the double-layer capacitive currents. Values of ECSA are given in Table 2 for each catalyst. As observed in a previous work [30], the modification of Pt by PSSA with a nominal degree of polymerization (DPn) of 2500 and a grafting density (GD) of ca. 2.82 × 10−7 molPSSA gPt−1 is beneficial for the ECSA, which increases from 80 m2 gPt−1 for Pt-NPs-(Nafion)/C catalyst up to 98 m2 gPt−1 for Pt-NPs-(PSSA)/C. Because the Pt nanoparticles and the Pt-NPs/C catalyst originate from the same synthesis method, the change in ECSA is not due to difference in their microstructures and/or morphologies. The higher surface area indicates a higher platinum utilization rate. The beneficial effect can be due to lower Pt surface coverage and contamination when modified by PSSA (the calculated number of feet per Pt Np is very low, e.g., 0.4 ft per Pt NP) than when impregnated by recast Nafion, which also is known to contain impurities [35, 36]. The modification of Pt-NPs by polymers bearing tetrafluorophenyl group, with DPn and GD close to those for PSSA, leads to a decrease of the active surface area of the catalysts (ECSA values of 71, 68, and 66 m2 g−1, for Pt-NPs-(PTFV-S-PSA)/C, Pt-NPs-(PTFV-O-PSA)/C, and Pt-NPs-(PTFSSNa)/C, respectively). For the same reasons, as explained earlier, the decrease of the surface area indicates lower Pt utilization rate, the lower value of ECSA being reached with the Pt-NPs-(PTFSSNa)/C catalyst without three-carbon spacer between the tetrafluorophenyl group and the sulfonic acid function. It is worth to note that the catalysts modified by grafting of polymers display lower charges for the Pt surface oxide reduction peak at ca. 0.75 V vs RHE. This phenomenon appears more important in the case of fluorinated polymers than in the case of PSSA and again is more pronounced for the Pt-NPs-(PTFSSNa)/C catalyst without three-carbon spacer between the tetrafluorophenyl group and the sulfonic acid function. This behavior is explained in terms of limited water accessibility to the Pt surface. The limited accessibility of water to the Pt surface can be due either to the higher affinity of fluorinated polymers towards Pt and stacking (adsorption) of the fluorophenyl groups on the Pt surface or to a higher hydrophobic character of the fluorinated polymers limiting water transport from the bulk electrolyte to the Pt-NP surface. Indeed, the presence of fluorine could make the polymer more hydrophobic, and the insertion of a spacer between the tetrafluorophenyl group and the sulfonic acid function could enhance the transport of water (increase of the Pt surface oxide reduction charge) and of protons (increase of ECSA) from the bulk electrolyte towards the Pt surface. Because in aqueous media protons are solvated by several water molecules, the hydrophobic character of the grafted polymers may also explain the higher resistance observed in the hydrogen adsorption/desorption region of the CVs, which indicates some limitations to the diffusion of proton in the organic crown. This is an important result which demonstrates that the electrochemical behavior of a Pt catalyst can be tuned by controlling the hydrophobic/hydrophilic character of the proton conductive polymer in electrodes.

Fig. 5
figure 5

Cyclic voltammograms recorded at 25 °C in N2-saturated 0.1 M HClO4 solution at the scan rate s = 5 mV s−1 of a of Pt-NPs-(Nafion)/C (black line) and Pt-NPs-(PSSA)/C (dashed-dotted black line) and b Pt-NPs-(PTFSSA)/C (gray line), Pt-NPs-(PTFV-S-PSA) (dashed black line), and Pt-NPs-(PTFV-O-PSA)/C (dotted black line)

Full size image

The activity and selectivity of the catalysts towards oxygen reduction were evaluated using the rotating disc electrode (RDE) technique. The hydrodynamic polarization curves recorded on all catalysts for different electrode rotation rates are presented in the Supporting Information together with the 1/j vs 1/ω1/2 curves at different electrode potentials (Koutecky-Levich straight lines) calculated from these hydrodynamic polarization curves (SI 10). Each hydrodynamic polarization curve was performed under quasi-stationary conditions [37] with negative-going potential scan to avoid an overestimation of the catalytic activity due to the reduction of H2O2 formed at low-electrode potentials (see RRDE measurements in Table 3). From the Koutecky-Levich straight lines, the exchange current density, the limiting current density in the catalytic film (mass transport and adsorption), the Tafel slope, and the kinetic current density at 0.9 V could be estimated using a generalized Koutecky-Levich equation (Eq. 1) for all catalysts [38,39,40].

$$ \frac{1}{j}=\frac{1}{j_e}+\frac{1}{j_1^{\mathrm{Diff}}}=\frac{1}{j_{\mathrm{k}}}+\frac{1}{j_1^{\mathrm{Film}}}+\frac{1}{j_1^{\mathrm{Diff}}} $$
(1)
Table 3 Electrochemical surface areas, kinetic current density at 0.9 V vs RHE (jkgeom and jkcata are expressed with respect to the electrode geometric surface area and to the Pt real surface area, respectively), mass activity at 0.9 V, limiting current density with respect to the electrode geometric surface area, Tafel slope determined from RDE experiments, and maximum percentage of H2O2 produced at 0.3 V vs RHE determined from RRDE experiments. The expanded uncertainties are calculated from equations detailed in supporting information (SI 11) and given for a level of confidence of 99%
Full size table

where j is the current density at a given electrode potential E, je is the current density resulting from processes occurring at the electrode, and the limiting current density of oxygen diffusion in the electrolyte (jlDiff) is expressed using the Levich equation (SI 9).

The mathematical treatment of the generalized Koutecky-Levich equation is given in SI 9. Figure 6 presents the kinetics current (jk) as a function of electrode potential for all catalysts. Kinetics data are given in Table 3. The values obtained of 0.058 mA cmPt−2 for the surface activity at 0.9 V cs. RHE are of the same order than that obtained by other authors on commercial Pt-NPs/C (0.040 mA cmPt−2 at 0.8 V vs RHE) [41]. The same trend as for the ECSA is observed for the electrocatalytic activity towards the ORR in terms of jk (kinetic current density at 0.9 V expressed with respect to the geometric surface area of the electrode), jkcata (kinetic current density at 0.9 V expressed with respect to the real surface area of platinum), and MA (specific current density with respect to the mass of platinum), with Tafel slope values ranging from 65 to 74 mV dec−1. The presence of a grafted polymer on Pt-NPs leads to a decrease of jlFilm, which could be due to limitations of oxygen and/or proton diffusion in the catalytic film towards the Pt sites or to oxygen adsorption on modified nanoparticles. It is also interesting to note that Pt-NPs-(PTFV-S-PSA)/C and Pt-NPs-(PTFV-O-PSA)/C catalysts lead to very close electrocatalytic performances, which indicate that the heteroatoms linking the spacer to the phenyl group (O or S) have no significant effect on the electrochemical behavior. The lower performances in terms of activity are achieved with the Pt-NPs-(PTFSSA)/C catalyst for which the sulfonic acid function is directly bonded to the tetrafluorophenyl group, without spacer. Considering the lower ECSA for Pt-NPs-(PTFSSA)/C and the quasi-invariance of jlFilm for all fluorinated nanocomposites, the important loss in activity of Pt-NPs-(PTFSSA)/C can be explained by the higher hydrophobic character of the grafted polymer, which limits the diffusion of proton rather than that of oxygen.

Fig. 6
figure 6

Kinetic current (jk) vs E curves for Pt-NPs-(Nafion)/C (square), Pt-NPs-(PSSA)/C (circle), Pt-NPs-(PTFSSA)/C (diamond), Pt-NPs-(PTFV-S-PSA)/C (inverted triangle), and Pt-NPs-(PTFV-O-PSA)/C (triangle)

Full size image

Rotating ring-disc electrode (RRDE) was used to evaluate the production of hydrogen peroxide for the different materials. The maximum ratios of H2O2 formed at 0.3 V vs RHE are given in Table 3. For all catalysts, no H2O2 oxidation current could be detected at the ring electrode for potentials higher than 0.75 V vs RHE. The formation of 5% H2O2 at 0.3 V vs RHE at a Pt-NPs-(Nafion)/C electrode is consistent with previous results. It is worth to note that this ratio is the lowest for Pt-NPs-(PSSA)/C material. It slightly increases for Pt-NPs-(PTFV-S-PSA)/C and Pt-NPs-(PTFV-O-PSA)/C nanocomposites with a spacer between the sulfonic acid function and the tetrafluorophenyl group and reaches its highest value for Pt-NPs-(PTFSSA)/C material without spacer. This behavior confirms the previous explanation invoking the hydrophobic character of the PTFSSA polymer limiting the proton accessibility towards the active sites.

Fuel Cell Measurements

At last, the performances of promising catalysts were evaluated and compared under real fuel cell working conditions. Fuel cell measurements were performed with the different MEAs fitted with 0.4 mgPt cm−2 cathodes and 0.2 mgPt cm−2 anodes. In the case of Pt-NPs-(PTFSSA)/C material, it was not possible to prepare homogeneous catalytic ink. Therefore, the electrode manufactured from this ink showed huge agglomerates of platinum (as evidenced by EDS microanalysis), with size reaching 100 μm (Fig. 7a). The inhomogeneity of the cathode and the high hydrophobicity of the catalytic material have led to very poor fuel cell performances with a maximum current density of 0.1 A cm−2 and fluctuant cell voltage. In contrary, the preparation of cathodes from Pt-NPs-(PTFV-S-PSA)/C and Pt-NPs-(PTFV-O-PSA)/C nanocomposites with a spacer between the sulfonic acid function and the tetrafluorophenyl group was very easy. Figure 7b shows a typical scanning electron microscopy image obtained on such electrodes. Very homogeneous catalytic layers were obtained with a good porosity. In addition, very interesting fuel cell performances could be obtained with these cathodes.

Fig. 7
figure 7

SEM images of electrodes prepared from a a Pt-NPs-(PTFSSA)/C catalytic ink and b a Pt-NPs-(PTFV-S-PSA)/C catalytic ink

Full size image

The polarization Ucell (j), where U is the cell voltage and j the current density, and power density P(j) curves were recorded under fuel cell working conditions at 60 °C (Fig. 8) for membrane electrodes assemblies (MEAs) fitted with a classical Pt-NPS-(Nafion)/C cathode as reference (solid line), a Pt-NPs-(PSSA)/C cathode (dash-dotted line), and a Pt-NPs-(PTFV-O-PSA)/C cathode (dotted line). Note that the MEA with Pt-NPS-(Nafion)/C cathode was manufactured by hot pressing of the electrodes against the Nafion 211 membrane, whereas those fabricated with Pt-NPs-(PSSA)/C and Pt-NPs-(PTFV-O-PSA)/C cathode did not contain any Nafion and were only mechanically pressed against the Nafion membrane during the assembling of the single fuel cell. For clarity reasons, the polarization and power density curves for the MEA fitted with a Pt-NPs-(PTFV-S-PSA)/C cathode are not reported because same results were obtained as for the MEA fitted with a Pt-NPs-(PTFV-O-PSA)/C cathode. As expected from cyclic voltammetry and RDE results, the atom separating the spacer from the polymer skeleton has no effect on the performance of the MEAs under fuel cell working condition. Lower performances were obtained at high current densities with the MEA fitted with a Pt-NPs-(PTFV-O-PSA)/C cathode. These observations could be explained in terms of lower electroactive surface area, lower specific activity, and higher limitation of reactant diffusion in the catalytic film of the nanocomposite catalysts modified with PTFV-O-PSA or PTFV-S-PSA, as determined from cyclic voltammetry and RDE measurements. But the three assemblies led to similar performances for current densities lower than 1.0 A cm−2, whereas the slope of the polarization curves in the 0.5 to 1.1 A cm−2 is higher for the MEA fitted with a Pt-NPs-(PTFV-O-PSA)/C cathode than for the two other MEAs, indicating higher cell resistance in the former case. On the other hand, in the polarization curves obtained with the MEAs fitted with Pt-NPS-(Nafion)/C and Pt-NPs-(PSSA)/C cathodes, an inflection point is visible at ca. 1.1 A cm−2 leading to more important slopes. This observation translates in the mass transport limitation and/or flooding of the electrode [42, 43]. In the case of the MEA fitted with a Pt-NPs-(PTFV-O-PSA)/C cathode, no inflection point is visible on the polarization curve. Because according to the RDE measurements the limitation of reactant diffusion in the Pt-NPs-(PTFV-O-PSA)/C catalytic film is higher than those in the case of Pt-NPs-(Nafion)/C and Pt-NPs-(PSSA)/C catalytic films, this translate to a better water management in the Pt-NPs-(PTFV-O-PSA)/C cathode.

Fig. 8
figure 8

Polarization and power density curves recorded in a 5-cm2 geometric surface area fuel cell at 60 °C for membrane electrodes assemblies (MEAs) fitted with a classical Pt-NPS-(Nafion)/C cathode as reference (black line), a Pt-NPs-(PSSA)/C cathode (dashed-dotted line), and a Pt-NPs-(PTFV-O-PSA)/C cathode (dotted black line)

Full size image

Durability tests were performed for 250 h at two different temperatures, first at 38 °C and second at 60 °C, and 1.0 A cm−2. Figure 9 shows the variation of the cell voltages as a function of time. At 38 °C (Fig. 9a), the MEA with a Pt-NPs-(PTFV-O-PSA)/C cathode led to a higher cell voltage than those obtained with MEA fitted with Pt-NPs-(Nafion)/C and Pt-NPs-(PSSA)/C cathodes. The voltage loss of the MEA with a Pt-NPs-(PTFV-O-PSA)/C cathode was estimated to be 16 μV h−1, i.e., twice and four times lower than those determined for MEAs with Pt-NPs-(PSSA)/C and Pt-NPs-(Nafion)/C cathodes, respectively. At 60 °C (Fig. 9b), the MEA with a Pt-NPs-(PTFV-O-PSA)/C cathode led to a lower cell voltage than those obtained with MEA fitted with Pt-NPs-(Nafion)/C and Pt-NPs-(PSSA)/C cathodes. This can be due to the increase of the resistance in the MEA fitted with a Pt-NPs-(PTFV-O-PSA)/C cathode which has shown better ability for the water removal from the active layer, leading to the drying of the membrane at high temperatures. The voltage loss at 60 °C is slightly higher than those at 38 °C. In this case, the MEA with a Pt-NPs-(PSSA)/C cathode led to lower cell voltage loss (ca. 50 μV h−1 against ca. 100 μV h−1 for the MEA with a Pt-NPs-(Nafion)/C cathode). The MEA with a Pt-NPs-(PTFV-O-PSA)/C cathode led to slightly higher cell voltage loss (ca. 60 μV h−1) than the MEA with a Pt-NPs-(PSSA)/C cathode, which may be explained by the higher hydrophobicity and water repulsion capacity of the fluorinated polymer than those of the PSSA. As a consequence, at 60 °C, the ionomer PTFV-O-PSA and the Nafion membrane becomes less and less humidified than at 38 °C. However, these experiments show the beneficial role of grafted PSSA and PTFV-O-PSA on the stability of cathode and further on long-term fuel cell performances.

Fig. 9
figure 9

Fuel cell stability at 1.0 A cm−2 recorded in a 5-cm2 geometric surface area fuel cell for membrane electrodes assemblies (MEAs) fitted with a classical Pt-NPS-(Nafion)/C cathode as reference (black line), a Pt-NPs-(PSSA)/C cathode (dashed-dotted black line), and a Pt-NPs-(PTFV-O-PSA)/C cathode (dotted black line) at a 38 °C (autothermal temperature) and b 60 °C

Full size image

Conclusion

Supported Pt-NPs/C were grafted by different non-fluorinated and fluorinated proton conducting polymers. In the case of fluorinated polymers, the sulfonyl functions, which allow proton conduction, were attached either directly or through spacers (-O-PSA and -S-PSA) to the tetrafluorovinylic groups. It was shown that the nature and structure of the grafted proton conducting polymer influenced oxygen and proton transport towards the oxygen reduction active sites and that this influence was directly linked to the hydrophobic character of the polymers. Particularly, the importance of the spacer was evidenced: both the Pt-NPs-(PTFV-O-PSA)/C and Pt-NPs-(PTFV-S-PSA)/C catalysts led to the same electrocatalytic behavior as determined in classical three-electrode setup by cyclic voltammetry and RDE measurements, higher than those obtained with Pt-NPs-(PTFSSA)/C in terms of active surface area, kinetic current density, and mass activity at 0.9 V, as well as in terms of limiting diffusion current density in the catalytic films. Fuel cell tests have also evidenced the influence of the grafted polymer on the water management in cathodes. Durability tests under fuel cell working conditions have also revealed very important results: the presence of the PSSA and PTFV-O-PSA conducting polymers in the cathode catalytic layer led to comparable electrical performances but to better stabilities of the fuel cell performances than in the case of a classical Pt-NPs-(25 wt% Nafion)/C cathode, the potential loss being at 38 °C twice and four times lower with a Pt-NPs-(PTFV-O-PSA)/C cathode than with a Pt-NPs-(PSSA)/C and a Pt-NPs-(Nafion)/C cathode, respectively. At 60 °C, the potential loss with a Pt-NPs-(PTFV-O-PSA)/C cathode remains twice lower than with a Pt-NPs-(Nafion)/C cathode.

The present study opens a new research field for the synthesis and grafting of new class of polymer with variable hydrophobic characters and further mass transport properties. For example, the influence of the spacer length between the fluorinated polymer skeleton and the sulfonyl group should be important; the copolymerization of fluorinated and non-fluorinated monomers and the ratio between both monomers could allow controlling the hydrophobic character of the polymer, etc. The optimal degree of polymerization (DPn) and grafting density for achieving the best fuel cell performances as possible likely depend on the polymer nature and structure, and the determination of the best configuration for Pt-NPs-(PSVF-O(S)-PSA)/C should be realized as it was previously done for Pt-NPs-(PSSA)/C. Moreover, only pure oxygen was used as an oxidant under fuel cell operations; the use of air and the comparison of results could bring additional evidences on the effect of materials on gas diffusivity. These works are currently under progress.