Reduced Graphene Oxide-Supported Pt-Based Catalysts for PEM Fuel Cells with Enhanced Activity and Stability

: Platinum (Pt)-based electrocatalysts supported by reduced graphene oxide (RGO) were synthesized using two different methods, namely: (i) a conventional two-step polyol process using RGO as the substrate, and (ii) a modiﬁed polyol process implicating the simultaneous reduction of a Pt nanoparticle precursor and graphene oxide (GO). The structure, morphology, and electrochemical performances of the obtained Pt/RGO catalysts were studied and compared with a reference Pt/carbon black Vulcan XC-72 (C) sample. It was shown that the Pt/RGO obtained by the optimized simultaneous reduction process had higher Pt utilization and electrochemically active surface area (EASA) values, and a better performance stability. The use of this catalyst at the cathode of a proton exchange membrane fuel cell (PEMFC) led to an increase in its maximum power density of up to 17%, and signiﬁcantly enhanced its performance especially at high current densities. It is possible to conclude that the optimized synthesis procedure allows for a more uniform distribution of the Pt nanoparticles and ensures better binding of the particles to the surface of the support. The advantages of Pt/RGO synthesized in this way over conventional Pt/C are the high electrical conductivity and speciﬁc surface area provided by RGO, as well as a reduction in the percolation limit of the components of the electrocatalytic layer due to the high aspect ratio of RGO.


Introduction
Platinum (Pt)-based electrocatalysts are commonly used in proton exchange membrane fuel cells (PEMFCs). The platinum nanoparticles must be dispersed uniformly on the surface of a suitable support (catalyst carrier) in order to maximize the active surface and the catalytic activity, to improve the stability, and to reduce the cost of electrodes [1]. Carbon blacks (Vulcan XC-72, Ketjen Black, etc.), which have the merit of being easily precursors. The obtained Pt/RGO(s) catalysts were characterized using different techniques, including X-ray diffraction (XRD) and X-ray photoemission spectroscopy (XPS), transmission electron microscopy (TEM), cyclic voltammetry and accelerated stress testing (AST) in comparison with the Pt/carbon black Vulcan XC-72 (C) and Pt/RGO catalysts obtained conventionally. Finally, the obtained catalysts were implemented at the cathodes of the H2/O2 PEMFC MEAs and tested.

Results and Discussion
The structure and morphology of the Pt/C, Pt/RGO, and Pt/RGO(s) catalysts were determined through TEM analysis. Figure 1 shows that instead of some agglomeration, the Pt nanoparticles of all of the studied catalysts were quite small and finely dispersed over the surface of the carbon substrate. Their size distributions (see insets of Figure 1) were obtained using the well-known ImageJ software [46]. The analysis was based on no less than 100 particles for each catalyst. The size of most Pt nanoparticles (ca. 80%) was between 3 and 5 nm. The mean particle sizes of the Pt/C, Pt/RGO, and Pt/RGO(s) catalysts were 3.6, 3.9, and 3.95 nm, respectively. Particle agglomeration made it difficult to precisely evaluate the nanoparticle size, thus XRD analysis was used to find the Pt crystallite size (see below). The reduction of GO led to significant structure changes. The XRD spectra of exfoliated GO usually shows an intensive (001) peak with an interplane distance of ca. 7.4 Å [47], which is higher than graphite interplane distance (3.37 Å), and because of the various functional groups formed and C-C bonds disruption during the oxidation [32] (see XPS spectra given below). The XRD spectra of RGO ( Figure 2) exhibited a main peak at 2θ = 30.5°, corresponding to the (002) plane with an interplane distance of ca. 3.4 Å, suggesting the partial restoration of the graphitic structure.
.  The reduction of GO led to significant structure changes. The XRD spectra of exfoliated GO usually shows an intensive (001) peak with an interplane distance of ca. 7.4 Å [47], which is higher than graphite interplane distance (3.37 Å), and because of the various functional groups formed and C-C bonds disruption during the oxidation [32] (see XPS spectra given below). The XRD spectra of RGO ( Figure 2) exhibited a main peak at 2θ = 30.5 • , corresponding to the (002) plane with an interplane distance of ca. 3.4 Å, suggesting the partial restoration of the graphitic structure. The reduction of GO led to significant structure changes. The XRD spectra of exfoliated GO usually shows an intensive (001) peak with an interplane distance of ca. 7.4 Å [47], which is higher than graphite interplane distance (3.37 Å), and because of the various functional groups formed and C-C bonds disruption during the oxidation [32] (see XPS spectra given below). The XRD spectra of RGO ( Figure 2) exhibited a main peak at 2θ = 30.5°, corresponding to the (002) plane with an interplane distance of ca. 3.4 Å, suggesting the partial restoration of the graphitic structure.
.   4-802, considering the Co Kα source being used here). The spectra of Pt/RGO(s) (Figure 2) shows the (002) peak of carbon at 30.7 • considering the multilayered RGO support [48,49]. The Scherrer equation was used to calculate the mean size of the Pt crystallites, as follows: where d XRD is the mean size of the Pt crystallites (in nm), K is the Scherrer constant (for spherical particles with cubic symmetry this is equal to 0.94), λ is the Co Kα radiation wavelength (0.178897 nm), β is the full peak width at half maximum (rad), and θ β is the Bragg angle (degrees). The mean sizes of the Pt crystallites of Pt/C, Pt/RGO, and Pt/RGO(s) were ca. 3.60, 3.55, and 3.50 nm, respectively. It should be mentioned that the mean size of the RGOsupported Pt nanoparticles is slightly higher in comparison with the Vulcan XC-72 supported ones, possibly because of some agglomeration of RGO during its reduction (thermal or chemical) [39]. The simultaneous reduction of precursors did not significantly affect the mean Pt crystallite size (measured by XRD) or the mean nanoparticles size (d TEM ) deduced from the analysis of the TEM images (see combined Pt crystallites/nanoparticles sizes data in Table 1). CasaXPS software [50] was used to analyze the experimental data and decompose the lines into constituent components. The Gaussian-Lorentzian sum functions were used to interpret the spectra of the individual lines. The National Institute of Standards and Technology (NIST) laboratory database was used to determine the positions and shapes of the lines [51]. According to the C 1s spectra (Figure 3a-c), RGO-supported catalysts primarily consist of sp 2 -carbon (at ca. 284.5 eV). The sp 2 /sp 3 ratios of Pt/C, Pt/RGO, and Pt/RGO(s) were 1.27, 36.2, and 38.6, respectively, suggesting that the graphitic structure of RGO-supported catalysts was successfully restored during GO reduction using both approaches-thermal reduction and simultaneous polyol reduction. Moreover, their C/O ratio values were close, and were equal to ca. 9.5. At a low binding energy, the spectra showed a few peak duplets (Figure 3d-f) corresponding to Pt at different oxidation states. To distinguish the metallic Pt(0), Pt(II) (PtO/Pt(OH) 2 ), and Pt (IV) (PtO 2 ), the obtained spectra were deconvoluted into a series of Pt 4f 7/2 and 4f 5/2 doublets, with a fixed ratio for the intensity and energy shift [52]. The obtained spectra suggest that Pt(0) is the predominant oxidation state in all catalysts (ca. 47-55%). On the other hand, all of the samples presented a variable amount of Pt(II) and Pt(IV) species deduced from the Pt 4f 7/2 peaks at binding energy values of ca. 72-75 eV, which could be attributed to the partial surface oxidation of small Pt metal nanoparticles [53]. The Pt(0) content of Pt/C, Pt/RGO, and Pt/RGO(s) were ca. 46.7, 43.2, and 55.2%, respectively. The lower amount of oxidized Pt in the case of Pt/RGO(s) could be explained by the particle size effect (smaller nanoparticles have more surface atoms, which are prone to oxidize) [53] or by the residual Pt(II) and Pt(IV) species from the polyol reduction process, in which Pt (IV) reduced to Pt (II) and then reached the metallic form [54]. It is worth noting that the Pt(0) surface provided an active site for electrochemical reactions, but not the Pt(II) and Pt(IV) species.   The accessibility of the Pt nanoparticles and the catalytic layer connectivity could be analyzed using the Pt utilization value. One hundred percent Pt utilization means that all Pt nanoparticles were integrated into the entire electron conductivity network, and were   The accessibility of the Pt nanoparticles and the catalytic layer connectivity could be analyzed using the Pt utilization value. One hundred percent Pt utilization means that all Pt nanoparticles were integrated into the entire electron conductivity network, and were  The accessibility of the Pt nanoparticles and the catalytic layer connectivity could be analyzed using the Pt utilization value. One hundred percent Pt utilization means that all Pt nanoparticles were integrated into the entire electron conductivity network, and were fully accessible by the electrolyte. The approximate value of the full surface area of the Pt nanoparticles can be calculated according to the following equation [57,58]: The accessibility of the Pt nanoparticles and the catalytic layer connectivity could be analyzed using the Pt utilization value. One hundred percent Pt utilization means that all Pt nanoparticles were integrated into the entire electron conductivity network, and were fully accessible by the electrolyte. The approximate value of the full surface area of the Pt nanoparticles can be calculated according to the following equation [57,58]: where S Pt is the total surface of Pt (m 2 g −1 Pt), ρ is the bulk Pt density (21.4 g cm −3 ), and d is the Pt nanoparticles mean size (nm; taken as the mean value between d XRD and d TEM ). Thus, the Pt utilization (u Pt , %) can be calculated as an EASA divided by the total surface area of Pt, as follows: The values of d XRD , d TEM , EASA, S Pt , and u Pt are collected in Table 1.
The obtained EASA values were lower than the values of the total area of the Pt calculated from Equation (2), because of the different use of Pt, which was generally less than 100%, even when the catalytic films that were used were thin. For instance, the u Pt value of a commercial E-TEK catalyst (with 20 wt.% of Pt) was 68% [59]. When graphenebased materials were used as a substrate, the Pt utilization of a catalyst strongly depended on the synthesis procedure, type of precursors, etc. [30,32,35]. According to the results of Sanli et al. [32], the u Pt of various RGO-supported catalysts prepared using different approaches could vary from 28 to 93%. Here, ethylene glycol (EG), simultaneously acting as a reductant, stabilizer, and dispersing agent [32], could form small and uniformly distributed Pt nanoparticles. Moreover, the use of RGO as a support, together with the simultaneous reduction approach, allowed for decreasing the agglomeration of particles and increasing the Pt utilization in Pt 40 /RGO(s). On the other hand, the stability of the obtained catalysts is another important factor that should be also considered. Figure 5 shows the variation of the EASA of the catalysts during AST.
= 6000 (2) where SPt is the total surface of Pt (m 2 g −1 Pt), ρ is the bulk Pt density (21.4 g cm −3 ), and d is the Pt nanoparticles mean size (nm; taken as the mean value between dXRD and dTEM). Thus, the Pt utilization (uPt, %) can be calculated as an EASA divided by the total surface area of Pt, as follows: The values of dXRD, dTEM, EASA, SPt, and uPt are collected in Table 1. The obtained EASA values were lower than the values of the total area of the Pt calculated from Equation (2), because of the different use of Pt, which was generally less than 100%, even when the catalytic films that were used were thin. For instance, the uPt value of a commercial E-TEK catalyst (with 20 wt.% of Pt) was 68% [59]. When graphene-based materials were used as a substrate, the Pt utilization of a catalyst strongly depended on the synthesis procedure, type of precursors, etc. [30,32,35]. According to the results of Sanli et al. [32], the uPt of various RGO-supported catalysts prepared using different approaches could vary from 28 to 93%. Here, ethylene glycol (EG), simultaneously acting as a reductant, stabilizer, and dispersing agent [32], could form small and uniformly distributed Pt nanoparticles. Moreover, the use of RGO as a support, together with the simultaneous reduction approach, allowed for decreasing the agglomeration of particles and increasing the Pt utilization in Pt 40 /RGO(s). On the other hand, the stability of the obtained catalysts is another important factor that should be also considered. Figure 5 shows the variation of the EASA of the catalysts during AST.  The EASA of all of the catalysts studied decreased significantly during AST. The main degradation mechanisms included (i) electrochemical Ostwald ripening, (ii) particle migration and coalescence, (iii) detachment from carbon support (mainly from carbon corrosion), and (iv) platinum dissolution and re-precipitation inside the ionomer phase [60][61][62][63][64][65][66]. As the initial particle size of the Pt nanoparticles of the different catalysts studied was similar, their different stability could be mainly attributed to the type of carbon support used. Moreover, corrosion of the carbon support was expected to be the prevailing degradation mechanism because of the potential range selected for the AST procedure [13]. After 3000 AST cycles, the EASA residual values of Pt/C, Pt/RGO, and Pt/RGO(s) were 18.4, 23.8, and 32.1 m 2 g −1 Pt, or 34, 49.5, and 53.9%, respectively. It is important to note that the electronic interactions between the Pt nanoparticles and graphene support could be responsible for the higher durability observed in the RGO-supported catalysts, as discussed in the literature [67]. Such enhanced metal-substrate interactions are evidenced by the negative shift in the Pt 4f binding energy on the XPS spectra [68], suggesting that electrons were transferred from the respective carbon supports to Pt. However, both RGO-supported catalysts showed a higher durability in comparison with Pt/C. A possible explanation is the hindered corrosion of the RGO support itself as a result of the smaller number of disordered sites in comparison with carbon black. Such disordered domains (sp 3 hybridized) are prone to be oxidized first at a high rate compared with the ordered sp 2 graphitic structure of RGO [69,70].
After characterization, the different catalysts (Pt/C, Pt/RGO, and Pt/RGO(s)) were studied as cathode catalysts in H 2 /O 2 PEMFC. The voltage and power density vs. current density are shown in Figure 6. was similar, their different stability could be mainly attributed to the type of carbon support used. Moreover, corrosion of the carbon support was expected to be the prevailing degradation mechanism because of the potential range selected for the AST procedure [13]. After 3000 AST cycles, the EASA residual values of Pt/C, Pt/RGO, and Pt/RGO(s) were 18.4, 23.8, and 32.1 m 2 g −1 Pt, or 34, 49.5, and 53.9%, respectively. It is important to note that the electronic interactions between the Pt nanoparticles and graphene support could be responsible for the higher durability observed in the RGO-supported catalysts, as discussed in the literature [67]. Such enhanced metal-substrate interactions are evidenced by the negative shift in the Pt 4f binding energy on the XPS spectra [68], suggesting that electrons were transferred from the respective carbon supports to Pt. However, both RGO-supported catalysts showed a higher durability in comparison with Pt/C. A possible explanation is the hindered corrosion of the RGO support itself as a result of the smaller number of disordered sites in comparison with carbon black. Such disordered domains (sp 3 hybridized) are prone to be oxidized first at a high rate compared with the ordered sp 2 graphitic structure of RGO [69,70].
After characterization, the different catalysts (Pt/C, Pt/RGO, and Pt/RGO(s)) were studied as cathode catalysts in H2/O2 PEMFC. The voltage and power density vs. current density are shown in Figure 6. According to Figure 6, the cell voltages measured at 0.5 A cm −2 on the PEMFC with Pt/C, Pt/RGO, and Pt/RGO(s) were ca. 0.593, 0.558, and 0.628 V, respectively. The use of RGO as a catalyst support, regardless of the technique used for the deposition of Pt nanoparticles, provided a better MEA performance at high current densities >1.2 A cm −2 because of the better reagents/products mass transfer in the cathode catalyst layer [71]. Significant mass transfer limitations were observed at high current densities with the Pt/Cbased MEA (Figure 6), while the polarization curves of Pt/RGO and Pt/RGO(s) remained According to Figure 6, the cell voltages measured at 0.5 A cm −2 on the PEMFC with Pt/C, Pt/RGO, and Pt/RGO(s) were ca. 0.593, 0.558, and 0.628 V, respectively. The use of RGO as a catalyst support, regardless of the technique used for the deposition of Pt nanoparticles, provided a better MEA performance at high current densities >1.2 A cm −2 because of the better reagents/products mass transfer in the cathode catalyst layer [71]. Significant mass transfer limitations were observed at high current densities with the Pt/Cbased MEA (Figure 6), while the polarization curves of Pt/RGO and Pt/RGO(s) remained linear up to 1.4 A cm −2 . The worse performance of Pt/RGO-based MEA could be ascribed to the lower catalyst EASA and Pt utilization discussed earlier, due to the agglomeration of RGO (during the catalyst synthesis and the catalyst deposition over the gas diffusion electrode (GDE) [21]); however, the open layered structure of RGO allowed for sustaining high current densities without significant transport limitations [12,72]. This suggestion is in good agreement with the results described in the literature [30], according to which the wrinkled sheet-like structure of RGO is superior to carbon black in terms of providing gas flow channels and a high accessibility for the reactant gases. The high aspect ratio of RGO allowed for reducing the percolation limits [72] of the electrocatalytic layer's components responsible for mass, electron, and proton transport.
The maximum power densities measured on PEMFC MEAs containing Pt/C, Pt/RGO, and Pt/RGO(s) were ca. 0.358, 0.337, and 0.421 W cm −2 , respectively. The simultaneous reduction of Pt and RGO precursors during synthesis provided a significantly better PEMFC MEA performance. This was mainly due to the higher catalyst EASA, higher Pt utilization, and less agglomerated support structure, which provided effective mass transport of the oxygen reduction reaction reagents and products. The sorption of H 2 PtCl 6 by the GO surface and its further simultaneous reduction were possibly responsible for the narrow and uniform Pt nanoparticles distribution over the RGO surface. The Pt nanoparticles nucleated during the synthesis might have enhanced the dispersibility of the RGO and prevent restacking, thus providing a high EASA and Pt utilization.

Preparation of Catalysts
Preparation of the Pt/C (C-carbon black Vulcan XC-72, purchased from Cabot Corporation, Boston, MA, USA) and Pt/RGO electrocatalysts (40 wt.% of Pt) was carried out using a modified polyol synthesis procedure, in which the reduction of the Pt precursor (H 2 PtCl 6 2H 2 O) was performed using ethylene glycol and formaldehyde as reductants, according to the literature [73]. The RGO support were obtained through the thermal reduction of GO obtained by modified Hummers' method, as described in the literature [9,10]. Pt 40 /RGO(s) were obtained by the same procedure, but with the addition of GO instead of carbon support, i.e., GO and Pt nanoparticles precursor were reduced "in one pot" in order to obtain the Pt/RGO(s) catalyst.

Structural and Morphological Studies
The XRD studies were performed using a D8 Advance diffractometer (Bruker, Billerica, MA, USA), using Co Kα radiation at 2θ = 20-120 • with a step of ∆2θ = 0.07 • . A scanning/transmission electron microscope, Titan 80-300 (Thermo-Fisher Scientific, Beverly, MA, USA), equipped with an energy dispersive X-ray spectrometer (EDAX, Mahwah, NJ, USA) was used to study the morphology of the catalysts. The XPS studies were performed using a PHOIBOS 150 (SPECS Surface Nano Analysis GmbH, Berlin, Germany) hemispherical analyzer and monochromatic Al Kα radiation, with a photon energy of 1486.61 eV and an end resolution of ∆E = 0.2 eV, located at the facility of the Kurchatov Complex of Synchrotron and Neutron Investigation (NRC "Kurchatov Institute", Moscow, Russian Federation). The powdered samples were pressed into carbon scotch tape and were transferred to the vacuum chamber of a spectrometer with a base pressure of 3 × 10 −9 mbar. All of the spectra were measured in fixed analyzer transmission mode, with a pass energy of 120 and 40 eV for the survey spectra and separated lines, respectively. To analyze the experimental data and decompose the lines into constituent components, CasaXPS software (v. 2.3.23, Casa Software Ltd., Teignmouth, UK) was used [50]. The NIST laboratory database (v. 4.1, National Institute of Standards and Technology, Gaithersburg, MD, USA) was used to determine the position of the lines [52].

Electrochemical Studies
Electrochemical studies were carried out using a Solartron 1280 (AMETEK, Inc., Berwyn, PA, USA), with thin catalytic films formed on the surface of the glassy carbon disk working electrodes, according to the previously described procedures [13,74,75]. Briefly, the EASA of the catalysts were determined from their CVs measured in deaerated 1 M H 2 SO 4 at 25 • C with a 20 mV/s sweep rate. The charges associated with the adsorption/desorption of hydrogen monolayers were taken to calculate the EASA (m 2 /g Pt) [74,76]. ASTs aimed at studying the impact of the corrosion of the support on the durability of the electrochemical performances were performed using an approach described earlier [13]. In brief, thin catalytic films were formed (as described above), and the working electrodes were subjected to potential cycling from 0.8 to 1.4 V vs. reversible hydrogen electrode at 0.1 V s −1 in an oxygen purged 1 M H 2 SO 4 solution.

Fuel Cell MEAs Fabrication and Testing
PEM fuel cell MEAs were prepared using Nafion ® 115 (DuPont Inc., Wilmington, DE, USA) as the polymer electrolyte membrane and Sigracet 39BC (SGL Group GmbH, Wiesbaden, Germany) as the anode and cathode GDEs. Pt 40 /C was used as the anode catalyst. The synthesized Pt/C, Pt/RGO, and Pt/RGO(s) materials were used as the cathode catalysts. The Pt loading of the anode and cathode electrocatalytic layers were 0.2 and 0.4 mg cm −2 , respectively. Catalytic layers were formed on the GDE surface by air-spraying catalytic ink containing 15 wt.% Nafion ® ionomer (Ion Power, Navarre, FL, USA), as described in the literature [20]. The MEA components were clamped in a circular titanium test cell with an active area of 7 cm 2 , as described elsewhere [12,20,77], and were kept at 90 • C for 2 h to ensure the tight binding of the MEA components. Such mild manufacturing conditions prevented the damage to the porous structure (the GDE and its microporous sublayer), which could be caused by any type of hot-press approach and ensured a high MEA performance at high current densities [78,79]. MEA testing was carried out using pure hydrogen and oxygen at a relative humidity of 100%, cell temperature of 60 • C, and at a bar gauge gas pressure of 1.

Conclusions
RGO-supported electrocatalysts designed for operation at the cathode of PEMFCs were synthesized using both the conventional two-step polyol process, and the simultaneous reduction of the Pt precursor and GO during a modified polyol process with EG and formaldehyde as the synthesis environment/reducing agents. The Pt/RGO(s) obtained by the optimized simultaneous reduction process showed high Pt utilization and EASA values, and a better durability during AST, provoking strong oxidation of the carbon support. The benefits of Pt/RGO(s) in comparison with conventionally prepared Pt/RGO and Pt/C are suggested to be as a result of the hindered RGO agglomeration during synthesis, optimized Pt nanoparticles morphology, and the ordered sp 2 graphitic structure of RGO. The implementation of the synthesized Pt/RGO(s) material as a cathode catalyst in H 2 /O 2 PEMFC allowed for increasing the maximum power density by up to 17% and to significantly enhance its performance, especially at high current densities. Funding: This work was partly supported by the scholarship of the President of the Russian Federation to young scientists and graduate students engaged in advanced research in priority areas of Russian economy modernization for 2019-2021 (CΠ-1764.2019.1), and partly within the framework of the project "Modified carbon nanomaterials for electrodes of fuel cells with solid polymer electrolyte", with the support of a grant from the National Research University "Moscow Power Engineering Institute" for the implementation of scientific research programs "Energy", "Electronics, Radio Engineering and IT", and "Industry 4.0 Technologies for Industry and Robotics in 2020-2022".

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.