Enhanced Performance of Pt Nanoparticles on Ni-N Co-Doped Graphitized Carbon for Oxygen Reduction Reaction in Polymer Electrolyte Membrane Fuel Cells

: Since the reaction rate and cost for cathodic catalyst in polymer electrolyte membrane fuel cells are obstacles for commercialization, the high-performance catalyst for oxygen reduction reaction is necessary. The Ni encapsulated with N-doped graphitic carbon (Ni@NGC) prepared with ethylenediamine and carbon black is employed as an efﬁcient support for the oxygen reduction reaction. Characterizations show that the Ni@NGC has a large surface area and mesoporous structure that is suitable to the support for the Pt catalyst. The catalyst structure is identiﬁed and the size of Pt nanoparticles distributed in the narrow range of 2–3 nm. Four different nitrogen species are doped properly into graphitic carbon structure. The Pt/Ni@NGC shows higher performance than the commercial Pt/C catalyst in an acidic electrolyte. The mass activity of the Pt/Ni@NGC in fuel cell tests exhibits over 1.5 times higher than that of commercial Pt/C catalyst. The Pt/Ni@NGC catalyst at low Pt loading exhibits 47% higher maximum power density than the Pt/C catalyst under H2-air atmosphere. These results indicate that the Ni@NGC as a support is signiﬁcantly beneﬁcial to improving activity.


Introduction
In a few decades, polymer electrolyte membrane fuel cells (PEMFCs) have rapidly developed and been considered as a promising candidate for the replacement of fossil fuels due to eco-friendly and highly efficient energy systems. Pt is known as the most efficient and active catalyst for the oxygen reduction reaction (ORR) [1,2]. However, there are significant issues for commercialization. Since Pt is a costly and scarce resource, it is required to use a small quantity and increase the catalytic performance.
Recently, the N-doped carbon structure has attracted much interest in the field of electrochemical energy conversion and storage due to its greater electron mobility like n-type or metallic behavior than un-doped carbon nanostructures [3]. Furthermore, doping nitrogen efficiently implants chemically active sites for catalysis and also plays a role in deposition sites for metal nanoparticle deposition. It has been reported that nitrogen-doped carbon as a support can increase the activity of a Pt catalyst toward the ORR [4][5][6][7]. Since those advantages of nitrogen-doped carbon structure were discovered, researchers have tried to replace the carbon black with nitrogen-doped carbon material as a support of electro-catalyst in PEMFCs [8,9]. Wang et al. reported that Pt@NiNC with low Pt loading (8 wt%) prepared by the galvanic replacement reaction exhibited the synergistic effect of Pt particles near atomic Ni-N-C complexes [10]. The Pt nanoparticles on Pt@NiNC carried more positive charges than the Pt foil, which indicated that the Pt atoms existed in the metallic form. Also, we demonstrated that the metallic Pt percentage of Pt nanoparticles anchored on N-doped KB) and Ni@NGC. The specific surface area of Ni@NGC was 425 m 2 g −1 , while the ox-KB had 634 m 2 g −1 . The change of specific surface area was also found in the previous results [23]. After pyrolysis, the total pore volume was reduced from 0.846 to 0.688 cm 3 g −1 and the peak pore diameter, which yields the peak differential pore volume in BJH PSD, as shown in Figure 1b, was slightly shifted from 3.9 (ox-KB) to 3.5 nm (Ni@NGC). The result could be revealed by the formation of the graphitized carbon layer derived from the Ni-EDA chelate and the shrinkage of the carbon frameworks due to the high-temperature pyrolysis [24,25]. The isotherms of the ox-KB and Ni@NGC corresponded to Type IV according to IUPAC classification. A hysteresis loop with sharp adsorption and desorption branches in a P/P0 range of 0.4-0.8 is observed in the isotherms. Besides, nitrogen uptake is observed at a relatively high P/P0 of 0.94-1.0, which indicates the presence of mesopores. Figure 2 presents X-ray diffraction (XRD) patterns of the ox-KB and Ni@NGC. The broad characteristic diffraction peaks of (002) and (101) were found at ca. 25 and 42° for carbon species. The diffraction peak of Ni@NGC was sharper, shifted to the positive angle, and had increased intensity. For example, the diffraction peak of Ni@NGC at the C(002) plane showed 26° while the peak of ox-KB exhibited 25.2°, which indicates that Ni@NGC was highly graphitized in the carbon matrix [3,7]. It can be seen that Ni@NGC showed characteristic diffraction peaks at 44.5, 51.8, and 76.4°, which correspond to the (111), (200), and (220) planes of face-centered cubic (fcc) structure of Ni metal particle (JCPDS 04-0850), respectively. It showed that the Ni metal remained after acid-leaching at 80 °C. The particle size of Ni calculated by the Scherrer equation was 15.7, 13.5, and 14.2 nm based on (111), (200), and (220).
High resolution transmission electron microscope (HR-TEM) images for ox-KB and Ni@NGC are shown in Figure 3. Figure 3a showed ox-KB with an amorphous structure without any additional layer. As shown in Figure 3b, however, the ox-KB was covered with the graphitic carbon layer. The Ni particles encapsulated within the graphitic carbon layer in Figure 3b,c were observed in the carbon framework. Also, some carbon tubes could be seen since the Ni particles are efficient catalysts for the graphitization of carbon materials. This is the reason why the peaks corresponding to Ni after acid-leaching were observed in the XRD pattern of Ni@NGC in Figure 2.
To further investigate the surface chemical analysis on Ni@NGC, X-ray photoelectron spectroscopy (XPS) analysis was employed. Figure 4 shows N 1s XPS spectra of Ni@NGC. N 1s XPS spectra were deconvoluted into four major components such as pyridinic, pyrrolic and/or pyridone, quarternary, and pyridinic-N + -O − (oxidized nitrogen) [11,26,27].  2 presents X-ray diffraction (XRD) patterns of the ox-KB and Ni@NGC. The broad characteristic diffraction peaks of (002) and (101) were found at ca. 25 and 42 • for carbon species. The diffraction peak of Ni@NGC was sharper, shifted to the positive angle, and had increased intensity. For example, the diffraction peak of Ni@NGC at the C(002) plane showed 26 • while the peak of ox-KB exhibited 25.2 • , which indicates that Ni@NGC was highly graphitized in the carbon matrix [3,7]. It can be seen that Ni@NGC showed characteristic diffraction peaks at 44.5, 51.8, and 76.4 • , which correspond to the (111), (200), and (220) planes of face-centered cubic (fcc) structure of Ni metal particle (JCPDS 04-0850), respectively. It showed that the Ni metal remained after acid-leaching at 80 • C. The particle size of Ni calculated by the Scherrer equation was 15.7, 13.5, and 14.2 nm based on (111), (200), and (220). The peak at 396.6 eV accounts for the presence of pyridinic-N, whereas the peak at 400.4 eV corresponds to the pyrrolic-N and/or pyridine-N. Peaks at 401.1 and 403.5 eV are ascribed to the presence of quaternary-N and pyridinic-N + -O − , respectively. The concentrations of pyridinic-N and pyrrolic-N and/or pyridine-N were 42.8 and 39.5% of total nitrogen, respectively, which indicated that most nitrogen existed in two components. Quaternary-N accounted for 4.1% while pyridinic-N + -O − occupied 13.6%. According to the previous results [27], the composition of quaternary-N can vary with the activity of the catalyst for the graphitization. The low composition of quaternary-N could be obtained since Ni is not highly active for the graphitization and the temperature is not high enough to form the highly graphitized carbon. It is well-known that pyridinic-N located on the edge of the graphite planes improves the ORR by donating one p-electron to the aromatic π system since it is located on the edge of the graphite planes [3,28,29]. Moreover, the latest investigations showed that quaternary-N plays a role in stable ORR active sites [20,30]. In Figure 4b, C 1s XPS spectra were shown. The main peaks of C 1s at ca. 284.4 and 284.7 eV are ascribed to the sp2 C of C=C bonding and the sp3 C of C-C, while the peaks at ca. 286.1, 288.1, and 288.9 eV are attributed to the carbon atoms in the C-OH, C=O, and O-C=O functional groups, respectively.  High resolution transmission electron microscope (HR-TEM) images for ox-KB and Ni@NGC are shown in Figure 3. Figure 3a showed ox-KB with an amorphous structure without any additional layer. As shown in Figure 3b, however, the ox-KB was covered with the graphitic carbon layer. The Ni particles encapsulated within the graphitic carbon layer in Figure 3b,c were observed in the carbon framework. Also, some carbon tubes could be seen since the Ni particles are efficient catalysts for the graphitization of carbon materials. This is the reason why the peaks corresponding to Ni after acid-leaching were observed in the XRD pattern of Ni@NGC in Figure 2.    To further investigate the surface chemical analysis on Ni@NGC, X-ray photoelectron spectroscopy (XPS) analysis was employed. Figure 4 shows N 1s XPS spectra of Ni@NGC. N 1s XPS spectra were deconvoluted into four major components such as pyridinic, pyrrolic and/or pyridone, quarternary, and pyridinic-N + -O − (oxidized nitrogen) [11,26,27]. The peak at 396.6 eV accounts for the presence of pyridinic-N, whereas the peak at 400.4 eV corresponds to the pyrrolic-N and/or pyridine-N. Peaks at 401.1 and 403.5 eV are ascribed to the presence of quaternary-N and pyridinic-N + -O − , respectively. The concentrations of pyridinic-N and pyrrolic-N and/or pyridine-N were 42.8 and 39.5% of total nitrogen, respectively, which indicated that most nitrogen existed in two components. Quaternary-N accounted for 4.1% while pyridinic-N + -O − occupied 13.6%. According to the previous results [27], the composition of quaternary-N can vary with the activity of the catalyst for the graphitization. The low composition of quaternary-N could be obtained since Ni is not highly active for the graphitization and the temperature is not high enough to form the highly graphitized carbon. It is well-known that pyridinic-N located on the edge of the graphite planes improves the ORR by donating one p-electron to the aromatic π system since it is located on the edge of the graphite planes [3,28,29]. Moreover, the latest investigations showed that quaternary-N plays a role in stable ORR active sites [20,30]. In Figure 4b, C 1s XPS spectra were shown. The main peaks of C 1s at ca. 284.4 and 284.7 eV are ascribed to the sp2 C of C=C bonding and the sp3 C of C-C, while the peaks at ca. 286.1, 288.1, and 288.9 eV are attributed to the carbon atoms in the C-OH, C=O, and O-C=O functional groups, respectively.   Figure 5 shows the cyclic voltammetry (CV) diagram of ox-KB and Ni@NGC. The measurements were performed in the nitrogen-saturated 0.1 M HClO4 at room temperature. The scan rate was 50 mV s −1 . In comparison with the Ni@NGC, the ox-KB showed well-developed redox peaks at ca 0.64 and 0.50 V during the positive and negative scans, respectively. These characteristic peaks correspond to the reversible quinone-hydroquinone redox coupling [31][32][33][34], which means oxygen groups were properly introduced through the acid-oxidation. According to the literature, the carbon black treated by the nitric acid shows about 100 mV less activation overpotential for the ORR compared to the carbon black [17]. The reaction related to redox coupling can be explained as follows:    Figure 5 shows the cyclic voltammetry (CV) diagram of ox-KB and Ni@NGC. The measurements were performed in the nitrogen-saturated 0.1 M HClO 4 at room temperature. The scan rate was 50 mV s −1 . In comparison with the Ni@NGC, the ox-KB showed well-developed redox peaks at ca 0.64 and 0.50 V during the positive and negative scans, respectively. These characteristic peaks correspond to the reversible quinone-hydroquinone redox coupling [31][32][33][34], which means oxygen groups were properly introduced through the acid-oxidation. According to the literature, the carbon black treated by the nitric acid shows about 100 mV less activation overpotential for the ORR compared to the carbon black [17]. The reaction related to redox coupling can be explained as follows: C=O + e − + H + ↔ C-OH In contrast to the ox-KB, the sharp peak at 0.50 and 0.65 V during the positive and negative scan, respectively, was not observed in the Ni@NGC. This is reasonable since the surface concentration of oxygen of Ni@NGC (1.7%) measured by XPS was even less than that of ox-KB (5.7%). The Linear sweep voltammetry (LSV) diagrams of ox-KB and Ni@NGC are shown in Figure 6. LSV was performed in the oxygen-saturated 0.1 M HClO4 at room temperature at a scan rate of 5 mV s −1 and a rotation speed of 1600 rpm. It was observed that the onset potential on the Ni@NGC for the ORR, which shifts positively by about 0.3 V compared to the ox-KB, was ca. 0.74 V. Also, Ni@NGC showed better kinetic than ox-KB. For example, the current density at 0.5 V was 0.32 mA cm −2 for Ni@NGC while only 0.01 mA cm −2 was observed with ox-KB. The diffusion-limited plateau of the polarization curve for both carbon materials was not well-defined as was usually observed in the Pt catalysts. Jiang et al. showed that the plateau was inclined when the active sites on the catalyst were not evenly distributed and the reaction was not fast [35].  In contrast to the ox-KB, the sharp peak at 0.50 and 0.65 V during the positive and negative scan, respectively, was not observed in the Ni@NGC. This is reasonable since the surface concentration of oxygen of Ni@NGC (1.7%) measured by XPS was even less than that of ox-KB (5.7%).

Electrochemical Characterization of Ni@NGC
The Linear sweep voltammetry (LSV) diagrams of ox-KB and Ni@NGC are shown in Figure 6. LSV was performed in the oxygen-saturated 0.1 M HClO 4 at room temperature at a scan rate of 5 mV s −1 and a rotation speed of 1600 rpm. It was observed that the onset potential on the Ni@NGC for the ORR, which shifts positively by about 0.3 V compared to the ox-KB, was ca. 0.74 V. Also, Ni@NGC showed better kinetic than ox-KB. For example, the current density at 0.5 V was 0.32 mA cm −2 for Ni@NGC while only 0.01 mA cm −2 was observed with ox-KB. The diffusion-limited plateau of the polarization curve for both carbon materials was not well-defined as was usually observed in the Pt catalysts. Jiang et al. showed that the plateau was inclined when the active sites on the catalyst were not evenly distributed and the reaction was not fast [35]. C=O + e − + H + ↔ C-OH In contrast to the ox-KB, the sharp peak at 0.50 and 0.65 V during the positive and negative scan, respectively, was not observed in the Ni@NGC. This is reasonable since the surface concentration of oxygen of Ni@NGC (1.7%) measured by XPS was even less than that of ox-KB (5.7%). The Linear sweep voltammetry (LSV) diagrams of ox-KB and Ni@NGC are shown in Figure 6. LSV was performed in the oxygen-saturated 0.1 M HClO4 at room temperature at a scan rate of 5 mV s −1 and a rotation speed of 1600 rpm. It was observed that the onset potential on the Ni@NGC for the ORR, which shifts positively by about 0.3 V compared to the ox-KB, was ca. 0.74 V. Also, Ni@NGC showed better kinetic than ox-KB. For example, the current density at 0.5 V was 0.32 mA cm −2 for Ni@NGC while only 0.01 mA cm −2 was observed with ox-KB. The diffusion-limited plateau of the polarization curve for both carbon materials was not well-defined as was usually observed in the Pt catalysts. Jiang et al. showed that the plateau was inclined when the active sites on the catalyst were not evenly distributed and the reaction was not fast [35].

Physical Characterization of Pt/Ni@NGC
The XRD patterns of the Pt/Ni@NGC and Pt/C are shown in Figure 7. Both the supported Pt catalysts exhibited the characteristics of the Pt fcc structure. The characteristic diffraction peaks at 39.8, 46.7, 67.7, and 81.8 • corresponded to the (111), (200), (220), and (311) planes of Pt nanoparticles, respectively, while those at 44.5, 51.8, and 76.4 • were ascribed to (111), (200), and (220) planes of Ni nanoparticles, respectively, as shown in Figure 2. Based on Figures 2 and 7, it was observed that the Pt/Ni@NGC did not form the alloy formation with existing Ni species. The Pt particle sizes, which were calculated using a Scherrer equation based on Pt (220) plane, were 2.4 nm and 2 nm for the Pt/Ni@NGC and Pt/C, respectively.

Physical Characterization of Pt/Ni@NGC
The XRD patterns of the Pt/Ni@NGC and Pt/C are shown in Figure 7.  Figures 2 and 7, it was observed that the Pt/Ni@NGC did not form the alloy formation with existing Ni species. The Pt particle sizes, which were calculated using a Scherrer equation based on Pt (220) plane, were 2.4 nm and 2 nm for the Pt/Ni@NGC and Pt/C, respectively.   Figure 8a reveals that the Pt nanoparticles prepared with the sodium formate as a reducing agent were successfully deposited on the Ni@NGC. The Pt nanoparticles were distributed evenly on the Ni@NGC with 2-3 nm of the mean particle size. The mean particle size observed in HR-TEM image of Pt/Ni@NGC was in good agreement with the result of XRD shown in Figure 7. Figure 8b is an HR-TEM image of Pt/C. The mean particle size of Pt/C was in the range of 2-3 nm.   Figure 8a reveals that the Pt nanoparticles prepared with the sodium formate as a reducing agent were successfully deposited on the Ni@NGC. The Pt nanoparticles were distributed evenly on the Ni@NGC with 2-3 nm of the mean particle size. The mean particle size observed in HR-TEM image of Pt/Ni@NGC was in good agreement with the result of XRD shown in Figure 7. Figure 8b is an HR-TEM image of Pt/C. The mean particle size of Pt/C was in the range of 2-3 nm.  Figures 2 and 7, it was observed that the Pt/Ni@NGC did not form the alloy formation with existing Ni species. The Pt particle sizes, which were calculated using a Scherrer equation based on Pt (220) plane, were 2.4 nm and 2 nm for the Pt/Ni@NGC and Pt/C, respectively.   Figure 8a reveals that the Pt nanoparticles prepared with the sodium formate as a reducing agent were successfully deposited on the Ni@NGC. The Pt nanoparticles were distributed evenly on the Ni@NGC with 2-3 nm of the mean particle size. The mean particle size observed in HR-TEM image of Pt/Ni@NGC was in good agreement with the result of XRD shown in Figure 7. Figure 8b is an HR-TEM image of Pt/C. The mean particle size of Pt/C was in the range of 2-3 nm.

Electrochemical Characterization and Performance
CV diagrams of Pt/Ni@NGC and Pt/C are presented in Figure 9. The experiments were achieved in the nitrogen-saturated 0.1 M HClO 4 at room temperature using a threeelectrode electrochemical cell. It was seen that the two redox reactions occurred on the Pt catalyst. One was hydrogen oxidation and reduction under ca. 0.4 V, while the other was oxygen oxidation and reduction on the Pt catalyst. Both catalysts showed the typical characteristic of polycrystalline Pt as shown in Figure 9. The ECSA in the CV diagrams was calculated from the hydrogen desorption region using the following equation: wherein Q H is hydrogen desorption charge in mC cm −2 , and L Pt is Pt loading in mg Pt cm −2 . The ECSA of Pt/C was 78.9 m 2 g −1 , while that of Pt/Ni@NGC showed 72.9 m 2 g −1 .

Electrochemical Characterization and Performance
CV diagrams of Pt/Ni@NGC and Pt/C are presented in Figure 9. The experiments were achieved in the nitrogen-saturated 0.1 M HClO4 at room temperature using a threeelectrode electrochemical cell. It was seen that the two redox reactions occurred on the Pt catalyst. One was hydrogen oxidation and reduction under ca. 0.4 V, while the other was oxygen oxidation and reduction on the Pt catalyst. Both catalysts showed the typical characteristic of polycrystalline Pt as shown in Figure 9. The ECSA in the CV diagrams was calculated from the hydrogen desorption region using the following equation: wherein QH is hydrogen desorption charge in mC cm −2 , and LPt is Pt loading in mgPt cm −2 . The ECSA of Pt/C was 78.9 m 2 g −1 , while that of Pt/Ni@NGC showed 72.9 m 2 g −1 . LSV diagrams of Pt/Ni@NGC and Pt/C are shown in Figure 10a. LSV was performed in the oxygen-saturated 0.1 M HClO4 at room temperature. The catalyst loading was 20 µgPt cm −2 for both catalysts. The diffusion-limited current of Pt/Ni@NGC was higher than that of Pt/C, while both Pt/Ni@NGC and Pt/C exhibited the identical on-set potential for ORR with ca. 1.0 V. The diffusion-limited current of Pt/Ni@NGC was observed as 5.34 mA cm −2 at 0.4 V whereas that of Pt/C showed 5.02 mA cm −2 . Figure 10b shows ORR activities of Pt/Ni@NGC and Pt/C in the range from 0.80 to 0.92 V. To compare those activities, the activity plot was prepared using the following equation [36]: where i is the measured current in LSV experiment, ik is the mass transport free kinetic current, and id is the diffusion-limited current. As shown in Figure 10b, Pt/Ni@NGC shows higher kinetic current than Pt/C. For example, at 0.85 V, the kinetic current of Pt/C was 1 mA, while that of Pt/Ni@NGC showed 1.2 mA. The Koutecky-Levich plots show a linear dependence as shown in Figure 10c. The linearity and the parallelism of these plots, in general, indicate first-order kinetics with respect to molecular oxygen [37]. Moreover, Pt/Ni@NGC shows to be more favorable to the four-electron pathway reaction over Pt/C. LSV diagrams of Pt/Ni@NGC and Pt/C are shown in Figure 10a. LSV was performed in the oxygen-saturated 0.1 M HClO 4 at room temperature. The catalyst loading was 20 µg Pt cm −2 for both catalysts. The diffusion-limited current of Pt/Ni@NGC was higher than that of Pt/C, while both Pt/Ni@NGC and Pt/C exhibited the identical on-set potential for ORR with ca. 1.0 V. The diffusion-limited current of Pt/Ni@NGC was observed as 5.34 mA cm −2 at 0.4 V whereas that of Pt/C showed 5.02 mA cm −2 . Figure 10b shows ORR activities of Pt/Ni@NGC and Pt/C in the range from 0.80 to 0.92 V. To compare those activities, the activity plot was prepared using the following equation [36]: where i is the measured current in LSV experiment, i k is the mass transport free kinetic current, and i d is the diffusion-limited current. As shown in Figure 10b, Pt/Ni@NGC shows higher kinetic current than Pt/C. For example, at 0.85 V, the kinetic current of Pt/C was 1 mA, while that of Pt/Ni@NGC showed 1.2 mA. The Koutecky-Levich plots show a linear dependence as shown in Figure 10c. The linearity and the parallelism of these plots, in general, indicate first-order kinetics with respect to molecular oxygen [37]. Moreover, Pt/Ni@NGC shows to be more favorable to the four-electron pathway reaction over Pt/C.   Figure 11a shows the mass activity of catalysts supplying H2 and O2 to the anode and cathode at 100% relative humidity (RH). The mass activity of Pt/C at 0.9 ViR-corrected was 0.21 A mgPt −1 , while Pt/Ni@NGC showed 0.35 A mgPt −1 , which means the mass activity of Pt/Ni@NGC is enhanced by 66.7% compared to that of Pt/C. The improvement in the mass activity proves that the use of active Ni@NGC gives rise to the synergetic effect with Pt for the ORR at the cathode. Figure 11b,c shows the polarization curves and their corresponding power density curves under H2 and air atmosphere at a 40% RH. The Membrane Electrode Assembly (MEA) with Pt/Ni@NGC as a cathode catalyst showed a maximum power density of 464 mW cm −2 , while the MEA with Pt/C used in the cathode catalyst showed a maximum power density of 350 mW cm −2 . At 800 mA cm −2 , the potential of Pt/Ni@NGC exhibited 0.565 V and that of Pt/C showed 0.437 V.
Furthermore, the open-circuit potential (OCP) of Pt/Ni@NGC was revealed at 0.988 V, which was even higher than that of Pt/C. The improved performance with Pt/Ni@NGC as a cathode catalyst can be attributed to the nitrogen-doped graphitic carbon layer containing pyridinic-N and quaternary-N components. As shown in Figure 6, Ni@NGC, which acts as a support, has activity toward the ORR. Its activity can be added to the activity of   Figure 11a shows the mass activity of catalysts supplying H 2 and O 2 to the anode and cathode at 100% relative humidity (RH). The mass activity of Pt/C at 0.9 V iR-corrected was 0.21 A mg Pt −1 , while Pt/Ni@NGC showed 0.35 A mg Pt −1 , which means the mass activity of Pt/Ni@NGC is enhanced by 66.7% compared to that of Pt/C. The improvement in the mass activity proves that the use of active Ni@NGC gives rise to the synergetic effect with Pt for the ORR at the cathode. Figure 11b,c shows the polarization curves and their corresponding power density curves under H 2 and air atmosphere at a 40% RH. The Membrane Electrode Assembly (MEA) with Pt/Ni@NGC as a cathode catalyst showed a maximum power density of 464 mW cm −2 , while the MEA with Pt/C used in the cathode catalyst showed a maximum power density of 350 mW cm −2 . At 800 mA cm −2 , the potential of Pt/Ni@NGC exhibited 0.565 V and that of Pt/C showed 0.437 V. Furthermore, the open-circuit potential (OCP) of Pt/Ni@NGC was revealed at 0.988 V, which was even higher than that of Pt/C. The improved performance with Pt/Ni@NGC as a cathode catalyst can be attributed to the nitrogen-doped graphitic carbon layer containing pyridinic-N and quaternary-N components. As shown in Figure 6, Ni@NGC, which acts as a support, has activity toward the ORR. Its activity can be added to the activity of Pt for the ORR in the cathode. It has been reported that nitrogen doping results in disorder in graphene stacking. Such disorder structures and defects can play an important role in providing anchoring sites for the deposition of Pt nanoparticles [38][39][40][41]. Also, the performance can be enhanced due to the improved binding of carbon and catalyst and electrical conductivity by nitrogen doping [42,43]. Groves et al. showed that the more nitrogen atoms and the closer they are to the carbon atom bonding directly to the Pt particle, the stronger the binding energy using density functional theory (DFT) [44].
Catalysts 2021, 11, x 10 of 14 Pt for the ORR in the cathode. It has been reported that nitrogen doping results in disorder in graphene stacking. Such disorder structures and defects can play an important role in providing anchoring sites for the deposition of Pt nanoparticles [38][39][40][41]. Also, the performance can be enhanced due to the improved binding of carbon and catalyst and electrical conductivity by nitrogen doping [42,43]. Groves et al. showed that the more nitrogen atoms and the closer they are to the carbon atom bonding directly to the Pt particle, the stronger the binding energy using density functional theory (DFT) [44].

Preparation of Ni@NGC and Pt/Ni@NGC
The Ni@NGC was synthesized with a similar method as described elsewhere [20,26]. Briefly, first, commercially available carbon black (Ketjen Black EC-300J, Akzo Nobel, Amsterdam, The Netherlands) was oxidized with concentrated nitric acid. Two milliliters of ethylenediamine (EDA, Aldrich, St. Louis, MO, USA) and 0.5 g of Ni(NO3)2•6H2O (Aldrich) as a nitrogen source and catalyst for the graphitization, respectively, were blended

Preparation of Ni@NGC and Pt/Ni@NGC
The Ni@NGC was synthesized with a similar method as described elsewhere [20,26]. Briefly, first, commercially available carbon black (Ketjen Black EC-300J, Akzo Nobel, Amsterdam, The Netherlands) was oxidized with concentrated nitric acid. Two milliliters of ethylenediamine (EDA, Aldrich, St. Louis, MO, USA) and 0.5 g of Ni(NO 3 ) 2 ·6H 2 O (Aldrich) as a nitrogen source and catalyst for the graphitization, respectively, were blended with ox-KB homogeneously in isopropyl alcohol (IPA, BDH). After reflux at 85 • C for 3 h, the solvent was removed by the rotary evaporator (Rotavapor ® R-210, Buchi, New Castle, DE, USA). After fully dried, the powder was moved to the tubular furnace (OTF-1200X-SNT-110, MTI Corp., Richmond, VA, USA) to produce the N-doped graphitic carbon structure at 800 • C for 1 h under nitrogen atmosphere. Unstable metal species in the sample were removed with 0.5 M H 2 SO 4 at 80 • C for 4 h. After filtering and washing with de-ionized (DI) water, the dried sample was utilized as a support of Pt catalyst.
Pt deposition was carried out using the electroless deposition method. The support was mixed with 0.02 M chloroplatinic acid hexahydrate (Alfa Aesar, Haverhill, MA, USA) in DI water. After mixing homogeneously, the solution was refluxed at 70 • C. 2 M sodium formate (Alfa Aesar, Haverhill, MA, USA) was added into the mixture slowly, followed by refluxing for 12 h. The resulting catalyst was washed with DI water and then fully dried at 80 • C. Consequently, 30% Pt/Ni@NGC was obtained.

Physical Characterization
The nitrogen adsorption/desorption isotherms were measured at −196 • C using a BET analyzer (Quantachrome NOVA 2000). Specific surface area for the sample was obtained by a multipoint Brunauer-Emmett-Teller (BET) analysis. XRD analysis was carried out using a Rigaku D/Max 2500 V/PC with a Cu Kα radiation (A tube voltage of 30 kV and a current of 15 mA).
XPS was performed with a Kratos AXIS 165 high-performance electron spectrometer to measure the elemental surface composition. HR-TEM was obtained by Hitachi 9500 HRTEM (300 kV accelerating voltage).

Electrochemical Characterization
Electrochemical characterization of the Ni@NGC and Pt/Ni@NGC was carried out in a rotating disk electrode (RDE) using a bi-potentiostat (Model AFCBP1, Pine Research Instrumentation, Durham, NC, USA). A glassy carbon disk electrode (0.247 cm 2 ) functioned as a working electrode, while the platinum mesh and Ag/AgCl electrodes acted as counter and reference electrodes, respectively. RDE tests were operated in 0.1 M HClO 4 electrolyte at ambient temperature. All potentials were reported against the reversible hydrogen electrode (RHE). Typically, for carbon species, 8 mg of carbon was ultrasonically dispersed in 1 mL of IPA. Fifteen microliters of the ink was dropped on the glassy carbon electrode. For the Pt catalyst experiment, the catalyst ink was made of Pt catalyst, absolute ethanol, and DI water ultrasonically. The catalyst ink was dropped onto the glassy carbon (catalyst loading: 20 µg Pt cm −2 ). To provide good adhesion, 5 µL of 0.25 wt% ionomer (Alfa Aesar, Haverhill, MA, USA) was added onto the catalyst layer. CV was performed at a scan rate of 20 mV s −1 in a range of 0.005 V and 1.0 V purging with nitrogen. LSV was carried out at a scan rate of 5 mV s −1 in a range of 0.2 V and 1.05 V purging with oxygen. The LSV curves shown in this work were adequately corrected with the background capacitance current that was obtained in the nitrogen atmosphere at a scan rate of 5 mV s −1 . For comparison purposes, the ox-KB and commercial Pt/C catalyst (TEC10EA50E, TKK, Tokyo, Japan) were evaluated at the same conditions.

MEA Fabrication and Single Cell Operation
For catalyst inks, a catalyst, IPA, ionomer, and DI water were ultrasonically mixed. The ionomer contents for the anode and cathode were 30% and 20%, respectively. To load 0.1 mg Pt cm −2 for both the anode and cathode, the thus prepared catalyst inks were manually sprayed on the gas diffusion layer (Sigracet GDL 10BC, SGL, Wiesbaden, Germany). To confirm the catalyst loading, we employed X-ray fluorescence (XRF, Model XDAL, Fisher, Waltham, MA, USA). The commercial Pt/C catalyst was utilized as a catalyst for the anode. Both electrodes were pressed onto both sides of a polymer electrolyte membrane (Nafion-212, Dupont, Wilmington, NC, USA) at 140 • C under 20 kg f cm −2 for 3 min. The thus fabricated MEA was cooled to ambient temperature and assembled in 25 cm 2 single cells. The single cell test was conducted using a fully automated test station (model 850e, Scribner Associates Inc., Southern Pines, NC, USA). The mass activity was obtained supplying 100% RH humidified H 2 and O 2 at the flow rates of 70 and 166 sccm, respectively. H 2 and air humidified at 59 • C flowed to the anode and cathode inlets, respectively, to obtain polarization curves at the single cell temperature of 80 • C. Flow rates of H 2 and air were controlled by the stoichiometric ratios of 1.5 and 1.8, respectively, under backpressure of 150 kPa abs . Below 0.2 A cm −2 , equivalent flows at 0.2 A cm −2 were supplied to both sides. For comparison purposes, MEA with the commercial Pt/C catalyst as a cathode was also fabricated and measured at the same conditions.

Conclusions
In this study, Ni@NGC was prepared with a facile approach and employed as a support of cathode catalyst in PEMFCs. Ni@NGC showed mesoporous characteristics and a large surface area with the graphitized carbon structure containing the nitrogen. Chelated-EDA was converted to the N-doped graphitic carbon layer, which involved pyrrolic and/or pyridone, pyridinic, quarternary, and pyridinic-N + -O − . pyridinic and/or quaterary-N. After these groups were doped in the carbon structure, the activity for the ORR was significantly enhanced. Also, we deposited Pt on the active Ni@NGC as a support, instead of conventional carbon black. According to HR-TEM and XRD, Pt particles were evenly deposited on the surface of Ni@NGC. RDE showed that Pt/Ni@NGC was more active and reduced the oxygen via a more effective four-electron pathway than the commercial Pt/C. Based on these beneficial properties of Pt/Ni@NGC, in the fuel cell test, the mass activity of Pt/Ni@NGC showed to be over 1.5 times higher than that of commercial Pt/C. As well, the maximum power density of Pt/Ni@NGC was 464 mW cm −2 , while that of Pt/C showed 350 mW cm −2 .