Nitrogen-Doped Reduced Graphene Oxide Supported Pd4.7Ru Nanoparticles Electrocatalyst for Oxygen Reduction Reaction

It is imperative to design an inexpensive, active, and durable electrocatalyst in oxygen reduction reaction (ORR) to replace carbon black supported Pt (Pt/CB). In this work, we synthesized Pd4.7Ru nanoparticles on nitrogen-doped reduced graphene oxide (Pd4.7Ru NPs/NrGO) by a facile microwave-assisted method. Nitrogen atoms were introduced into the graphene by thermal reduction with NH3 gas and several nitrogen atoms, such as pyrrolic, graphitic, and pyridinic N, found by X-ray photoelectron spectroscopy. Pyridinic nitrogen atoms acted as efficient particle anchoring sites, making strong bonding with Pd4.7Ru NPs. Additionally, carbon atoms bonding with pyridinic N facilitated the adsorption of O2 as Lewis bases. The uniformly distributed ~2.4 nm of Pd4.7Ru NPs on the NrGO was confirmed by transmission electron microscopy. The optimal composition between Pd and Ru is 4.7:1, reaching −6.33 mA/cm2 at 0.3 VRHE for the best ORR activity among all measured catalysts. Furthermore, accelerated degradation test by electrochemical measurements proved its high durability, maintaining its initial current density up to 98.3% at 0.3 VRHE and 93.7% at 0.75 VRHE, whereas other catalysts remained below 90% at all potentials. These outcomes are considered that the doped nitrogen atoms bond with the NPs stably, and their electron-rich states facilitate the interaction with the reactants on the surface. In conclusion, the catalyst can be applied to the fuel cell system, overcoming the high cost, activity, and durability issues.


Introduction
As the demand for renewable energy technology increases for a zero-emission society, energy conversion and storage devices are actively being studied. Among them, the fuel cell is one of the promising energy systems due to H 2 fuel, which has high energy density. Additionally, it emits only water as a by-product, without any pollutants such as carbon dioxide. Despite these advantages, it has a significant limitation: oxygen reduction reaction (ORR). ORR, the complex four electrons and multi-stepped reaction, works as the rate-determining step in fuel cells. For this reason, the use of catalysts is inevitable to facilitate the reaction of the cells. To date, carbon black supported Pt (Pt/CB) catalyst has been used the most widely due to its remarkable ORR performance. However, it has fatal disadvantages, including its high price, the scarcity of platinum, and the deterioration of carbon black under long operation periods. Therefore, innovative alternatives of Pt/CB with lower prices, abundance, and durability are essential for the development of renewable energy devices.
Regarding the active material, palladium has attracted attention as a promising ORR catalyst for its greater abundance than Pt and comparable properties, with the same crystal structure and group of the periodic table as Pt [1]. One of the efficient ways to improve its catalytic property and reduce its price is alloying Pd with other elements by rearranging the electronic structure, which is called the ensemble effect [2]. Among many metal elements,

Synthesis of Pd-Ru NPs on NrGO and rGO
Pd-Ru NPs/NrGO catalysts were synthesized by a microwave-assisted method [16][17][18]. First, 55 mg of NrGO was dispersed in 50 mL of diethylene glycol (DEG, Junsei Chemical Co., Tokyo, Japan) by ultrasonication for 3 h. After that, 0.1 M palladium (II) chloride (PdCl 2 , Wako Chemical Inc., Osaka, Japan) and ruthenium (III) chloride hydrate (RuCl 3 ·6H 2 O, Sigma-Aldrich Co., Burlington, MA, USA) were added into the solutions, totaling 1 mL in various ratios of 1:0, 4:1, 3:1, 1:1, and 0:1, respectively, to compare the catalytic activity depending on the elemental composition of Pd and Ru. At the same time, 1 mL of 0.5 M sodium hydroxide (NaOH, Junsei Chemical Co., Tokyo, Japan) was added to control reaction kinetics. Subsequently, the solutions were mixed by ultrasonication for another 1 h. After that, the prepared solutions were heated in a microwave oven at 700 W for 80 s. After cooling it to room temperature, DEG was separated by centrifugation at 8000 rpm for 50 min. Then, the catalysts were rinsed twice with acetone to remove residual DEG and impurities.

Materials Characterization
The surface morphologies of the fabricated catalysts were characterized by transmission electron microscopy (FE-TEM, Talos F200X, Thermo Fisher Scientific, Waltham, MA, USA and Cs-corrected STEM, JEM-ARM200F, JEOL, Tokyo, Japan) equipped with energy dispersive X-ray spectroscopy (EDS, Talos F200X, Thermo Fisher Scientific, Waltham, MA, USA). The crystalline structures were determined with selected area diffraction (SAED) and X-ray diffraction (XRD, Xpert 3, Malvern Panalytical, Malvern, UK, Cu Kα anode). The chemical states and bonding characteristics were analyzed through X-ray photoelectron spectrometer (XPS, K-alpha System, Thermo Fisher Scientifics, Waltham, MA, USA) with a monochromatic Al Kα (1486.6 eV). Raman spectroscopy (Micro Raman Spectrometer, NRS-5100, JASCO International Co., Tokyo, Japan) with laser excitation line of 512 nm was used to analyze defective and graphitic structures of the NrGO and rGO based catalysts. Elemental compositions and contents of Pd and Ru were investigated by ICP-OES (Optima 8300, PerkinElmer, Waltham, MA, USA).

Electrochemical Measurements
All electrochemical measurements were conducted using a standard three-electrode cell system connected to an electrochemical workstation (VSP, Biologic, Knoxville, Tennessee, USA) with a rotating ring-disk electrode rotator (RRDE-3A, ALS, Tokyo, Japan) in 0.1 M KOH (85%, Junsei Chemical Co., Tokyo, Japan) electrolyte. Pt coil (EC Frontier, Kyoto, Japan) and Hg/HgO (saturated 20% KOH) were used as a counter electrode and a reference electrode, respectively. Catalyst ink was prepared by adding 3 mg of the prepared catalyst powder into 1 mL of isopropyl alcohol (IPA, OCI Co., Seoul, Korea) and 0.1 mL of Nafion solution (5 wt %, Alfa aesar., Haverhill, MA, USA), followed by dispersing with ultrasonication. After that, 20 µL of the catalyst ink was deposited on the glassy carbon working electrode (GCE, 5 mm diameter, ALS, Tokyo, Japan) with~0.278 mg/cm 2 of loading mass and dried in the air. ORR activities were estimated by linear sweep voltammetry (LSV). LSV measurements were carried out before and after the accelerated degradation test (ADT) by rotating the working electrode at 1600 rpm in an O 2 -saturated electrolyte at 1 mV/s scan rate from 1.2 V RHE to 0 V RHE . To characterize the long-term durability of the catalysts with ADT, cyclic voltammetry (CV) was carried out for 1000 cycles at 50 mV/s scan rate from 1.2 V RHE to 0 V RHE . Additionally, electrochemical impedance spectroscopy (EIS) measurements were conducted before and after the ADT with a frequency range from 100 kHz to 100 mHz at 0.3 V RHE . Chrono-amperometry (CA) measurements were carried out for 15 h at 0.3 V RHE and 0.75 V RHE with rotating the working electrode. ECSA was compared by calculating electrochemical double-layer capacitance (C dl ) from CV tests at scan rates with 5, 10, 25, 50, 100, 150, and 200 mV/s between 1.1 and 1.0 V RHE . In addition, methanol and CO poisoning tests were conducted. For the methanol poisoning test, LSV curves were collected under the same conditions after dropping 1 mL of 1 M CH 3 OH from 1.2 V RHE to 0 V RHE 1 mV/s of scan rate. In case of CO poisoning test, 5 cycles of CV under CO saturation at 50 mV/s scan rate and following LSV tests under O 2 saturation at 1 mV/s scan rate were carried out to estimate the CO poisoning at the same scan range. All measured potential values of V vs. Hg/HgO were calibrated to the reversible hydrogen electrode (RHE) scale by Equation (1), and the pH of 0.1 M KOH electrolyte was 13.01. E RHE = E Hg/HgO + 0.098 + 0.059 pH (1) Figure 1 displays the Pd 4.7 Ru NPs/NrGO and ORR on the catalyst surface. During the thermal annealing with NH 3 flow, graphene oxide (GO) is reduced to be the rGO, and nitrogen atoms form covalent bondings with the carbon atoms in the graphene lattice. Several N sites, such as pyrrolic, graphitic, and pyridinic N, were fabricated, and Pd 4.7 Ru NPs were synthesized by the microwave-assisted method in a few nanometer size, especially anchored on the pyridinic N sites. These could facilitate the ORR activity of the catalyst due to the higher electron conductivity from heterogenous atoms and improved O 2 adsorption by the effect of C atoms neighboring pyridinic N sites [12,13]. Furthermore, strong bonding between the catalyst and N sites might prevent the NPs from agglomeration or dissolution from the surface. Pd 4.7 Ru NPs grow in a form of a Pd-based face-centered cubic (FCC) structure, and the lattice could be also distorted when Ru atoms are introduced, resulting in higher ORR activity than merely being in a single metal, and it was maximized when the ratio of Pd and Ru was 4.5:1-4.7:1.

Results and Discussions
at 50 mV/s scan rate and following LSV tests under O2 saturation at 1 mV/s scan rate were carried out to estimate the CO poisoning at the same scan range. All measured potential values of V vs. Hg/HgO were calibrated to the reversible hydrogen electrode (RHE) scale by Equation (1), and the pH of 0.1 M KOH electrolyte was 13.01. ERHE = EHg/HgO + 0.098 + 0.059 pH (1) Figure 1 displays the Pd4.7Ru NPs/NrGO and ORR on the catalyst surface. During the thermal annealing with NH3 flow, graphene oxide (GO) is reduced to be the rGO, and nitrogen atoms form covalent bondings with the carbon atoms in the graphene lattice. Several N sites, such as pyrrolic, graphitic, and pyridinic N, were fabricated, and Pd4.7Ru NPs were synthesized by the microwave-assisted method in a few nanometer size, especially anchored on the pyridinic N sites. These could facilitate the ORR activity of the catalyst due to the higher electron conductivity from heterogenous atoms and improved O2 adsorption by the effect of C atoms neighboring pyridinic N sites [12,13]. Furthermore, strong bonding between the catalyst and N sites might prevent the NPs from agglomeration or dissolution from the surface. Pd4.7Ru NPs grow in a form of a Pd-based face-centered cubic (FCC) structure, and the lattice could be also distorted when Ru atoms are introduced, resulting in higher ORR activity than merely being in a single metal, and it was maximized when the ratio of Pd and Ru was 4.5:1-4.7:1.  Figure 2c,f, demonstrate that Pd4.7Ru NPs/NrGO has 2.4 nm on average, while Pd4.5Ru NPs/rGO presents 3.4 nm on average. The well-doped nitrogen acts as an efficient particle generation point and alleviates to form small and uniform particles with wide surface area, which might consequently improve catalytic reaction. Moreover, Figure S1a,c show that single Pd NPs on NrGO and rGO are shown agglomerated distribution, whereas single Ru NPs on the supports were dispersed in a sub-nanometer size, as shown in Figure  S1b,d. These results suggest that the introduction of Ru in Pd lattice might affect the size and dispersity of NPs when synthesized.  Figure 2c,f, demonstrate that Pd 4.7 Ru NPs/NrGO has 2.4 nm on average, while Pd 4.5 Ru NPs/rGO presents 3.4 nm on average. The well-doped nitrogen acts as an efficient particle generation point and alleviates to form small and uniform particles with wide surface area, which might consequently improve catalytic reaction. Moreover, Figure S1a,c show that single Pd NPs on NrGO and rGO are shown agglomerated distribution, whereas single Ru NPs on the supports were dispersed in a sub-nanometer size, as shown in Figure S1b,d. These results suggest that the introduction of Ru in Pd lattice might affect the size and dispersity of NPs when synthesized.

Results and Discussions
To confirm the crystallinity of NPs, we collected selected area electron diffraction (SAED) patterns from the catalysts. The Pd 4.7 Ru NPs/NrGO is similar to Pd/NrGO, representing the rings of Pd (111), (200), (220), (311), and (331) planes (see Figure 2g and Figure S2a). Meanwhile, only Ru NPs/NrGO in Figure S2b shows Ru (101) and (200) facets, which could be distinguished from the abovementioned Pd planes. Additionally, the calculated Pd d-spacing value from the Pd (111) ring of the Pd 4.7 Ru NPs/NrGO (0.22 nm, shown in Table S2) was consistent with the value in Figure 2b. Interestingly, Table S2 [19,20]. Regarding the Pd 4.7 Ru particles, when Ru atoms that have a smaller atomic radius than Pd are substituted in the Pd lattice, it might shrink the Pd lattice; this was confirmed by the decrease in d-spacing value. These results also support that Pd and Ru coexist as alloys, not as separate phases. For further characterization, we conducted mapping and line scanning elemental analyses by HAADF-EDS. Figure 2h demonstrates that N, Pd, and Ru are clearly detected. This demonstrates that Pd and Ru appear in the form of NPs on the N-doped graphene. Moreover, the line scanning image in Figure 2i shows that Pd and Ru are observed simultaneously, indicating that they coexist in the NP. Consequently, Pd and Ru are synthesized in a nanometer-size on the graphene lattice, forming the non-separated alloy structure. To confirm the crystallinity of NPs, we collected selected area electron diffraction (SAED) patterns from the catalysts. The Pd4.7Ru NPs/NrGO is similar to Pd/NrGO, representing the rings of Pd (111), (200), (220), (311), and (331) planes (see Figure 2g and Figure  S2a). Meanwhile, only Ru NPs/NrGO in Figure S2b shows Ru (101) and (200) facets, which Additional structural analyses for the NPs and the graphene-based supports were conducted by XRD and Raman spectroscopy. Figure 3a,b show the XRD patterns of the NrGO and rGO-based catalysts, respectively. All measured catalysts have common peaks around 25.0 • from the graphite (002). However, there is no peak corresponding to either Pd or Ru in the Pd-Ru alloy and Ru-based catalysts. It is attributed to the result that small particles under 4 nm are anchored uniformly on the graphene-based supports, which is identical to the TEM results in Figure 2a,d and Figure S1b,d [21]. On the other hand, only Pd NPs both on the NrGO and rGO showed aggregated shapes and sized over 5 nm, as shown in Figure S2a,c, thus displaying a distinct peak at 40.4 • in the XRD spectra. As it corresponds to FCC Pd (111) peak, we could confirm the FCC crystallinity of Pd along with the SAED patterns in Figure S2a. loys, not as separate phases. For further characterization, we conducted mapping and scanning elemental analyses by HAADF-EDS. Figure 2h demonstrates that N, Pd, an are clearly detected. This demonstrates that Pd and Ru appear in the form of NPs o N-doped graphene. Moreover, the line scanning image in Figure 2i shows that Pd an are observed simultaneously, indicating that they coexist in the NP. Consequently, Pd Ru are synthesized in a nanometer-size on the graphene lattice, forming the non-sepa alloy structure.
Additional structural analyses for the NPs and the graphene-based supports conducted by XRD and Raman spectroscopy. Figure 3a,b show the XRD patterns o NrGO and rGO-based catalysts, respectively. All measured catalysts have common p around 25.0° from the graphite (002). However, there is no peak corresponding to e Pd or Ru in the Pd-Ru alloy and Ru-based catalysts. It is attributed to the result that particles under 4 nm are anchored uniformly on the graphene-based supports, wh identical to the TEM results in Figure 2a,d and Figure S1b,d [21]. On the other hand, Pd NPs both on the NrGO and rGO showed aggregated shapes and sized over 5 n shown in Figure S2a,c, thus displaying a distinct peak at 40.4° in the XRD spectra. corresponds to FCC Pd (111) peak, we could confirm the FCC crystallinity of Pd a with the SAED patterns in Figure S2a. To compare the status of two graphene supports, we additionally conducted Ra spectroscopy. As shown in Figure 3c,d, all samples have two separate peaks around and 1590 cm −1 (Table 1) of the D and G bands, which contain the information of de To compare the status of two graphene supports, we additionally conducted Raman spectroscopy. As shown in Figure 3c,d, all samples have two separate peaks around 1345 and 1590 cm −1 (Table 1) of the D and G bands, which contain the information of defects and graphitic degree [22,23]. The NrGO has a lower I D /I G ratio (1.04) than the rGO (1.19), indicating the ordered graphene system by the introduced nitrogen atoms [24,25]. Additionally, this tendency is consistent with Pd-Ru-anchored catalysts with values of 1.02 and 1.14 for the Pd 4.7 Ru NPs/NrGO and Pd 4.5 Ru NPs/rGO, respectively. Furthermore, blue-shifts of the D and G bands are observed in Pd-Ru NPs samples both on the NrGO and rGO, compared with the pristine NrGO and rGO sheets due to the compressive strain by anchoring particles [26,27]. Moreover, in the case of NrGO-based catalysts, the degree of the shifted D and G band is higher than the rGO-based catalyst, indicating that more compressive strain is applied over the NrGO after particle anchoring. It might connect to the stronger bonding of Pd 4.7 Ru particles and NrGO than Pd 4.5 Ru and rGO as scattered at higher frequency. We conducted XPS analyses to characterize the chemical states and bonding configurations of Pd 4.7 Ru NPs/NrGO and Pd 4.5 Ru NPs/rGO. In Figure 4a of C 1s spectra, common peaks at 284.7, 286.3, and 289.1 eV are observed from both samples, which corresponds to C-C, C-O, and C=O bonding, respectively [28,29]. C-C bonding originated from the graphene lattice, and C-O and C=O bondings are from the residual hydroxyl, carbonyl, and epoxy on the NrGO and rGO surfaces. In addition, only in the Pd 4.7 Ru NPs/NrGO sample, C-N bonding appears at 285.8 eV, indicating doped N atoms form stable covalent bonding with the neighboring C atoms. To clarify the bonding states of N atoms, N 1s spectra were also recorded at Figure 4b. The spectra are deconvoluted into pyridinic, pyrrolic, graphitic, and oxidized N peaks at 398.7, 399.8, 401.2, and 402.5 eV with ratios of 36.5, 31.9, 17.7, and 13.9%, respectively [30]. Previous research demonstrated that doped nitrogen atoms in carbon-based support, especially the pyridinic N, supply the particle anchoring sites, and the pyridinic N is dominant in this sample with the contents of 36.5% [7,8,18,31]. Likewise, it was proved in this research by comparing the N 1s peak of Pd 4.7 Ru NPs/NrGO and NrGO in Figure S3a. The pyridinic N peak in NrGO is displayed at 398.2 eV and shifted to 398.7 eV after synthesizing with Pd 4.7 Ru NPs, and it could be inferred as the NPs were anchored on pyridinic N [8,32]. Pd 3p photoemission lines in Figure 4c display two distinct peaks at 335.7 and 340.9 eV, corresponding with the metallic Pd 3d 5/2 and Pd 3d 3/2 , respectively. Additionally, two oxidic states are shown simultaneously at 337.4 eV and 342.5 eV, which are matched with the Pd 2+ 3d 5/2 and Pd 2+ 3d 3/2 , respectively. These peaks are detected at the same position in the Pd NPs/NrGO and Pd NPs/rGO, as shown in Figure S3b. Interestingly, Ru 3p spectra show only metallic peaks of Ru 3p 1/2 and Ru 3p 3/2 without any oxidized peak in the Pd-Ru-based catalysts (see Figure 4d), whereas both metallic and oxidic states are seen in Ru NPs (see Figure S3c). It is considered that the residual oxygen at graphene sheets might be reacted with Pd 2+ ions preferentially than Ru 4+ because it is reported that PdO is formed at lower temperature than RuO 2 [33][34][35]. From these results, we could assume that the Pd 4.7 Ru NPs/NrGO and the Pd 4.5 Ru NPs/rGO catalysts have metal-dominant phases of Pd and Ru.
To investigate electrochemical performance of the catalysts, we carried out the ORR analyses of the Pd 4.7 Ru NPs/NrGO, Pd 4.5 Ru NPs/rGO, and Pt/CB with a three-electrode system in O 2 gas saturated 0.1 M KOH electrolyte. Figure 5a shows the LSV measurement of cathodic scanning plots. Among the catalysts, the Pd 4.7 Ru NPs/NrGO achieves the highest onset potential at −0.1 mA/cm 2 and half-wave potential with 0.913 V RHE and 0.792 V RHE , followed by the Pt/CB with 0.908 V RHE and 0.785 V RHE , and Pd 4.5 Ru NPs/rGO with 0.863 V RHE and 0.713 V RHE , respectively (see Table 2). Additionally, the Pd 4.7 Ru NPs/NrGO, Pd 4.5 Ru NPs/rGO, and Pt/CB exhibited the limited current density of −6.33 mA/cm 2 , −5.46 mA/cm 2 , and −5.36 mA/cm 2 at 0.3 V RHE, and the specific activities were also plotted at Figure 5b. These results could be explained by the previous XPS analyses that doped pyridinic nitrogen atoms contribute to increasing the onset potential value of ORR. Additionally, we compared the activity of Pd 4.7 Ru NPs/NrGO to other previously studied Pd-based catalysts on their onset potential values, which are summarized in Table S3. Among them, nitrogen-doped carbon-supported Pd catalysts show relatively high onset potentials over 0.89 V RHE than undoped catalysts, and it is explained by the role of pyridinic N [36]. Moreover, the Pd 4.7 Ru catalyst has superior activity to Pd single element catalysts, confirming the alloying effect. We also compared the charge transfer resistance of the catalysts by EIS at 0.3 V RHE . As represented in Figure 5c, three components constitute the Randles circuit: solution resistance R s , charge transfer resistance R ct , and C dl [37]. R s and R ct were measured at 100 kHz and 100 mHz, respectively, and C dl is shown in form of a semicircle as the frequency decreased. In the graph, the Pd 4.7 Ru NPs/NrGO has the lowest charge transfer value, of which R ct is 115.2 Ω, followed by the Pd 4.5 Ru NPs/rGO and Pt/CB with 327.7 Ω and 336.7 Ω, respectively. This result confirms the higher conductivity of NrGO with more valence electrons of nitrogen atoms in honeycomb structure with carbon atoms. Additionally, it is consistent with the former results that oxygen reduction occurs faster on the Pd 4.7 Ru NPs/NrGO than commercial Pt/CB and Pd 4.5 Ru NPs/rGO. Additionally, to investigate the ORR activity according to the ratio of Pd and Ru, we carried out the LSV analyses on the prepared Pd-Ru alloy catalysts. LSV curves of the catalysts supported on the NrGO ( Figure S4a) and rGO ( Figure S4c) are exhibited, and Figure S4b  To investigate electrochemical performance of the catalysts, we carried out the ORR analyses of the Pd4.7Ru NPs/NrGO, Pd4.5Ru NPs/rGO, and Pt/CB with a three-electrode system in O2 gas saturated 0.1 M KOH electrolyte. Figure 5a shows the LSV measurement of cathodic scanning plots. Among the catalysts, the Pd4.7Ru NPs/NrGO achieves the highest onset potential at −0.1 mA/cm 2 and half-wave potential with 0.913 VRHE and 0.792 VRHE, followed by the Pt/CB with 0.908 VRHE and 0.785 VRHE, and Pd4.5Ru NPs/rGO with 0.863 VRHE and 0.713 VRHE, respectively (see Table 2). Additionally, the Pd4.7Ru NPs/NrGO, Pd4.5Ru NPs/rGO, and Pt/CB exhibited the limited current density of −6.33 mA/cm 2 , −5.46 mA/cm 2 , and −5.36 mA/cm 2 at 0.3 VRHE, and the specific activities were also plotted at Figure 5b. These results could be explained by the previous XPS analyses that doped pyridinic nitrogen atoms contribute to increasing the onset potential value of ORR. Additionally, we compared the activity of Pd4.7Ru NPs/NrGO to other previously studied Pd-based catalysts on their onset potential values, which are summarized in Table S3. Among them, nitrogen-doped carbon-supported Pd catalysts show relatively high onset potentials over 0.89 VRHE than undoped catalysts, and it is explained by the role of pyridinic N [36]. Moreover, the Pd4.7Ru catalyst has superior activity to Pd single element catalysts, confirming the alloying effect. We also compared the charge transfer resistance of the catalysts by EIS at 0.3 VRHE. As represented in Figure 5c, three components constitute the Randles circuit: solution resistance Rs, charge transfer resistance Rct, and Cdl [37]. Rs and Rct were measured at 100 kHz and 100 mHz, respectively, and Cdl is shown in form of a semicircle as the  Additionally, we carried out the CV analyses at various scan rates with 5, 10, 25, 50, 100, 150, and 200 mV/s in the potential window between 1.10 V RHE and 1.00 V RHE to identify the ECSA of the catalysts. It could be calculated from the C dl according to the equation: ECSA = C dl /C s . Thus, the ECSA is proportional to the C dl , as the specific capacitance of a flat surface (C s ) is a constant value [38,39]. We plotted the difference of current densities |j a − j c |/2 at 1.05 V RHE against the scan rate, and the slope of the linear trend was C dl [40]. Figure S5a-c exhibits the CV curves of the Pd 4.7 Ru NPs/NrGO, Pd 4.5 Ru NPs/rGO, and Pt/CB, respectively. We plotted the calculated |ja − jc|/2 and C dl at Figure S5d and compared the ECSA of the catalysts. It is found that the Pd 4.7 Ru NPs/NrGO has the highest C dl and ECSA, attributed to the well-dispersed NPs on graphene support, which was shown in the TEM images and corresponding size distribution histograms. This tendency is also coincident with the aforementioned LSV and EIS results, supporting the large active surface area of the Pd 4.7 Ru NPs/NrGO for promoting the electrocatalytic reaction. Table 2. Onset potentials, half-wave potential, and current densities of the Pd4.7Ru NPs/NrGO, Pd4.5Ru NPs/rGO, and Pt/CB, determined from LSV at initial and after ADT. Rct values of the Pd4.7Ru NPs/NrGO, Pd4.5Ru NPs/rGO, and Pt/CB, determined from EIS at initial and after CV ADT.   To observe the durability of the catalyst, we conducted an ADT by repeating CV for 1000 cycles from 0 V RHE to 1.2 V RHE at a scan rate of 50 mV/s. Figure 5d,e represent the LSV plots and corresponding specific activity histograms of the Pd 4.7 Ru NPs/NrGO, Pd 4.5 Ru NPs/rGO, and Pt/CB after ADT. The onset potential was shifted −43 mV and −76 mV for the Pd 4.7 Ru NPs/NrGO and Pd 4.5 Ru NPs/rGO, respectively, also achieving −6.30 mA/cm 2 and −4.49 mA/cm 2 at 0.3 V RHE . Furthermore, EIS results in Figure 5f, also demonstrate that R ct of the Pd 4.7 Ru NPs/NrGO was barely increased (119.2 Ω), whereas the resistance of the Pd 4.5 Ru NPs/rGO and Pt/CB were increased around 6.3 times (2063.0 Ω) and 1.5 times (428.5 Ω), respectively. In conclusion, the incorporation of N atoms in graphene lattice could facilitate enhanced catalytic activity during its repeated redox reaction for a long time.

Sample
To confirm the stability of the catalysts under the constant potential, we conducted a CA test at the saturated current potential with 0.3 V RHE and the initiation potential of ORR with 0.75 V RHE for 15 h. In Figure 5g,h, the NrGO-based catalysts show the retention of 98.3% and 93.7%, which are the highest values among all catalysts, whereas the rGObased catalysts have the lowest value of the retention with 84.2% at 0.3 V RHE and 51.8% at 0.75 V RHE . These results are consistent with the abovementioned CV and EIS graphs and support our opinion that nitrogen introduction can affect the enhancement of catalytic durability. After the durability tests, the surface morphologies of the Pd-Ru alloy catalysts were observed by TEM. As shown in Figure S6a, the Pd 4.7 Ru particles are somewhat aggregated on the NrGO sheet, whereas the cohesive image of the Pd 4.5 Ru particles on the rGO sheet is clearly visible in Figure S6d. After, CA at 0.3 V RHE (Figure S6b,e) and 0.75 V RHE ( Figure S6c,f) also appear the same tendency, indicating that the catalysts deteriorate their initial shape and distribution more during the reaction when supported on rGO than NrGO. Furthermore, we plotted the LSV graphs of the Pd 4.7 Ru NPs/NrGO, Pd 4.5 Ru NPs/rGO, and Pt/CB at Figure S7a-c, respectively, which were conducted under a CH 3 OH and CO atmosphere to estimate their methanol tolerance and CO poisoning. The results confirm stability of the Pd 4.7 Ru NPs/NrGO, showing high retention of its current even under harsh condition, although the current of Pt/CB is decreased.
In general, we confirmed that the Pd 4.7 Ru NPs/NrGO catalyst exhibited prior catalytic performance in ORR to Pd 4.5 Ru NPs/rGO and even Pt/CB, the commercial catalyst. When we compare the performance of the NrGO and rGO-based catalysts intuitively, the result could be explained with the previous XPS analyses that the carbon atoms bonding with dominant pyridinic N are active sites for O 2 adsorption, and they contribute to the high onset potential of ORR by acting as Lewis bases [12,13]. In addition, the N atoms, which have a different number of valence electrons from C, could promote faster electronic conduction and the EIS data supports this effect. Moreover, the well-distributed, small-size particles of the Pd 4.7 Ru NPs/NrGO make their active area broader, generating high current density, and it was proven by the ECSA results. The particles, grown on N anchoring sites, have strong bondings with the carbon support, preventing aggregation during the reaction.

Conclusions
We fabricated Pd 4.7 Ru NPs decorated catalysts supported by NrGO and characterized them with a series of analytical methods. The NrGO was fabricated by thermal annealing of the GO in Ar/NH 3 atmospheres, followed by anchoring 4.1 wt % of Pd 4.7 Ru NPs on the carbon supports with the microwave-assisted method. XPS results displayed pyridinic, graphitic, pyrrolic, and oxidized N, which are doped on graphene sheets, and among them, the pyridinic N exists abundantly, with 36.5% contents in the Pd 4.7 Ru NPs/NrGO catalyst. It facilitates the ORR because the C atoms, bonding with pyridinic N, act as active O 2 adsorption sites. Additionally, doped nitrogen atoms contributed to well-dispersed Pd 4.7 Ru NPs with an average size of 2.4 nm on the NrGO, confirmed by TEM analysis, which might be attributed to its high ECSA. Moreover, the nitrogen atoms could transfer their valence electrons to the NPs. It could decrease charge transfer resistance on the catalyst surface and improve conductivity confirmed from EIS results. These factors promoted the ORR activity in the electrochemical test, showing higher limited current density, onset potential, half-wave potential, and lower charge transfer resistance than the Pd 4.5 Ru NPs/rGO and commercial Pt/CB with the value of −6.33 mA/cm 2 , 0.913 V RHE , 0.792 V RHE , and 115.2 Ω, respectively. Moreover, the Pd 4.7 Ru NPs/NrGO catalyst showed the highest retention among all measured catalysts with 98.3% at 0.3 V RHE and 93.3% at 0.75 V RHE . Furthermore, it maintained initial morphology even after 1000 cycles of CV ADT and 15 h of CA durability test. Consequently, we demonstrated that the NrGO is remarkable carbon-based support with the Pd 4.7 Ru NPs for durable and low costed ORR catalysts as a promising alternative of the Pt/CB.