Nitrogen doped nanoporous graphene: an e ﬃ cient metal-free electrocatalyst for the oxygen reduction reaction †

The oxygen reduction reaction (ORR) is an important half reaction, which occurs at the cathode within a fuel cell and limits their range of applications due to slow electrochemical kinetics. To overcome this issue, electrocatalysts are sought, which need to be an alternative to expensive and unsustainable metallic catalysts. Herein we report the synthesis of nitrogen doped nanoporous graphene (NPG), which is a competitive alternative to currently employed metallic catalysts. The NPG is synthesised through a chemical vapour deposition methodology followed by a chemical functionalization step introducing oxygen functional groups (carbonyl and hydroxyl groups), which is then doped with nitrogen via ortho phenylenediamine (OPDA). The NPG is physiochemically and electrochemically characterised. The NPG demonstrates outstanding electrocatalytic activity towards the ORR in alkaline media proceeding via a favourable 4-electron pathway and is comparable to commercially available platinum – carbon (20%). We demonstrate that the electrochemical activity of the NPG is tailorable such that through increased nitrogen doping the ORR transforms from a 2-electron process to that of the favourable 4-electron process via increasing the proportion of pyridinic nitrogen while the content of graphitic nitrogen remains almost constant. The NPG exhibits excellent electrochemical performance towards the ORR in alkaline media, long-term stability and appropriate methanol crossover as benchmarked to commercialised Pt/C electrodes; this outstanding electrocatalytic activity is related to the high proportion of defects, high porosity and (pyridinic) doping.


Introduction
As the global energy economy reacts to the negative effects of anthropogenic climate change, more sustainable energy systems are being developed. A promising clean energy generation technology involves the use of hydrogen as a fuel source for proton exchange membrane fuel cells (PEMFC), which subsequently generate electrical currents. The key reactions that allow a PEMFC to produce a current are the hydrogen oxidation (HOR) and oxygen reduction reactions (ORR), which occur at the anode and cathode, respectively. Of these two reactions it is the ORR that is the most studied, as it has the most signicant inuence upon the performance and efficiency of PEMFCs. [1][2][3] This is due to the ORR having a large kinetic inhibition, arising due to the strength of the (di)oxygen double bond. 4,5 The reaction mechanism for the ORR is different for acidic and basic electrolytes as described below and generally an efficient catalyst operates via the 4 electron pathway: Acidic media: Attempts to increase the performance of PEMFCs have therefore focused on developing efficient, economical and sustainable new materials to reduce the over-potentials, increase the achievable current density and allow for a 4-electron pathway associated with the ORR. Platinum (Pt), and compounds containing it, are considered the most effective catalysts for ORR allowing the reaction mechanism to occur via a 4-electron pathway. 6,7 This is a consequence of its low (near negligible) binding energy of adsorbates like O 2 and H + . However, the high cost and poor global distribution of platinum limit the attractiveness of its application. 8 Attempts have been made to develop non-precious metal catalysts as cost effective alternatives to Pt. Recently, carbon nanostructures, such as carbon nanotubes, 9,10 graphene 11-13 carbon quantum dots, 14 and mesoporous carbon, 15 which have been doped with heteroatoms (N, S, P, B, I, or F), have shown promising metal free ORR catalysis. 16 Table 1 presents a thorough overview of the surrounding literature. The introduction of heteroatoms increases the electron density near the Fermi level, which induces an increase in the electropositive charge on the adjacent carbon atom, via a polarization effect, thus facilitating the binding of the electronegative O atoms within the electrolyte and making the electropositive carbon atoms the sites responsible for the ORR electrocatalytic activity. 17,18 Of particular interest within this eld has been the utilisation graphene as a carbon material for doping, due to it processing unique and benecial electrochemical properties compared to other carbon-based materials. [19][20][21] whilst, pristine (defect free) graphene possesses poor ORR activity it has been shown that by synthesising mesoporous graphene one can greatly increase graphene's ORR activity. There are several reported methods within the literature for the fabrication of mesoporous graphene, such as; chemical etching, 22,23 photocatalytic reaction, 24 lithography, 25 electron beam 26 and chemical activation. 27 An elegant study by Lin et al. 28 prepared mesoporous nitrogen doped graphene, through pyrolysis of graphene oxide and polyaniline at 1000 C, and observed it to have benecial ORR electrocatalytic activity and lowered the electronegativity of the ORR onset allowing the mechanism to occur via the desirable 4 electron pathway, whilst also increasing the stability and methanol tolerance, compared to undoped graphene. This study clearly highlights the potential for heteroatom doped graphene to be a cost effect alternative to Pt based electrocatalysts, however by it utilising chemically exfoliated graphene oxide as the carbon precursor and implementing a chemical activation method, the resulting graphene based material is prone to restacking and crumpling of its sheets, which limits the stability and electrocatalytic potential of the material. Lin et al. 28 is not alone in their use of graphene oxide as a precursor, the vast majority of the literature utilises a similar fabrication methodology and thusly encounters the same detrimental effects.
In order to address the problem of restacking and crumpling graphene sheets, which is endemic within the literature, this paper presents a facile two-step methodology for the fabrication of nitrogen-doped nanoporous graphene. This involves, producing the nanoporous graphene via a chemical vapour deposition (CVD) method followed by introducing nitrogen groups via pyrolysis using ortho-phenylenediamine as a precursor. Table 1 summarises the current literature devoted to developing heteroatom-doped catalysts, which demonstrates that our approach is a novel methodology to realise nitrogen doped nanoporous graphene with outstanding electrochemical properties. The nitrogen doped nanoporous graphene electrocatalyst does not suffer from the restacking and crumpling problems associated with other heteroatom doped graphene based ORR electrocatalysts. This study therefore signicantly increases the attractiveness of graphene-based electrocatalysts as potential cost-effective alternatives for Pt in PEMFCs.

Preparation of nanoporous graphene
Nanoporous graphene was synthesised by a chemical vapour deposition (CVD) process using methane as the carbon precursor. 29 The reaction was carried out in a tubular quartz reactor (diameter ¼ 5 cm, length ¼ 120 cm) in the presence of hydrogen and nitrogen as the reducing and carrier gas, respectively. The optimum ratio of methane, hydrogen and nitrogen was found to be 4 : 1 : 1. The CVD processes were conducted in a horizontal furnace at 900-1000 C for 5-30 minutes at a rate of 5 C min À1 . Subsequently, the reactor was allowed to naturally cool to room temperature under a nitrogen atmosphere, aer which the nanoporous graphene was collected. The obtained product was puried by reuxing in 20% HCL solution for 24 hours at room temperature. The puried sample was repeatedly washed and neutralised with deionized water and dried at 80 C for 48 hours. The obtained product, typically ca. 2 g was oxidised via functionalization with a 200 mL mixture of concentrated sulfuric and nitric acids (3 : 1 v/v). The mixture was mechanically stirred and sonicated for 10 Table 1 A comparison of heteroatom doped carbon-based materials that were produced via metal free synthesis then explored towards the ORR. Adapted from previous work by Zhang  The commercial Pt/C catalysts are commonly used as references: 20 wt% Pt on Vulcan X72R from ElectroChem, Inc., 20 wt% Pt on carbon black from Alfa Aesar, PANI; polyaniline, GO; graphene oxide, rGO; reduced graphene oxide, -; value unobtainable, 3-MBP-dca; 3-methyl-1-butylpyridine dicyanamide, CMIM-Cl; 3-(3-cyanopropyl)-1-methyl-1H-imidazol-3-iumchloride, GQD; graphene quantum nanodots, SWCNTs; single walled carbon nanotubes, CVD; chemical vapour deposition, DNA; deoxyribose nucleic acid. and 180 minutes, respectively. Next, the functionalised porous graphene was cooled down to room temperature and diluted with 500 mL deionised water. Subsequently, the solution was ltrated, neutralised and dried at 60 C overnight. The oxidised porous graphene is denoted as PG herein.

Nitrogen doped nanoporous graphene (NPG)
Nitrogen doped nanoporous graphene was prepared using orthophenylenediamine (OPDA) as an aromatic nitrogen precursor. Initially, 0.2 g of the as-prepared NPG was ultrasonically dispersed in 200 mL ethanol. Different amounts of OPDA were then gradually added into separate suspensions of: 0.1, 0.2, 1, and 2 g which are hereaer denoted as NPG 10.5, NPG 1-1, NPG 1-5, and NPG 1-10, respectively. The suspensions were further sonicated for 10 minutes in order to dissolve the OPDA. Next, the suspension was mechanically stirred (600 rpm) at room temperature for 6 hours and dried at 80 C in order to allow complete solvent evaporation. The remnant solid was then collected and grinded with a pestle and mortar to obtain a uniform powder. Next, the obtained solid was placed into a quartz boat and pyrolyzed in a tubular furnace at 900 C for 2 hours within a nitrogen atmosphere with a heating rate of 5 C min À1 . The reactor was allowed cooled down to room temperature in the presence of nitrogen and the nal product obtained was the NPG. As a control, oxidised nanoporous graphene was subjected to the same procedure without the addition of OPDA.

Preparation of electrodes
A rotating glassy carbon electrode (RGCE) (diameter: 3 mm) was utilised in all electrochemical investigations. In each test, the RGCE was diligently polished using a microcloth and decreasing sizes of alumina. (0.3 and 0.05 mm alumina slurry). The polished electrode was sequentially washed with deionised water and sonicated in ethanol to remove any residual alumina powder. For each electrocatalyst, 5 mg powder was ultrasonically dispersed in 950 mL of ethanol, ultrapure water (1 : 1 v/v) and 50 mL of Naon solution (0.05 wt%) for 15 minutes. Finally, 5 mL of the as-prepared "ink" electrocatalysts were loaded upon the surface of the RGCE and allowed to dry at room temperature. The catalyst loading for all electrodes was estimated to correspond to ca. 0.35 mg cm À2 .

Physicochemical characterisation
The crystallinity of the samples was investigated by X-ray diffraction (XRD) with a PW 1840 Philips device in the range of 10-90 with the rate of 0.02 (2q/s). The morphology and elemental composition of the nanoporous graphene was examined by eld emission scanning electron microscopy FEI Quanta 650F Environmental (SEM), energy-dispersive X-ray spectroscopy (EDX) and transmission electron microscopy (Tecnai G2 F20 S TWIN HR(S) (TEM), FEI). X-ray photoelectron spectroscopy (XPS) was performed using a hemispherical analyzer supplied by (XPS, SPECS Germany, PHOIBOS 150). The CasaXPS soware (2.3.17PR1.1) was used for deconvolution of the XPS results by subtracting Shireley background. Raman spectra were carried out on an Almega Thermo Nicolet with an Ar ion laser source and excitation of 532 nm. The Brunauer-Emmett-Teller (BET) and the Barrett-Joyner-Halenda (BJH) methods were used for the estimation of the surface area and pores size distribution of the nanoporous graphene. The BET surface area, pore volume, and pore size distribution were assessed by nitrogen adsorption-desorption isotherms at 77 K utilizing an ASAP-2010 porosimeter (Micromeritics Corporation, GA).

Electrochemical analysis
Electrochemical analyses were performed using an Autolab potentiostat/galvanostat instrument (PGSTAT30 Eco Chemie, B.V, Netherlands), along with a rotating disk electrode (RDE) system in a conventional three electrode system. The RGCE was used as the working electrode, while Ag/AgCl and Pt wire were used as the reference and counter electrode, respectively. Prior to electrochemical evaluation, including cyclic voltammetry (CV), linear sweep voltammetry (LSV), chronoamperometry (CA), and electrochemical impedance spectroscopy (EIS), of 0.1 M KOH were gently purged with pure oxygen or nitrogen for 30 minutes to get a saturated electrolyte. All potentials in this study are reported versus reversible hydrogen electrode (RHE).

Results and discussion
The overall fabrication process of the nitrogen doped nanoporous graphene (NPG) is summarised in Fig. 1 and detailed within the Experimental section. In the rst step, the oxygen moieties on the nanoporous graphene likely attract the -NH 2 in OPDA through electrostatic interactions, increasing the level of nitrogen doping. 30 The decomposition of the adsorbed OPDA at 900 C may also create new C-N bonds in the nanoporous graphene lattice mainly as pyridinic, pyrrolic, and graphitic. The aromatic structure and the presence of two nitrogen atoms in the OPDA reactant leads to the production of nitrogen doped nanoporous graphene with a high content of nitrogen atoms.

Physiochemical characterisation of the PG and NPG variants
A full independent physicochemical characterisation was performed on the PG and NPG variants, this includes: Raman Spectroscopy, SEM, TEM, XPS and XRD. A brief summary of the characterisation is given below with interested readers directed to the ESI † where a full description of each physicochemical characterisation technique can be found. Fig. 2A and S1 † show the SEM images and corresponding elemental mapping analysis of the PG and NPG 1-5. The 3D morphological structure can be observed to be one of high porosity for the both PG and NPG 1-5. The TEM images of the PG and NPG 1-5 were also taken to further investigate the crumpled, porous, and corrugated structure of PG and NPG 1-5 (see Fig. 2B and C). As presented in the TEM images, the more transparent and apparently fewer layers of the NPG 1-5, compared to PG, might be attributed to the nitrogen doping effect in agreement with the XRD and Raman results; this indicates that the NPG does not suffer from restacking and crumpling problems which is associated with other non-doped and other heteroatom doped graphenes.
The interconnected framework and ultrathin nanosheets of the NPG 1-5 maintain lots of slit-shaped porous structures as the desirable sites for ORR. 31 Simultaneously, the presence of amorphous carbon and highly graphitized structure of the NPG 1-5 can be obviously seen in the TEM images.
The XRD patterns of PG, NPG 1-0.5, NPG 1-1, NPG 1-10 and NPG 1-5 samples are shown within Fig. 2D. The XRD patterns reveal the (002) reection of graphene structure at around 2q $ 29 and 26 for non-doped and nitrogen doped samples, respectively, suggesting that nanoporous graphene exists in single or few  layers. The scattered peaks at ca. 44-45 can be assigned to the (100) crystal plane of graphite, implying the presence of both crystalline and amorphous carbon structures. 32 As observed, the strong peaks of nitrogen doped nanoporous graphene are broadened and shied toward lower angles in comparison with non-doped nanoporous graphene, indicating that the pyrolysis treatment with OPDA moderately disorders the graphene framework, possibly due to the creation of more defects resulting from heteroatom doping effect. 33 The number of graphene sheets can be estimated by calculating the interlayer spacing of the lattices and the mean crystallite size of the powder, which is obtained by the Scherrer equation: 34 where L hkl is the mean crystallite size of powder for the related Miller indices (hkl), l is the wavelength of radiation (1.542Å), b is the full widths at half maximum (FWHM) in radians, and q is the scattering angle. The interlayer spacing of graphene sheets can be calculated by Bragg's law: 34 where d hkl is the interlayer spacing of the lattices for the related Miller indices (hkl) and n represents an integer for the order of the diffraction peak. Accordingly, the number of graphene layers can be estimated trough the following equa- The number of layers, FWHM, L hkl , and d hkl are summarized in Table 2 (main paper) with respect to the XRD spectra.
The number of graphene sheets for PG and NPG 1-0.5 was estimated by calculating the interlayer spacing of the lattices and the mean crystallite size of the powder and found to be 7 and 5, respectively. Whilst the number of graphene layers for NPG 1-1, NPG 1-5, and NPG 1-10 is equal to 4 (see Table 2). The decreasing number of graphene layers can be attributed to increasing the space layer for nitrogen doped nanoporous samples versus un-doped nanoporous graphene.
The augmentation of FWHM quantities of nitrogen doped graphene compared to the pristine nanoporous graphene further veries the successful doping of N atoms into the carbon skeleton.
Furthermore, by increasing the OPDA concentration, the (002) peaks get somewhat broader from NPG 1-0.5 to NPG 1-5 (FWHM in Table 2), signifying that the higher concentration of OPDA leads to the creation of more disorders in the graphene network. However, further increasing the concentration of OPDA to NPG 1-10 reduces the FWHM, suggesting the NPG 1-5 as the optimized ratio for the preparation of nitrogen doped nanoporous graphene.
The Raman spectra of the samples are presented in Fig. 2E. The three expected characteristic peaks are observed at ca. 1350, 1580, and 2685 cm À1 for all samples correspond to D, G, and 2D bands, respectively (see Table 3). The three distinctive peaks at around 1350, 1580, and 2685 cm À1 for all samples correspond to D, G, and 2D bands, respectively ( Table 3). The G band is considered as the rst-order scattering of the E 2g mode of sp 2 carbon atoms in graphene matrix, whereas the D band is related to the breathing mode of A 1g symmetry, indicating the presence of defects, microstructure disordering and amorphous carbon. 35 Commonly, the I D /I G reveals the information about the relation of the sp 3 hybridized carbon (disordered structures) to sp 2 hybridized carbon (graphitic structures). It should be taken into account that the higher ratio of I D /I G signies the higher in plane and edge defects emanating from doping, functionalization and exfoliation of graphene. 36 Obviously, the intensity of D and G band of PG is approximately equal, suggesting the presence of crystallite and amorphous structures in nanoporous graphene at the same time. The I D /I G ratio of PG is about 0.93, while doping nitrogen increases the amount of I D /I G ratio to 1.1, 1.23, 1.42, and 1.54 for NPG 1-0.5, NPG 1-1, NPG 1-10, and NPG 1-5, respectively ( Table 3). The augmentation of I D / I G can be attributed to the disruption of aromatic p-p electrons in the graphene matrix conrming successful nitrogen doping procedure. The I D /I G ratio increases by increasing the concentration of OPDA in the ratio of (1 : 5), despite higher concentrations of OPDA (1 : 10) reducing the of I D /I G ratio. This indicates NPG 1-5 is the best proportion. Apart from that, the shape and intensity of the 2D bands are known as one of the most prominent features of Raman spectra for identifying the number of layers in two-dimension carbon nanostructures. As the 2D band gets broader and positively shied toward higher wave number, the number of graphene layers increases. 37 The 2D bands of nitrogen doped porous graphene are partially sharper than that of non-doped porous graphene (Fig. 2E), indicating that the number of graphene layers can be decreased via nitrogen doping. However, as the XRD results showed, there is no signicant alteration between the number of graphene layers before and aer nitrogen doping. Also, the intangible augmentation of the I 2D by introducing nitrogen atoms (Table 3) leads to the increase in the I 2D /I G ratio, which further veries the diminishing the number of graphene sheets. It can be concluded that the nitrogen doped samples are composed of few-layered graphene as well as sufficient defects as the desirable sites for oxygen reduction reaction.
XPS was performed to investigate the elemental composition of the electrocatalysts and verify the degree and type of nitrogen doping. The XPS surveys of porous graphene presented in Fig. S2 and S3 † shows two distinctive peaks at 285 and 531 eV corresponding to C 1s and O 1s, respectively, without any other impurities. In the NPG spectra, the appearance of the characteristic peaks at about 400 eV conrms the successful doping of nitrogen atoms in graphene matrix. Last, BET surface area (m 2 g À1 ), pore volume (cm 3 g À1 ) and mean pore size (nm) for PG sample was determined to correspond to 721.3, 2.3 and 12.6, respectively (see ESI and Fig. S4 †). Also, the BET surface area (m 2 g À1 ), pore volume (cm 3 g À1 ) and mean pore size (nm) for NPG 1-5 sample is 665.2, 1.8 and 13.7, respectively. In addition to BET surface area, the pore volume and pore size of the PG and NPG 1-5 remains approximately unchanged, indicating that nitrogen doping does not principally deteriorate the mesoporous structure of porous graphene, which is in good agreement with TEM images and Raman spectra. Thus, the PG and NPG variants utilised throughout this study have been fully characterised and revealed to be of high quality and can be accurately described as such.

Electrochemical evaluation of ORR performance
Initially it was vital to assess the intrinsic electrochemical response of the NPG variants in N 2 saturated 0.1 M KOH (50 mV s À1 ) to ensure that there was no visible redox peaks within the potential region where the O 2 reduction peak is expected, as this could result in convolution of the observed signal output. Fig. S5 Fig. 3 for comparative purposes. As stated above the ORR onset potential for the NPG 1-5 is more electropositive than the bare/ unmodied PG and comparable to that of Pt/C 20%. Upon inspection of this gure it is also evident that NPG 1-5 has a larger peak potential than either the bare/unmodied PG or Pt/C 20%. This supports the interference that N doping the PG increases its catalytic activity towards the ORR. In order to assess the limiting current and whether there is a change in the ORR mechanism associated with the observed increase in the ORR, we employed rotating disk electrochemistry as is common within the literature. This was performed by carrying out LSV in O 2 saturated 0.1 M KOH at scan rate of 5 mV s À1 , utilising various rotation rates. The LSV plots at the rotation rates from 250 to 3500 rpm show that by increasing the rotation rate, the limiting current increases as expected (see Fig. S6, † with a comparison of the LSVs at 1500 rpm being given in Fig. 4b). As presented, for all of the nitrogen doped samples, the ORR  . In order to deduce the ORR mechanism the number of electron transfers, n, participating in the electrochemical reaction need to be deduced. A commonly employed approach is to calculate this via rotating disc analysis via the Koutecky-Levich equation: 10-12,38 In these equations, J is the experimental current density, J K is the kinetic-limiting current density, J d is the diffusion-limiting current density, u is the rotation speed (rpm), n is the electron transfer number in ORR, F is the Faraday constant (96 485 C mol À1 ), D is the diffusion coefficient of O 2 in 0.1 M KOH solution (1.9 Â 10 À5 cm 2 s À1 ), y is the electrolyte velocity (0.01 cm 2 s À1 ) and C 0 is the bulk concentration of O 2 (1.2 Â 10 À6 mol cm À3 ). The Koutecky-Levich (K-L) plots of J À1 versus u À1 at various potentials (RHE) from RDE measurements exhibit straight and parallel lines, implying the rst-order reaction (See Fig. S7 †). Analysis of the K-L slope plots for the PG suggest that ORR mechanism occurs via a two electron pathway, whilst the nitrogen doped samples where observed to have an ORR mechanism that occurred by the desirable fourelectron pathway, suggesting the almost complete O 2 reduction to OH À . The exact number electrons involved in the PG and NPG variants reaction mechanism are displayed within Fig. 4a and Table 5. It can be inferred that the doping of PG with N increases the ORR activity by enabling the reaction mechanism to occur by the desirable 4 electron pathway.
The obtained results suggest that the high electrocatalytic activity of the NPG can be attributed to the degree and type of nitrogen doping. Whilst the detailed role of different nitrogen containing phases is still unclear, 39 with various studies within the literature offering alternative theories. For example Lai et al. 40 reported that the graphitic nitrogen phase determines the limiting current density, while the pyridinic phase has an undeniable role in onset potential improvement. 40 Contrary to this, it has been reported that the high electrocatalytic activity of nitrogen doped carbon nanostructures can be attributed to pyrrolic nitrogen content. 39 It was therefore important to assess the surface groups present on the NPG samples to determine their involvement in observed ORR activity. It was initially theorised as the incorporation of OPDA increased the percentage of nitrogen groups present on the graphene's surface would increase, however, according to the XPS results, the coverage of graphitic nitrogen remained almost unchanged with increasing usage of OPDA. This suggests that the electocatalytic activity can be much more affected by pyridinic nitrogen as the favourable active sites for ORR rather than the graphitic nitrogen. The pyridinic percentage in NPG 1-5 is higher than other N-doped electrocatalysts, possibly explaining why it displayed the most benecial ORR activity. The pyridinic nitrogen doping into graphene network can increase the 2pp state of C-N, as well as, weaken the C-C electron density. 41 Moreover, the increase in N/C ratio (Table 4) would alter the electronic density of state close to the Fermi level, which enhances the adsorption of O 2 and facilitates the electron transfer between catalyst and oxygen molecules. 17,41 Thus, the best electrocatalytic performance of NPG 1-5 among the other

Stability and methanol tolerance of the NPG variants
The tolerance of the electrocatalysts towards methanol crossover was evaluated by CV and chronoamperometry (CA). Fig. S5 † shows the typical CV response obtained for the NPG variants and Pt/C 20% when 3 M methanol was introduced to an oxygenated solution of 0.1 M KOH. It is clear that the addition of methanol does not signicantly alter the CV proles on the nitrogen doped samples and PG, implying promising selectivity toward ORR against methanol oxidation, whereas methanol oxidation leads to the appearance of large anodic peak at Pt/C electrode showing the poisoning effect upon fuel crossover. The stability of the electrocatalysts upon addition of methanol was further conrmed by CA measurements through applying a constant potential (1.0 V vs. RHE) and a rotation of 1500 rpm, in O 2 saturated electrolyte, for a duration of 1000 seconds (see Fig. S8 †). An appropriate stability, as well as an excellent methanol tolerance of the nitrogen doped samples, can be observed. The loss of current density of the nitrogen doped samples was estimated to be less than 15% aer addition of 3 M methanol at 500 th second. In the case of PG, the loss of current density is more than nitrogen doped samples, accentuating the role of nitrogen atoms and porous structure on the stability of the electrocatalysts for ORR. According to Fig. S8f, † the Pt/C electrocatalyst shows a signicant current decline of about 73% compared to nitrogen doped, representing a rapid poisoning of the platinum active sites, with minimal restoration of the current density over the remaining duration. The longterm durability of the electrocatalysts were examined by exposing the electrode to equal conditions for ca. 27 hours. As observed in Fig. S9(a), † the nitrogen doped samples display remarkable stability over the duration of the experiment, this is due to the slower attenuation of current density compared to PG and Pt/C. The CA responses and the corresponding normalised current is shown in Fig. S9(b). † The normalized current also indicates the higher stability of the nitrogen doped electrocatalysts compared to Pt/C. The results show that the asprepared metal free catalysts not only possess high selectivity toward oxygen reduction, but also have outstanding stability in comparison to Pt/C. Electrochemical impedance spectroscopy (EIS) was employed to further elucidate the electrocatalytic performance of the samples. Nyquist plots of the samples and Pt/C at different potential are presented in Fig. S10. † The Nyquist curves show an arc-like or semicircle prole with different diameter corresponding to the charge transfer resistance (R ct ). For all of the nitrogen doped samples and Pt/C, the diameter of the semicircles shrinks by decreasing the applied potential to +0.87 V (vs. RHE) and slightly increases at +0.82 V, indicating higher electrical conductivity in the vicinity of +0.865 V (vs. RHE), which is close to the peak potential in cyclic voltammetry. A similar trend does is not observed for the non-doped porous graphene, further highlighting the role of nitrogen atoms and high specic surface area, which facilitates electron transfer in the conductive channels of NPG. The comparison of Nyquist plots at +0.87 V (vs. RHE) and the corresponding equivalent circuit are shown in Fig. S11. † Based on the equivalent circuit tting, the R ct of PG, NPG 1-0.5, NPG 1-1, NPG 1-10, NPG 1-5, and Pt/C were calculated to be 2335, 283, 255, 212, 155, and 496 U, respectively. The semicircle diameter of NPG 1-5 is slightly smaller than other samples. The R ct of all nitrogen doped samples are noticeably lower than that of PG and Pt/C, suggesting improved electron and ion transfer between electrode and electrolyte. Prior literature has demonstrated that the nitrogen atoms, especially pyridinic groups, next to the carbon atoms provide benecial electrochemical/electrocatalytic sites due to their localised density of states in the occupied region near the Fermi level. 41 Moreover, density functional theory (DFT) indicates that carbon atoms in the vicinity of the pyridinic nitrogen can act as Lewis bases because of its tendency to donate the electron pair. 42 Therefore, in accordance to the XPS results, the high concentration of pyridinic nitrogen's of the electrocatalysts (i.e. NPG 1-5 and NPG 1-10, Table 5) result in enhanced electron transport and reduced charge transfer resistance.
In summary, this study has shown that by producing nitrogen doped nanoporous graphene, via a CVD method followed by introducing nitrogen groups via pyrolysis using orthophenylenediamine as a precursor, a stable and cost effective alternative to the Pt/C can be fabricated for use as a cathodic material in a PEMFC.

Conclusions
NPG electrocatalysts have been successfully prepared and are shown to be useful electrocatalysts towards the ORR with comparable electrocatalytic activity to a commercially available Pt/C. Table 1 provides a thorough literature overview of nitrogen-doped electrocatalysts, where it can be seen that our fabrication approach is novel with the electrochemical performance demonstrated to be at least comparable, if not better than those reported within Table 1. The outstanding electrocatalytic activity of the nitrogen doped electrocatalysts are due to its high surface area, high nitrogen content, large amount of defects and a nanoporous structure; the NPG does not suffer from restacking and crumpling problems. The homogenously distributed N-species not only provide numerous ORR active sites, including graphitic and pyridinic phases, but also contribute to the highly disordered and nanoporous structure, all of which enhances electron transfer processes between the NPG and oxygen. Furthermore, the long-term stability and methanol crossover effect of the optimised electrocatalyst (16% signal output loss) towards the ORR is optimal over that of Pt/C (73% signal output loss). In summary, the prepared electrocatalysts have potential as an excellent alternative to precious metal based catalysts for oxygen reduction reaction.

Conflicts of interest
There are no conicts to declare.