Journal of Catalysis 314 (2014) 66–72 Contents lists available at ScienceDirect Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat N-doped graphene as catalysts for oxygen reduction and oxygen evolution reactions: Theoretical considerations Mingtao Li a,b, Lipeng Zhang a, Quan Xu a, Jianbing Niu a, Zhenhai Xia a,c,⇑ a Department of Materials Science and Engineering, University of North Texas, Denton, TX 76203, USA International Research Center for Renewable Energy, State Key Laboratory of Multiphase Flow in Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, PR China c Department of Chemistry, University of North Texas, Denton, TX 76203, USA b a r t i c l e i n f o Article history: Received 7 December 2013 Revised 11 March 2014 Accepted 19 March 2014 Available online 26 April 2014 Keywords: N-doped graphene Oxygen evolution reaction Oxygen reduction reaction Catalysts Fuel cells First principles calculation a b s t r a c t Electrocatalysts are essential to two key electrochemical reactions, oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) in renewable energy conversion and storage technologies such as regenerative fuel cells and rechargeable metal–air batteries. Here, we explored N-doped graphene as costeffective electrocatalysts for these key reactions by employing density functional theory (DFT). The results show that the substitution of carbon at graphene edge by nitrogen results in the best performance in terms of overpotentials. For armchair nanoribbons, the lowest OER and ORR overpotentials were estimated to be 0.405 V and 0.445 V, respectively, which are comparable to those for Pt-containing catalysts. OER and ORR with the minimum overpotentials can occur near the edge on the same structure but different sites. These calculations suggest that engineering the edge structures of the graphene can increase the efficiency of the N-doped graphene as efficient OER/ORR electrocatalysts for metal–air batteries, water splitting, and regenerative fuel cells. Ó 2014 Elsevier Inc. All rights reserved. 1. Introduction The oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) play pivotal roles in various renewable energy conversion and storage technologies, such as fuel cells [1], direct solar driven water splitting [2] hydrogen production from water electrolysis [3], rechargeable metal–air batteries [4], and regenerative fuel cells [5]. The current bottleneck of fuel cells lies in the sluggish ORR on the cathode side, while OER, a reverse reaction of ORR, plays an important role in efficiency of energy storage such as direct solar driven water splitting systems, Li–air batteries and regenerative fuel cells. Extensive efforts have been devoted to the development and understanding of electrocatalysts for OER/ORR. To date, platinum and its alloy have been used as catalysts for ORR [6] while platinum and metal oxides/hydroxides of iridium, ruthenium, nickel and other metals for OER [7]. However, the OER/ORR is sluggish, even facilitated by these catalysts. Furthermore, the limited resources and high cost of the platinum catalysts has been shown to be the major ‘‘showstopper’’ to mass market fuel cells for commercial applications. Even though the amount of platinum needed for desired catalytic effect could be reduced, ⇑ Corresponding author at: Department of Materials Science and Engineering, University of North Texas, Denton, TX 76203, USA. Email: [email protected] E-mail address: [email protected] (Z. Xia). http://dx.doi.org/10.1016/j.jcat.2014.03.011 0021-9517/Ó 2014 Elsevier Inc. All rights reserved. the commercial mass production will still require large amount of platinum. Therefore, it is highly desirable to develop low-cost OER and ORR catalysts. In particular, bifunctional OER/ORR catalysts are needed for rechargeable metal–air batteries and regenerative fuel cells. Graphene, a two-dimensional one-atom-thick single layer of graphite, has attracted intense scientific and technological interest since its freestanding form was isolated in 2004 [8]. With its unique properties such as huge surface area [9], high mobility of charge carriers [10], excellent mechanical properties [11], and superior thermal conductivity [12], graphene could be an excellent materials candidate for energy conversion and storage such as batteries [13], supercapacitors [14], and fuel cells [15,16]. Nitrogendoped graphene has been demonstrated as an excellent catalyst for electrochemical reactions, such as oxygen reduction [17,18] and hydroperoxide oxidation [19,20] due to their unique electronic properties derived from the conjugation between the nitrogen lone-pair electrons and the graphene p system. N-doped graphene as bifunctional catalysts for OER and ORR has also been demonstrated recently [21,22]. These carbon-based electrocatalysts have the potential to replace noble metals for energy applications, particularly for Li–air batteries and regenerative fuel cells [15]. Understanding of the OER/ORR mechanisms of N-doped graphene could provide design guidelines for materials and process development, and discovery of new catalysts. Many 67 M. Li et al. / Journal of Catalysis 314 (2014) 66–72 of 50 meV. All calculations were spin polarized and were done until the force of the system converged to about 0.02 eV/Å. The lattice a was optimized previously by fitting the energy-volume relationship with Murnaghan Equation of State for further absorption and reaction free energy calculations. All structures were fully relaxed. Bader charge analysis was performed to calculate the effective charge for these N-doped graphene nanoribbons [31]. The formation energy for a nitrogen dopant, Ef, was calculated by Ef ¼ ENGNR þ mlC ðEGNR þ mlN Þ Fig. 1. (a) Armchair and (b) Zigzag N-doped graphene structures used in the calculations. The numbers denote substitutional sites and reaction sites. Symbols a, b, c, d, e, and f denote reaction sites apart from 1, 2, 3, 4, 5, and 6. researchers have been devoted to theoretical studies to understand the high ORR activity in N-doped carbon-based materials. Based on DFT investigations, Okamoto suggested that the complete fourelectron ORR activation can be achieved with an adequate binding energy of the oxygen atoms on multiple N-doped graphitic sites [23]. Zhang et al. found that the ORR over N-doped graphene is a four-electron pathway but pure graphene does not have such catalytic activities. They also concluded that the catalytic active sites on the N-doped graphene depend on spin density distribution and atomic charge distribution on the neighboring carbon atoms to the doping sites [17,24]. Kim et al. proposed that the ORR activation initiates at the outermost graphitic nitrogen site with the first reduction step, the latter graphitic nitrogen becomes pyridinic-like in the next reduction steps via the ring-opening of a cyclic C–N bond [25]. However, to our knowledge, little work is done on single OER or bifunctional OER/ORR over N-doped graphene. In this work, we studied OER/ORR mechanism of N-doped armchair and zigzag graphene nanoribbons by using density functional theory (DFT) calculations, in an effort to understand how ORR/OER occur over N-doped graphene, and provide design principles how the doping structure can be translated into enhanced performance of electrocatalysts for renewable energy conversion and storage. 2. Computational models and methods First-principle calculations were performed within the framework of density functional theory (DFT) as implemented in the plane wave set Vienna ab initio Simulation Package (VASP) code [26,27]. Nuclei–electron interactions were described by the projector augmented wave (PAW) pseudo-potentials [28,29], while the electronic exchange and correlation effects were described within the generalized gradient approximation (GGA) as parameterized by Perdew, Burke, and Ernzerhof [30]. The C-2s22p2, O-2s22p4, H-1s1 were treated as valence electrons. All nanoribbons were modeled as three-dimensional periodic structures, where a rectangle periodic box was represented with a dashed line in Fig. 1 and vacuum layers were set around 14 and 18 Å in the y- and z-directions, respectively to avoid interaction between slabs. The k-point sampling of the Brillioun zone was obtained using a 4 1 1 grid centered at the gamma (C) point using Monkhorst Pack Scheme. The plane wave basis set with a high cut off energy of 450 eV was used throughout the computations. The k point meshes and energy cut off were chosen to ensure that the energies were converged within 1 meV/per atom. The Fermi level was slightly broadened using a Fermi–Dirac smearing ð1Þ where ENGNR is the total energy of the supercell with N-doped graphene nanoribbon, and EGNR is the total energy of the pristine graphene nanoribbon. lC is the chemical potential of carbon defined as the total energy of graphene per carbon atom. lN is the chemical potential of N taken as one-half the total energy of N2 molecule. And m is the number of nitrogen atoms in the model. 3. Results and discussion We developed a series of models for graphene nanoribbons with a single nitrogen dopant at different sites, as shown in Fig. 1. For each N-doped structure, we consider all the sites at which ORR/OER could occur. These models are labeled as Ax–y, or Zx–y, where A and Z refer to armchair and zigzag graphene, respectively, and symbols x and y denote substitutional sites of nitrogen atoms and reaction sites of ORR/OER, respectively. Also, we use x = h to denote pristine graphene nanoribbons. For example, Model A1–2 (Fig. 1a) represents an armchair graphene nanoribbon with a nitrogen dopant on site 1, and reactions occur on site 2. The location of nitrogen dopants is important to OER/ORR activities. In order to identify efficiently catalytic sites of N-doped graphene for OER/ORR, we began with calculations of formation energies Ef of nitrogen dopants. The calculated formation energy and minimum C–N bond length are plotted as a function of the distance (de) from the dopants to the edge, as shown in Fig. 2. It can be seen that the formation energy is relatively low when the dopant locates at the edge, but it increases and becomes constant with increasing the distance de. Thus, nitrogen atoms prefer to substitute the carbon atoms near the edge. Among all the models, Model Z1 is the most stable one from the formation energy viewpoint, which is consistent with the results reported in Ref. [32]. The minC—N imum C–N bond length dmin also varies with regard to the distance de. Besides, the N doping changes the C–C bond length due to larger electronegativity of nitrogen atoms. From the general trends of the formation energy and bond length, shown in Fig. 2, the changes in C—N the formation energy Ef and minimum C–N bond length dmin are C—N Fig. 2. Formation energy Ef and minimum C–N bond length dmin as a function of the distance de from the nitrogen dopants to the edge of the graphene nanoribbons. 68 M. Li et al. / Journal of Catalysis 314 (2014) 66–72 correlated with each other, and an edge effect exists within the range of 2.5 Å from the edge. Such edge effect may have pronounced influence on catalytic activities of the N-doped graphene. The ORR and OER activities on active sites of N-doped graphene were studied in detail. In acidic environment, OER could occur over N-doped graphene in the following four electron reaction paths, adsorption free energy of O, OH and OOH. The absorption energies were calculated as follows [35], H2 OðlÞ þ ! OH þ ðHþ þ e Þ DEO ¼ EðO Þ EðÞ EH2 O EH2 ð2Þ OH ! O þ ðHþ þ e Þ ð3Þ O þ H2 OðlÞ ! OOH þ ðHþ þ e Þ ð4Þ OOH ! þ O2 ðgÞ þ ðHþ þ e Þ ð5Þ where stands for an active site on the graphene surface, (l) and (g) refer to gas and liquid phases, respectively, and O, OH and OOH are adsorbed intermediates. The ORR can proceed incompletely through a two-step two-electron pathway that reduces O2 to hydrogen peroxide, H2O2, or completely via a direct four-electron process in which O2 is reduced directly to water, H2O, without involvement of hydrogen peroxide. Here, we study the complete reduction cycle because the previous results showed that the ORR proceeds on Ndoped graphene through the four-electron mechanism [17]. The ORR mechanism is summarized using the following elementary steps [17], O2 þ ðHþ þ e Þ ! OOH ð6Þ OOH þ ðHþ þ e Þ ! O þ H2 OðlÞ ð7aÞ OOH þ ðHþ þ e Þ ! OH þ OH ð7bÞ O þ ðHþ þ e Þ ! OH ð8aÞ OH þ OH þ ðHþ þ e Þ ! OH þ H2 OðlÞ ð8bÞ OH þ ðHþ þ e Þ ! þ H2 OðlÞ ð9Þ There are two branching paths for the 2nd and 3rd steps, where OOH is reduced to O and H2O in path (a) defined in Eq. (7a), or to 2OH in path (b) defined in Eq. (7b). Both paths lead to the same final products in the third step as shown in Eqs. (8a) and (8b). These reactions (6), (7a), (8a), and (9) for ORR are inversed from the reactions (2)–(5) for OER. In OER, the potential-determining steps can either be the formation of O from OH (Eq. (3)) or the transformation of O to OOH (Eq. (4)) [33]. However, in ORR, it was reported that the rate determining steps can either be the adsorption of O2 as OOH (Eq. (6)) or the desorption of OH as water (Eq. (9)) [33]. Here, we took reactions (2)–(5) to derive the thermochemistry of both OER and ORR, because the reactions (6), (7a), (8a), and (9) are inversed from reactions (2)–(5). The overpotentials of the ORR/OER processes can be determined by examining the reaction free-energies of the different elementary steps. The thermochemistry of these electrochemical reactions was obtained by using DFT calculations in conjunction with SHE model developed by Nørskov and co-workers [34,35]. This thermodynamic approach establishes a minimum set of requirements for the reactions based on the binding of the intermediates and the assumption that there are no extra barriers from adsorption/dissociation of O2 or proton/electron transfer reactions. In our calculations, the OER and ORR were analyzed using intermediate species associated with one electron transfer at a time, which is energetically more favorable than the simultaneous transfer of more than one electron. In order to obtain the rate limiting step of OER and ORR on different sites for different model structures, we calculated the DEOH ¼ EðOH Þ EðÞ EH2 O 1=2EH2 DEOOH ¼ EðOOH Þ EðÞ 2EH2 O 3=2EH2 ð10Þ ð11Þ ð12Þ in which, E( ), E(OH ), E(O ), and E(OOH ) are the ground state energies of a clean surface and surfaces adsorbed with OH, O, and OOH, respectively. EH2 O and EH2 are the calculated DFT energies of H2O and H2 molecules in the gas phase using the approaches outlined by Nørskov et al. [35]. Also, we considered the ZPE and entropy corrections here. These calculations transform DFT binding energies, DEDFT ads , into free energies of adsorption, DGads, by the following equation [35], DGads ¼ DEDFT ads þ DZPE T DS ð13Þ where T is the temperature and DS is the entropy change. For the zero-point energy (ZPE), the vibrational frequencies of adsorbed species (O, OH, and OOH) were calculated with the N-doped graphene nanoribbons fixed to obtain ZPE contribution in the free energy expression. Moreover, only vibration entropy contributions were considered for adsorbates and total entropies for solvent molecules were taken from standard thermodynamic tables (see the Supporting information). In present study, we did not use any solvent corrections to the adsorbed species. Fig. 3(a) shows the adsorption free energies of OOH and OH for various N-doped structures. The free energies of OOH are linearly related to that of OH by y = x + 3.177 with a constant of approximate 3.177 eV, independent of the binding strength to the surface. The slope of unity in the linear fit is motivated by the single bond between O and the carbon of N-doped graphene nanoribbons for both OH and OOH, which is very similar to that on the surface of ABO3 perovskite [29]. The constant energy difference between the binding energies of OH and OOH implies that there is a scaling relation between OH and OOH. Fig. 3(b) shows the adsorption free energies of OH and O for various N-doped structures. These is also a scaling relation with a slope of 0.5 between OH and O except some sites where oxygen atom is adsorbed in bridge-on mode, such as A2–1, Z1–2, Z2–1 and Ah–1. Due to the scaling relation between OH and OOH, the total reaction enthalpy differences for reaction (3), (4), (7a), and (8a) should be a constant, which results in a lower limit over potential of OER and ORR as described below. For each step, the reaction free energy DG is defined as the difference between free energies of the initial and final states and is given by the expression [36], DG ¼ DE þ DZPE T DS þ DGU þ DGpH ð14Þ where DE is the reaction energy of reactant and product molecules adsorbed on catalyst surface, obtained from DFT calculations, DGU = eU, where U is the potential at the electrode, and e is the charge transferred. DGpH is the correction of the H+ free energy by the concentration dependence of the entropy: DGpH ¼ kB T ln½Hþ ð15Þ The free energy of reaction (2)–(5) can be calculated using Eq. (14). For the OER reactions, Nørskov et al. developed a method to determine the overpotentials [35], GOER ¼ maxfDG1 ; DG2 ; DG3 ; DG4 g ð16Þ gOER ¼ GOER =e-1:23 V ð17Þ where DG1, DG2, DG3, and DG4 are the free energy of reaction (2)– (5), respectively. An ideal catalyst should be able to facilitate water 69 M. Li et al. / Journal of Catalysis 314 (2014) 66–72 (a) (b) Fig. 3. (a) Adsorption energies of OOH versus adsorption energies of OH and (b) adsorption energies of OH versus adsorption energies of O on different sites of armchair and zigzag graphene nanoribbons. oxidation just above the equilibrium potential, but requires all the four charge transfer steps to have reaction free energies of the same magnitude at zero potential (i.e., 4.92 eV/4 = 1.23 eV). This is equivalent to all the reaction free energies being zero at the equilibrium potential, 1.23 V. Since there is a scaling relation between OH and OOH, this set a constraint on DG2 and DG3, i.e., DG2 + DG3 = DGOOH* DGOH* = 3.177 eV, resulting a lower limit of OER overpotential. When DG2 = DG3 = (GOOH* GOH*)/2 = 1.589 eV, the OER overpotential has the minimum value, gOER limit ¼ 1:589—1:23 ¼ 0:359 V. For ORR, there is also a lower limit of the overpotential. The overall free energy of reaction (2)–(5) is 4.92 eV, leading to DG1 + DG2 + DG3 + DG4 = 4.92 eV. Since DG2 + DG3 = constant, we derived that DG1 + DG4 = constant. When DG1 = DG4 = 0.871 eV, the ORR overpotential has its minimum value, gOER limit ¼ 1:23—0:871 ¼ 0:359 V. To derive the minimum overpotential in N-doped graphene systems, we calculated the overpotenials for different reaction sites on different structures employing a descriptor DG0O DG0HO : Fig. 4(a) shows the volcano plot, i.e., overpotential gOER versus the descriptor for various reaction sites on armchair and zigzag graphene structures. From this theoretical analysis, Model A2–3 is identified to have a minimum OER overpotential (gOER min ¼ 0:405 V). For the ORR activity, previous results showed that the pathway (6), (7a), (8a), and (9) had close reaction enthypl distribution for each steps [17], which would result in small overall ORR overpotentail. Using similar methods described above, we calculated the overpotenials of this ORR pathway gORR for various reaction sites on armchair/ zigzag graphene structures. A volcano plot was made using the descriptor DG0OH . As shown in Fig. 4(b), among the structures studied, Model A2–1 has the lowest ORR overpotential, which was estimated to be 0.445 V. These values of the overpotential for ORR and OER are comparable to those of Pt containing catalysts (0.4 V for OER on PtO2-rutile and 0.45 V for ORR on Pt [34]), indicating that (a) N-doped graphene as bifunctional catalysts may have as good performance as its counterparts. To determine at what condition the OER or ORR can spontaneously occur, we calculated the free energy under different electrode potentials U. Fig. 5 shows the diagrams of OER substeps on reaction site A2–3 and ORR substeps on reaction site A2–1. For sites on A2–3, the OER is uphill when the electrode potential is 0 V. At U = 1.23 V, an ideal water splitting potential, the transformation of OOH to O2 becomes downhill, but the reactions (2)–(4) are still uphill. Only when the potential increases to 1.635 V (i.e., 0.405 V in overpotentail), can all the elementary reaction steps become downhill. So, 1.635–1.23 = 0.405 V is the overpotentail for this reaction site, and the transformation of O to OOH is the rate determination step. Since the OER overpotential is reduced by nitrogen doping, the above OER is facilitated overall by the Ndoped graphene. For ORR on A2–1, when the electrode potential is 0 V, the ORR substeps (6), (7a), (8a), and (9) are all downhill, corresponding to a short circuit condition of fuel cells. As the electrode potential increased to 1.23 V, the reaction steps (7a), (8a), and (9) are all uphill, corresponding to an open circuit condition of fuel cells. Since the reaction (6), adsorption of O2 as OOH becomes uphill while other subreactions still keep downhill at U = 0.785 V, the adsorption of O2 as OOH must be the rate determination step in ORR. Thus, the minimum ORR overpotential is 1.23– 0.785 = 0.445 V. The most active sites identified above are attributed to the redistribution of surface charge induced by the incorporation of nitrogen atoms into carbon lattice. As shown in Fig. 6, some carbon atoms become positively charged while others are negative after a nitrogen atom is doped on the graphene. Those carbon atoms with positive effective charge will facilitate the adsorption of some species with negative charges [17]. However, if the adsorption free energy of O is too high due to high positive charge or edge effect, (b) Fig. 4. Volcano plots for (a) OER and (b) ORR on different sites of armchair and zigzag graphene nanoribbons. 70 M. Li et al. / Journal of Catalysis 314 (2014) 66–72 (b) (a) Fig. 5. Free energy diagram for (a) the OER on site 3 for A2 structure (A2–3), and (b) the ORR on site 1 for Z2 structure (Z2–1) at different electrode potential U. Fig. 6. Bader effective charges of (a) A2 and (b) Z2 structures. it in turn becomes a barrier for O to transfer to OOH in OER. For example, sites A1–a, A1–2, A2–1 and Z2–1 have relatively large positive charge (>0.21–0.38) or edge effect (de = 0–1.2 Å), leading to high adsorption energy, and consequently high overpotentials. For site 3 of structure A2 (Fig. 6(a)), with moderate charge (0.27) and edge effect (de = 2.4 Å), the adsorption free energies of O and OOH both have moderate values, which result in low OER overpotentail (Fig. 4). It was found that there is a belt region near the edge (de 2.4 Å) but within the edge effect range identified in Fig. 2, where active sites are generated by nitrogen dopants, resulting in high OER activity because of the moderate edge and charge effect. Similar phenomena were observed in ORR, but the reaction active sites were quite different from those in OER. For most of armchair nanoribbons, the reaction (9) has moderate free energy, but the reaction (6) has the largest free energy, making this the rate determination step. An example is site A2–1, which has relatively high positive charge (0.209) and strong edge effect (de = 0), resulting in the lowest OER overpotential, 0.445 V. Since reaction (6) involves two substeps, the adsorption of O2 on graphene and reaction with a proton, the adsorption of O2 may also be the critical step. We calculated the adsorption O2 on different sites of model A2 and found that O2 can be adsorbed on the edge in side-on mode and nitrogen doping increases the adsorption energy, which means nitrogen doping can promote the adsorption process. An exception is site A1–a. Although the adsorption of O2 is relative easy on this site (in side-on mode), the overall potential is still larger than that of A2–1, due to a very small free energy of the reaction (9), making it the rate determination step. For zigzag graphene, although from the overpotential viewpoint, the site Zh-1 has a relatively low overpotential 0.468 V, a small adsorption energy of O2 makes adsorption O2 inefficient. For site Z2–1 with large positive charge and de = 0 (Fig. 6(b)), the adsorption of O2 and the subsequent transformation of OH to H2O are easy on this site and the calculated overpotential is relative lower than those of other zigzag graphene nanoribbon cases. However, the reaction (7a), transformation OOH to O, involves a ring-opening of a C–N bond on this site, which is consistent with the results of Ref. [25]. Even in disso- 71 M. Li et al. / Journal of Catalysis 314 (2014) 66–72 (a) (b) Fig. 7. (a) OER and (b) ORR overpotential verse the distance from edge to N atom. ciative reaction pathway, the dissociation of O2 as 2O involves a ring-opening of a C–N bond, which makes this site inefficient. In all cases, these most active sites for ORR are located at a distance of de = 0. These results suggest that while ORR takes place at the edge of the graphene, OER usually occurs near the edge but within the range of edge effect. Fig. 7 shows the plot of ORR and OER overpotentials versus the distance of N atom from the edge. Generally speaking, the ORR and OER overpotentials for zigzag graphene are larger than those for armchair graphene nanoribbons, but in some cases, the potential for the zigzag can be as small as that for the armchair. In most cases, the low OER/ORR overpotentails can be achieved by doping nitrogen atom near the edge within the distance of edge effect. For ORR, near-edge doping of N makes the adsorption O2 as OOH easier, except some structures such as A1–2, which has a high OER overpotential up to 1.96 V. It should be noted that the large OER overpotential for the armchair graphene nanoribbons with nitrogen atom near edge originates from different oxygen adsorption configuration. For example, the O atom adsorbed near A2–1 in bridge mode rather than end-on mode is unfavorable to the efficient transformation of O to OOH. Overall, edge doping plays an important role in reducing the overpotentail and enhancing OER/ ORR catalytic capability of graphene. Also, under sufficiently high potentials, the nitrogen present in the graphene nanoribbons could be removed as NH3. From the formation energy viewpoint, the structure with the substituting nitrogen atom near the edge is more stable. Since the graphene edge structures could be controlled by various methods [37–41], engineering the edge structure of the graphene could significantly increase the efficiency of the Ndoped graphene as OER/ORR electrocatalysts for energy conversion and storage. 4. Conclusion Oxygen reduction reaction and oxygen evolution reaction on nitrogen doped graphene nanoribbons were analyzed by density functional theory calculations. It was found that there is a linear relation between the binding energy of OOH and OH for these structures. The OER active sites were identified on the armchair nanoribbons at the carbon atoms near the nitrogen atom, while the ORR active sites are those on the edge carbon near the nitrogen atom. The armchair nanoribbons with nitrogen dopants near the edge have the minimum theoretical OER overpotential, which was estimated to be 0.405 V; while the minimum overpotential for ORR was calculated to be 0.445 V. Those values of ORR and OER overpotentials are comparable to those of Pt containing catalysts. These theoretic calculations suggest that N doped graphene nanoribbons are a highly promising OER/ORR catalyst for metal–air batteries, water splitting systems, and regenerative fuel cells. Acknowledgments We thank Air Forces MURI program for the support of this research under the contract #FA9550-12-1-0037, and National Science Foundation (NSF) for the suppoet under the contract: #IIP1343270. Computational resources were provided by UNT high performance computing initiative, a project of academic computing and user services within the UNT computing and information technology center. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcat.2014.03.011. References [1] X. Ji, K.T. Lee, R. Holden, L. Zhang, J. Zhang, G.A. Botton, M. Couillard, L.F. Nazar, Nat. Chem. 2 (2010) 286. [2] H.B. Gray, Nat. Chem. 1 (2009) 7. [3] M.W. Kanan, D.G. Nocera, Science 321 (2008) 1072. [4] Y.-C. Lu, Z. Xu, H.A. Gasteiger, S. Chen, K. Hamad-Schifferli, Y. Shao-Horn, J. Am. Chem. Soc. 132 (2010) 12170. [5] J. Suntivich, K.J. May, H.A. Gasteiger, J.B. Goodenough, Y. Shao-Horn, Science 334 (2011) 1383. [6] H.A. Gasteiger, S.S. Kocha, B. Sompalli, F.T. Wagner, Appl. Catal. B: Environ. 56 (2005) 9. [7] J. Ahn, R. Holze, J. Appl. Electrochem. 22 (1992) 1167. [8] K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, Y. Zhang, S.V. Dubonos, I.V. Grigorieva, A.A. Firsov, Science 306 (2004) 666. [9] C.N.R. Rao, A.K. Sood, K.S. Subrahmanyam, A. Govindaraj, Angew. Chem. Int. Ed. Engl. 48 (2009) 7752. [10] S. Park, R.S. Ruoff, Nat. Nanotechnol. 4 (2009) 217. [11] C. Lee, X. Wei, J.W. Kysar, J. Hone, Science 321 (2008) 385. [12] A.A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao, C.N. Lau, Nano Lett. 8 (2008) 902. [13] B. Lung-Hao Hu, F.-Y. Wu, C.-T. Lin, A.N. Khlobystov, L.-J. Li, Nat. Commun. 4 (2013) 1687. [14] Y. Zhu, S. Murali, M.D. Stoller, K.J. Ganesh, W. Cai, P.J. Ferreira, A. Pirkle, R.M. Wallace, K.A. Cychosz, M. Thommes, et al., Science 332 (2011) 1537. [15] Y. Liang, Y. Li, H. Wang, J. Zhou, J. Wang, T. Regier, H. Dai, Nat. Mater. 10 (2011) 780. [16] E. Yoo, J. Nakamura, H. Zhou, Energy Environ. Sci. 5 (2012) 6928. [17] L. Zhang, Z. Xia, J. Phys. Chem. C 115 (2011) 11170. [18] Q. Liu, H. Zhang, H. Zhong, S. Zhang, S. Chen, Electrochim. Acta 81 (2012) 313. [19] W.-J. Lin, C.-S. Liao, J.-H. Jhang, Y.-C. Tsai, Electrochem. Commun. 11 (2009) 2153. [20] Y. Wang, Y. Shao, D.W. Matson, J. Li, Y. Lin, ACS Nano 4 (2010) 1790. [21] Z. Lin, G.H. Waller, Y. Liu, M. Liu, C. Wong, Carbon 53 (2013) 130. [22] Z. Lin, G.H. Waller, Y. Liu, M. Liu, C. Wong, Nano Energy 2 (2013) 241. [23] Y. Okamoto, Appl. Surf. Sci. 256 (2009) 335. [24] L. Zhang, J. Niu, L. Dai, Z. Xia, Langmuir 28 (2012) 7542. [25] H. Kim, K. Lee, S.I. Woo, Y. Jung, Phys. Chem. Chem. Phys. 13 (2011) 17505. [26] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 6 (1996) 15. [27] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169. [28] P.E. Blöchl, Phys. Rev. B 50 (1994) 17953. [29] G. Kresse, Phys. Rev. B 59 (1999) 1758. [30] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865. [31] W. Tang, E. Sanville, G. Henkelman, J. Phys.: Condens. Matter 21 (2009) 084204. 72 M. Li et al. / Journal of Catalysis 314 (2014) 66–72 [32] J. Jiang, J. Turnbull, W. Lu, P. Boguslawski, J. Bernholc, J. Chem. Phys. 136 (2012) 014702. [33] F. Calle-Vallejo, M.T.M. Koper, Electrochim. Acta 84 (2012) 3. [34] J.K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J.R. Kitchin, T. Bligaard, H. Jónsson, J. Phys. Chem. B 108 (2004) 17886. [35] I.C. Man, H.-Y. Su, F. Calle-Vallejo, H.A. Hansen, J.I. Martínez, N.G. Inoglu, J. Kitchin, T.F. Jaramillo, J.K. Nørskov, J. Rossmeisl, ChemCatChem 3 (2011) 1159. [36] S. Zuluaga, S. Stolbov, J. Chem. Phys. 135 (2011) 134702. [37] L. Ci, L. Song, D. Jariwala, A.L. ElÃas, W. Gao, M. Terrones, P.M. Ajayan, Adv. Mater. 21 (2009) 4487. [38] L. Ci, Z. Xu, L. Wang, W. Gao, F. Ding, K.F. Kelly, B.I. Yakobson, P.M. Ajayan, Nano Res. 1 (2008) 116. [39] X. Jia, M. Hofmann, V. Meunier, B.G. Sumpter, J. Campos-Delgado, J.M. RomoHerrera, H. Son, Y.-P. Hsieh, A. Reina, J. Kong,, Science 323 (2009) 1701. [40] X. Wang, X. Li, L. Zhang, Y. Yoon, P.K. Weber, H. Wang, J. Guo, H. Dai, Science 324 (2009) 768. [41] I.-Y. Jeon, S. Zhang, L. Zhang, H.-J. Choi, J.-M. Seo, Z. Xia, L. Dai, J.-B. Baek, Adv. Mater. 25 (2013) 6138.
© Copyright 2026 Paperzz