View PDF VersionPrevious ArticleNext Article
 
DOI: 10.1039/C6RA17651H (Paper) RSC Adv., 2016, 6, 79662-79667

DFT-based study on the mechanisms of the oxygen reduction reaction on Co(acetylacetonate)2 supported by N-doped graphene nanoribbon†

Nan Wang, Yuejie Liu*, Jingxiang Zhao* and Qinghai Cai
Key Laboratory of Photonic and Electronic Bandgap Materials, Ministry of Education, College of Chemistry and Chemical Engineering, Harbin Normal University, Harbin, 150025, China. E-mail: xjz_hmily@163.com; zjx1103@hotmail.com

Received 11th July 2016 , Accepted 15th August 2016

First published on 16th August 2016

Abstract

Development of low-cost and highly efficient electrocatalysts for oxygen reduction reaction (ORR) is still a great challenge for the large-scale application of fuel cells and metal–air batteries. In this work, by means of density functional theory (DFT) computations, we have systemically explored the anchoring of Co(acac)2 (acac = acetylacetonate) on N-doped graphene nanoribbon and its potential as the ORR electrocatalyst. Our DFT computations revealed that N-doped graphene nanoribbon can be used as the anchoring material of the Co(acac)2 complex due to the formation of a Co–O4–N moiety, thus ensuring its high stability. Especially, an O2 molecule can be moderately activated on the surface of the anchored Co(acac)2 complex, and the subsequent ORR steps prefer to proceed though a more efficient 4e pathway with a small overpotential (0.67 V). Therefore, the hybridization of Co(acac)2 with N-doped graphene can give rise to outstanding catalytic performance for ORR in fuel cells.

1. Introduction

In recent years, graphene electrochemistry has gained growing interest due to its potential applications for energy conversion and storage devices such as fuel cells.1–22 One of potential applications of graphene is its superior electrocatalytic performance for the oxygen reduction reaction (ORR) occurring at the cathode of fuel cells.1,3,9,21 Thus, graphene-based electrocatalysts are regarded as one of the most promising alternatives to the precious Pt-based catalysts owing to their low cost, fuel tolerance, and long-term durability. Especially, the ORR performance of graphene-based materials can be further enhanced by tuning their charge/spin distributions through mono- and co-doping heteroatoms (e.g., N, B, S, P, I, or metal atoms) into a graphene matrix.4,8,11,18 Among various explored materials, N-doped graphenes have been extensively studied because they are considered as the most promising ORR electrocatalysts of low cost, high stability, and high efficiency to replace Pt-based catalysts for fuel cells.18,23–25 In spite of superior catalytic performance, some challenges still exist for N-doped graphenes as ORR electrocatalysts. For example, the exact catalytic mechanism of N-doped graphene is still under debate because the contribution of three nitrogen doping configurations (e.g., graphitic, pyridinic, and pyrrolic N) to the catalytic performance is unclear.26–31

Very recently, Han et al. have synthesized a novel hybrid material by the reaction of an organometallic complex, [Co(acac)2](acac = acetylacetonate), with N-doped graphene-based materials at room temperature.32 The hybrid material shows high electrocatalytic activity for the ORR. Remarkably, the room temperature coordination of organometallic molecules and graphene-based materials preserves the precursor structures and allows greater predictability of local structure around the metal center,32 thus leading to the design of novel electrocatalysts in a more rational fashion.

Inspired by this pioneered work, in this work, by means of density functional theory (DFT) computations, we investigated the entire oxygen reduction on Co(acac)2 complex supported by N-doped graphene under an acidic environment. Our results revealed that the Co(acac)2 complex possesses good stability on N-doped graphene and exhibits excellent ORR catalytic activity. Thus, coordination of organometallic complexes and graphene-based materials can lead to a new class of efficient and low-cost ORR catalysts.

2. Computational methods and models

All computations were performed with the spin-polarized DFT framework as implemented in the DMol3 code.33,34 The generalized gradient approximation (GGA) with PBE functional35 was utilized to describe the exchange and correlation effects. The double numerical plus polarization (DNP) basis set was adopted, and its accuracy is comparable to that of Pople's 6-31G**. Remarkably, the PBE/DNP method in Dmol3 code has been widely employed for evaluating the potential of transition metal-based electrocatalysts for the ORR,36–41 although the GGA with the correction of a Hubbard U term could be more suitable to compute electronic structures of graphene with the transition metal dopant.42

Due to the existence of transition metal atoms, the DFT semicore pseudopotential (DSPP) method43 was employed to treat the relativistic effect, which introduces some degree of relativistic corrections to the core. To accurately describe the long-range electrostatic interactions of ORR species with catalysts, the PBE + D2 method with the Grimme vdW correction44 was employed. In the geometry structural optimization, the convergence tolerances of energy, maximum force, and displacement were 1.0 × 10−5 Ha, 0.002 Ha Å−1, and 0.005 Å, respectively. To ensure high quality results, the real-space global orbital cutoff radius was used as high as 4.7 Å in all the computations. A conductor-like screening model (COSMO)45 was used to simulate a H2O solvent environment throughout the whole process. The dielectric constant was set as 78.54 for H2O solvent.

In Han et al.' study, ten possible binding sites for Co(acac)2 complex on various kinds N-doped graphene, including N-containing aromatic rings such as pyrroles, pyridines, and imidazoles.32 Their computations of all possible binding configurations suggest that the surface pyridine-type edge in zigzag graphene nanoribbon has the strongest binding strength for Co(acac)2 complex due to an electrostatic interaction.32 On the basis of this study, in this work, the graphene edge with the pyridine-type zigzag edge was employed as the anchoring material for Co(acac)2 complex. Notably, the armchair graphene nanoribbon was not considered because it possesses lower chemical reactivity than zigzag one.46 This formed hybrid material is periodic in the z direction and adopts a supercell length of 19.68 Å, and in the x and y dimensions were chosen to be 30 × 20 Å2, which are large enough to avoid the interactions with their periodic images. A 1 × 1 × 5 Monkhorst–Pack k-point mesh was chosen in geometry optimizations. The minimum energy path (MEP) for O2 dissociation was computed by LST/QST tools in Dmol3 code. The adsorption energy (Eads) of adsorbate on substrate was computed as: Eads = Eadsorbate/substrate − Eadsorbate − Esubstrate, where Eadsorbate/substrate, Eadsorbate, and Esubstrate are the total energies of the adsorbed systems, an adsorbate species, and the substrate, respectively. According to this equation, a negative value of Eads denotes an exothermic adsorption process. The charge transfer was computed by the Hirshfeld method.47

The change in free energy (ΔG) of the rate-determining step is an important parameter to compute the overpotential, which is essential to evaluate the catalytic performance of the proposed electrocatalyst. As described by Nørskov et al.,48–50 the reaction free energies can be computed by referencing to computational hydrogen electrode (CHE) model. The ΔG of each elementary step can be was determined as follows: ΔG = ΔE + ΔEZPE − TΔS + ΔGpH + ΔGU, where ΔE was the reaction energy directly obtained from DFT calculations, ΔEZPE is the change in the zero-point energy, T is the temperature (298.15 K, if note mentioned clearly), and ΔS is the change in entropy. ΔGU = −neU, where n was the number of electrons transferred and U was the applied electrode potential. ΔGpH = kbT × ln[thin space (1/6-em)]10 × pH represents the free energy contribution due to the variations in the H concentration, and in this work the value of pH was assume to 0 in acidic medium. In reaction free energy calculations, zero point energy (ZPE) values of the adsorbed species were obtained from the vibrational frequencies calculations using the DMol3 program, and the entropy of the adsorbed species was ignored. The entropies and vibrational frequencies of gaseous molecules were taken from the standard tables in the Physical Chemistry text book.48

3. Results and discussion

3.1. Stability and properties of the Co(acac)2 complex supported by N-doped graphene nanoribbon

Before exploring the mechanisms of the ORR on Co(acac)2/N-doped graphene nanoribbon, a thermodynamic prediction was firstly carried out to examine its stability, which is the ultimate prerequisite for a material to become an ORR catalyst. Our results indicated that the Co atom in Co(acac)2 complex points to the center of the N-containing hexagonal ring with the shortest distance of ∼3.49 Å, forming the O4–Co–N moiety (Fig. 1). The adsorption energy of Co(acac)2 complex on N-doped graphene nanoribbon is computed to be −1.03 eV, which is well agreement with previous report.32 The large adsorption energy ensures the good stability of the Co(acac)2 complex on N-doped graphene nanoribbon. About 0.10 electrons are transferred from Co(acac)2 to N-doped graphene nanoribbon, indicating that the substrate can withdraw electrons from Co(acac)2 complex. Additionally, our computations demonstrate that when the Co(acac)2 complex is noncovalently deposited on the N-doped graphene zigzag nanoribbon, the antiferromagnetic spin configuration (with up- and down-spin states localized on opposite zigzag edges) is the ground state with the net magnetic moment of 1.37 μB, which mainly originates from the contribution of the Co atom (1.03 μB, Fig. S1 in ESI†). It is noted that the antiferromagnetic effects of the nanoribbon can also be induced by the noncovalent functionalization with other organic molecules.51–55
image file: c6ra17651h-f1.tif
Fig. 1 Optimized structure of Co(acac)2 anchored on N-doped graphene nanoribbon (top and side views).

3.2. O2 activation

To begin the ORR reaction, an O2 molecule was first introduced on the surface of Co(acac)2/N-doped graphene nanoribbon. To obtain the most energetically favorable configuration for O2 adsorption, we considered various initial configurations by placing O2 in three typical sites: (1) O2 adsorption on the Co(acac)2 complex (O2/Co(acac)2/graphene), (2) O2 adsorption on the outside surface of the N-doped graphene nanoribbon, and (3) O2 embedded in the interlayer of Co(acac)2/graphene (Co(acac)2/O2/graphene). A full geometry optimization for the structure of O2 adsorption is performed by considering two initial configurations, including the end-on configuration (superoxo form) and the side-on configuration (peroxo form).

Our DFT results demonstrated that the side-on configuration is unstable and it will spontaneously convert to the end-on configuration. The most stable adsorption site of O2 molecule is the Co site, and O2 molecule is energetically favorable to locate the outside of the Co(acac)2 complex (Fig. 2a) with the adsorption energy of −0.96 eV (the ΔG value is −0.34 eV after taking account of ΔEZPE and entropy.). In addition, two meta-stable configurations are also obtained as shown in Fig. S2,† in which O2 molecule is attached to the interlayer of Co(acac)2/graphene (Eads = −0.31 eV) and the outside surface of N-doped graphene (−0.19 eV). Although O2 molecule has a negative adsorption energy, the ΔG turns to be positive 0.33 and 0.45, suggesting that the O2 molecule is unable to be sufficiently activated on the two sites. Thus, in the following sections, we will focus on the most stable configuration.


image file: c6ra17651h-f2.tif
Fig. 2 (a) Optimized configuration of O2 adsorption on Co(acac)2/N-doped graphene nanoribbon and (b) the corresponding projected density of states. The unit of the bond length is Å.

In the most sable configuration (Fig. 2a), the distance of the adsorbed O atom in O2 molecule (labeled as Oa) and the Co site is 1.87 Å, and the other O atom pointing outward (labeled as Ob) with an angle (∠Co–Oa–Ob) of ∼117.1° (Fig. 2a). The O–O bond length is elongated to 1.29 Å from that of 1.21 Å for an isolated O2 molecule. According to the Hirshfeld charge population analysis, the O2 molecule gains a total 0.24 e− extra charge during chemisorption, in which only a 0.06 e− charge accumulated on outer Ob atom, while the remaining 0.18 e− charge accumulated on the inner Oa atom bonded to the Co atom.

The elongation of O–O bond due to excess charge transfer to the O2 molecule and moderate adsorption strength of O2 on Co(acac)2/N-doped graphene nanoribbon indicate that the chemisorbed O2 molecule has been sufficiently activated, which would initiate its subsequent reduction reaction. To further gain deeper insight into O2 activation, we also computed the PDOSs for O2 chemisorbed on Co(acac)2/N-doped graphene nanoribbon (Fig. 2b). Our results revealed that there is an obvious hybridization between the O2-2p orbitals and Co-3d orbitals in both spin-up and spin-down channels, and the O2-2π* state is partially occupied mostly due to the charge transfer from Co(acac)2/N-doped graphene to O2 molecule.

3.3. Mechanisms of the ORR on Co(acac)2/N-doped graphene nanoribbon

Once O2 molecule is activated by the anchored Co(acac)2 complex, we explored the mechanisms of the oxygen reduction on its surface under an acidic environment.

The ORR at cathodes could proceed through two reaction mechanisms: (I) the chemisorbed O2 molecule is directly dissociated into two O* species, or (II) is hydrogenated by reacting with one proton and electron to form OOH species, followed by its further hydrogenation to two water molecules. The atomic configurations and the corresponding free energy changes (ΔG) of each elementary reaction in the ORR are displayed in Fig. 3 and 4.


image file: c6ra17651h-f3.tif
Fig. 3 The optimized geometric structures of various states (OOH*, O* + H2O, O*, OH*, and H2O) along the reaction path of ORR proceeded on Co(acac)2/N-doped graphene nanoribbon. The unit of the bond length is Å.

image file: c6ra17651h-f4.tif
Fig. 4 The computed free-energy diagram of ORR on the Co(acac)2/N-doped graphene nanoribbon surface. Black and red lines represent reactions at zero electrode potential (U = 0 V) and the equilibrium potential (U = 1.23 V), respectively.

Providing that the ORR follows the dissociative mechanism, the first elementary step is where the chemisorbed O2 molecule is dissociated into two separate O atoms on the surface, where one O atom bonded to Co atom of Co(acac)2 and the other O atom is adsorbed between two neighboring O atoms. Our computations demonstrated that this dissociation reaction is endothermic by 0.57 eV and has to overcome an energy barrier of 3.15 eV. The high energy barrier for the O–O bond dissociation step clearly indicates that the complete ORR steps on the supported Co(acac)2 complex by N-doped graphene nanoribbon via this mechanism is almost impossible at normal fuel cells operating conditions (approximately 350 K).

When the ORR follows the associative mechanism, the adsorbed O2 molecule is firstly hydrogenated by adsorbing a proton coupled with an electron transfer to form an OOH species, which is still located on the central Co site. The hydrogen binds to the outer O site with the Ob–H bond length of 0.98 Å, and the O–O bond is elongated from 1.29 Å of image file: c6ra17651h-t1.tif species to 1.43 Å of OOH* species (Fig. 3a). Remarkably, this step (i.e., image file: c6ra17651h-t2.tif + H+ + e− → OOH*) is exothermic in the free energy profile by 0.56 eV on Co(acac)2/N-doped graphene nanoribbon (Fig. 4), respectively, revealing that OOH species can be easily formed.

The formed OOH species can be further hydrogenated by reacting with another proton and an electron transfer. Due to random nature, the second hydrogen is possible to interact with both two O sites (i.e., Oa and Ob sites in Fig. 3a), thus leading to two different pathways, one is the four-electron (4e) reduction pathway in which O2 is reduced to two H2O molecules, and the other is the two-electron (2e) reduction pathway in which O2 is reduced to a H2O2 molecule.

In the 4e reduction pathway, the first H2O molecule would be created after the adsorption of H on the pre-hydrogenated Ob site of OOH group. The H2O molecule is attached to the Oa atom, which still locates on central metal site with a distance of 1.77 Å, via H-bonding with a distance of 1.54 Å (Fig. 3b). This process, (i.e., OOH* + H+ + e− → O* + H2O) is downhill in the free energy profile by 1.45 eV (Fig. 4). Subsequently, we introduced a third H+ and an additional electron to the system. The H atom interacts with the remaining Ob atom on the Co site to yield an OH species, which is adsorbed at the Co site and has a Co–O bond of 1.85 Å (Fig. 3c). The reaction free energy for this step is −1.80 eV. Finally, a fourth H+ was introduced with an additional electron to react with the adsorbed OH group to form the second H2O molecule, which will be easily released from the surface due to its weak adsorption on this catalyst (−0.30 eV, Fig. 3d). This final step is also downhill in the free energy profile by 0.78 eV.

Providing that the oxygen reduction takes place via the 2e pathway, the proton will interact with the pre-unhydrogenated Oa site of the OOH species to form a H2O2 molecule. The computed free energy for this process, OOH* + H+ + e− → HOOH on the Co(acac)2/N-doped graphene is −0.59 eV, which is much less favorable than that of the reduction of OOH* group to O* + H2O (−1.45 eV) in the 4e pathway. Thus, it is energetically favorable for the ORR on Co(acac)2/N-doped graphene to take place via the 4e reduction pathway rather than the 2e reduction pathway.

In Fig. 4, the free energy diagrams of complete ORR via the association mechanism on Co(acac)2/N-doped graphene nanoribbon were plotted at two different electrode potentials, U = 0 and 1.23 V (the ideal electrode potential for ORR vs. the reversible hydrogen electrode at T = 298.15 K). Our DFT results revealed that, at zero potential, all elementary steps in the ORR catalyzed by the Co(acac)2/N-doped graphene nanoribbons are downhill in the free energy profile. However, at U = 1.23 V, some of the intermediate steps turn to be uphill. For example, the reduction of image file: c6ra17651h-t3.tif to OOH* and OH* to H2O is uphill in the free energy profile by 0.67 and 0.45 eV, respectively. Thus, the rate-limiting step of oxygen reduction on Co–O4–N moiety lies in the second (H+ + e−) transfer step (i.e., image file: c6ra17651h-t4.tif → OOH*) with the energy barrier (or overpotential, defined as the minimum energy required for the ORR on the surface) of 0.67 eV, which is far lower than that of Co–N co-doped graphene (1.00 eV),42 graphene-supported Pt nanoparticles (1.13–1.68 eV).56 Since smaller value of free energy change in the rate-determining step indicates faster ORR reaction, the Co(acac)2 complex supported by N-doped graphene nanoribbon exhibits superior ORR catalytic performance. Especially, the further optimization for the coupling between organometallic complex containing Mn, Co, Fe, and Ni and nonmetal elements (e.g., N, B, P, and S) doped graphene may induce graphene-based electrocatalysts with higher catalytic activity due to an optimum surface activity to interact with oxygenated species.

To deeply understand the synergistic effect between N-doped graphene and Co(acac)2 complex, we further the free energies of the ORR occurring on the pristine N-doped graphene, isolated Co(acac)2 complex, the outside surface of graphene in the Co(acac)2/N-doped graphene, and the interlayer of the Co(acac)2/N-doped graphene. As shown in Fig. S3,† the computed overpotentials for the four cases are 0.72, 0.69, 0.70, and 1.05 eV, which are smaller than that of on the outside surface of Co(acac)2 complex in the Co(acac)2/N-doped graphene (0.67 eV). Thus, the interaction between Co(acac)2 complex and N-doped graphene has a co-operative effect for promoting the ORR.

4. Conclusions

In summary, our spin-polarized DFT computations demonstrated that the Co(acac)2 complex can be stably anchored on N-doped graphene due to the certain amount of charge transfer between each other. Especially, this hybrid material can moderately activate O2 molecule, and the subsequent ORR steps prefer to take place though a more efficient 4e pathway with a considerably small overpotential of 0.67 eV. Therefore, our theoretical results suggest a quite promising alternative non-Pt ORR catalyst for fuel cells, whose ORR activity and durability can be further improved by the careful optimization for the coupling between organometallic complex and non-carbon dopants in the graphene.

Acknowledgements

Support by NSFC (21203048) and the Excellent Young Foundation of Harbin Normal University (XKYQ201304) are gratefully acknowledged.

References

  1. M. Zhou, H.-L. Wang and S. Guo, Chem. Soc. Rev., 2016, 45, 1273–1307 RSC.
  2. Y. Yang, C. Han, B. Jiang, J. Iocozzia, C. He, D. Shi, T. Jiang and Z. Lin, Mater. Sci. Eng., R, 2016, 102, 1–72 CrossRef.
  3. D. Higgins, P. Zamani, A. Yu and Z. Chen, Energy Environ. Sci., 2016, 9, 357–390 CAS.
  4. D. Deng, K. S. Novoselov, Q. Fu, N. Zheng, Z. Tian and X. Bao, Nat. Nanotechnol., 2016, 11, 218–230 CrossRef CAS PubMed.
  5. A. Ambrosi, C. K. Chua, N. M. Latiff, A. H. Loo, C. H. A. Wong, A. Y. S. Eng, A. Bonanni and M. Pumera, Chem. Soc. Rev., 2016, 45, 2458–2493 RSC.
  6. Y. Zhai, Z. Zhu and S. Dong, ChemCatChem, 2015, 7, 2806–2815 CrossRef CAS.
  7. Y. Jiao, Y. Zheng, M. Jaroniec and S. Z. Qiao, J. Am. Chem. Soc., 2014, 136, 4394–4403 CrossRef CAS PubMed.
  8. X. Wang, G. Sun, P. Routh, D.-H. Kim, W. Huang and P. Chen, Chem. Soc. Rev., 2014, 43, 7067–7098 RSC.
  9. D.-W. Wang and D. Su, Energy Environ. Sci., 2014, 7, 576–591 CAS.
  10. M. Liu, R. Zhang and W. Chen, Chem. Rev., 2014, 114, 5117–5160 CrossRef CAS PubMed.
  11. X.-K. Kong, C.-L. Chen and Q.-W. Chen, Chem. Soc. Rev., 2014, 43, 2841–2857 RSC.
  12. A. Ambrosi, C. K. Chua, A. Bonanni and M. Pumera, Chem. Rev., 2014, 114, 7150–7188 CrossRef CAS PubMed.
  13. C. Xu, B. Xu, Y. Gu, Z. Xiong, J. Sun and X. S. Zhao, Energy Environ. Sci., 2013, 6, 1388–1414 CAS.
  14. D. S. Su, S. Perathoner and G. Centi, Chem. Rev., 2013, 113, 5782–5816 CrossRef CAS PubMed.
  15. T. Kuila, A. K. Mishra, P. Khanra, N. H. Kim and J. H. Lee, Nanoscale, 2013, 5, 52–71 RSC.
  16. L. Dai, Acc. Chem. Res., 2013, 46, 31–42 CrossRef CAS PubMed.
  17. H. Chang and H. Wu, Energy Environ. Sci., 2013, 6, 3483–3507 CAS.
  18. H. Wang, T. Maiyalagan and X. Wang, ACS Catal., 2012, 2, 781–794 CrossRef CAS.
  19. L. Dai, D. W. Chang, J.-B. Baek and W. Lu, Small, 2012, 8, 1130–1166 CrossRef CAS PubMed.
  20. H.-J. Choi, S.-M. Jung, J.-M. Seo, D. W. Chang, L. Dai and J.-B. Baek, Nano Energy, 2012, 1, 534–551 CrossRef CAS.
  21. F. Cheng and J. Chen, Chem. Soc. Rev., 2012, 41, 2172–2192 RSC.
  22. Y. Sun, Q. Wu and G. Shi, Energy Environ. Sci., 2011, 4, 1113–1132 CAS.
  23. W. Liang, J. Chen, Y. Liu and S. Chen, ACS Catal., 2014, 4, 4170–4177 CrossRef CAS.
  24. J. Wang, L. Li and Z. D. Wei, Acta Phys.-Chim. Sin., 2016, 32, 321–328 CAS.
  25. D. Deng, L. Yu, X. Chen, G. Wang, L. Jin, X. Pan, J. Deng, G. Sun and X. Bao, Angew. Chem., Int. Ed., 2013, 52, 371–375 CrossRef CAS PubMed.
  26. C. V. Rao, C. R. Cabrera and Y. Ishikawa, J. Phys. Chem. Lett., 2010, 1, 2622–2627 CrossRef CAS.
  27. N. P. Subramanian, X. Li, V. Nallathambi, S. P. Kumaraguru, H. Colon-Mercado, G. Wu, J.-W. Lee and B. N. Popov, J. Power Sources, 2009, 188, 38–44 CrossRef CAS.
  28. H. Niwa, K. Horiba, Y. Harada, M. Oshima, T. Ikeda, K. Terakura, J.-i. Ozaki and S. Miyata, J. Power Sources, 2009, 187, 93–97 CrossRef CAS.
  29. H. Kim, K. Lee, S. I. Woo and Y. Jung, Phys. Chem. Chem. Phys., 2011, 13, 17505–17510 RSC.
  30. C. H. Choi, H.-K. Lim, M. W. Chung, J. C. Park, H. Shin, H. Kim and S. I. Woo, J. Am. Chem. Soc., 2014, 136, 9070–9077 CrossRef CAS PubMed.
  31. D. Guo, R. Shibuya, C. Akiba, S. Saji, T. Kondo and J. Nakamura, Science, 2016, 351, 361–365 CrossRef CAS PubMed.
  32. J. Han, Y. J. Sa, Y. Shim, M. Choi, N. Park, S. H. Joo and S. Park, Angew. Chem., Int. Ed., 2015, 54, 12622–12626 CrossRef CAS PubMed.
  33. B. Delley, J. Chem. Phys., 1990, 92, 508–517 CrossRef CAS.
  34. B. Delley, J. Chem. Phys., 2000, 113, 7756–7764 CrossRef CAS.
  35. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS PubMed.
  36. P. Zhang, X. Hou, L. Liu, J. Mi and M. Dong, J. Phys. Chem. C, 2015, 119, 28028–28037 CAS.
  37. P. Zhang, X. Hou, J. Mi, L. Liu and M. Dong, Chem. Phys. Lett., 2015, 641, 112–116 CrossRef CAS.
  38. Z. Lu, G. Xu, C. He, T. Wang, L. Yang, Z. Yang and D. Ma, Carbon, 2015, 84, 500–508 CrossRef CAS.
  39. Y. Wang, H. Yuan, Y. Li and Z. Chen, Nanoscale, 2015, 7, 11633–11641 RSC.
  40. N. Wang, L. Feng, Y. Shang, J. Zhao, Q. Cai and P. Jin, RSC Adv., 2016, 6, 72952–72958 RSC.
  41. L. Feng, Y. Liu and J. Zhao, Phys. Chem. Chem. Phys., 2015, 17, 30687–30694 RSC.
  42. F. Li, H. Shu, C. Hu, Z. Shi, X. Liu, P. Liang and X. Chen, ACS Appl. Mater. Interfaces, 2015, 7, 27405–27413 CAS.
  43. B. Delley, Phys. Rev. B: Condens. Matter Mater. Phys., 2002, 66, 155125 CrossRef.
  44. S. Grimme, J. Comput. Chem., 2006, 27, 1787–1799 CrossRef CAS PubMed.
  45. A. Klamt and G. Schuurmann, J. Chem. Soc., Perkin Trans. 1, 1993, 2, 799–805 RSC.
  46. T. Wassmann, A. P. Seitsonen, A. M. Saitta, M. Lazzeri and F. Mauri, Phys. Rev. Lett., 2008, 101, 096402 CrossRef PubMed.
  47. F. L. Hirshfeld, Theor. Chim. Acta, 1977, 44, 129–138 CrossRef CAS.
  48. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jónsson, J. Phys. Chem. B, 2004, 108, 17886–17892 CrossRef.
  49. J. Rossmeisl, A. Logadottir and J. K. Nørskov, Chem. Phys., 2005, 319, 178–184 CrossRef CAS.
  50. A. A. Peterson, F. Abild-Pedersen, F. Studt, J. Rossmeisl and J. K. Nørskov, Energy Environ. Sci., 2010, 3, 1311–1315 CAS.
  51. Y.-W. Son, M. L. Cohen and S. G. Louie, Nature, 2006, 444, 347–349 CrossRef CAS PubMed.
  52. Y.-L. Lee, S. Kim, C. Park, J. Ihm and Y.-W. Son, ACS Nano, 2010, 4, 1345–1350 CrossRef CAS PubMed.
  53. B. M. Wong, S. H. Ye and G. O'Bryan, Nanoscale, 2012, 4, 1321–1327 RSC.
  54. S. Tang and X. Cao, Phys. Chem. Chem. Phys., 2014, 16, 23214–23223 RSC.
  55. D. Hou, J. Wei and S. Xie, Phys. Chem. Chem. Phys., 2011, 13, 13202–13206 RSC.
  56. D.-H. Lim and J. Wilcox, J. Phys. Chem. C, 2012, 116, 3653–3660 CAS.

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra17651h
This journal is © The Royal Society of Chemistry 2016
Click here to see how this site uses Cookies. View our privacy policy here.