First-Principles Study on the Synergistic Mechanism of SnO2 and

Article pubs.acs.org/JPCC

First-Principles Study on the Synergistic Mechanism of SnO2 and Graphene As a Lithium Ion Battery Anode Ling Miao,† Jiangbin Wu,† Jianjun Jiang,*,† and Pei Liang‡ †

Department of Electronic Science and Technology, Huazhong University of Science and Technology, Wuhan, Hubei 430074, People’s Republic of China ‡ College of Optical and Electronic Technology, China Jiliang University, Hangzhou, Zhejian 310018, People’s Republic of China ABSTRACT: Using first-principles calculations, we investigated the effect of the synergistic mechanism on the properties of a SnO2 and graphene hybrid structure, including the stability, electronic, and Li diffusion performance. The stable interface formed by C−O covalent bonds not only improves the structural stability but also makes the hybrid structure more conductive than the semiconducting SnO2. The calculated binding energies and diffusion barriers show that Li tends to insert into the interface, and a new pathway along the SnO2(110) direction for Li rapid diffusion with a low barrier is found.

■

INTRODUCTION Lithium ion batteries (LIBs) have attracted special attention in scientific and industrial fields due to their high electromotive force and high energy density.1,2 Since first reported as active anode materials for LIBs, transition-metal oxides (MOs), generally at nanoscale, such as NiO,3 CoO,4 Fe3O4,5 Mn3O4,6 CuO,7 and SnO2,8 have caused great concern due to their higher specific capacity9,10 than other materials. The nanostructured MO has shown advantages compared with microsized structures for LIBs, owing to the shortened diffusion length for both Li ions and electrons and larger specific surface area for the electrode/electrolyte interaction. However, drawbacks such as poor conductivity and the large volume expansion during charge/discharge remain problematic and require continuous development. Graphene is a new elastic and flexible material with very high conductivity.11 Therefore, many researchers have attempted to reassemble graphene nanosheets (GNS) in the presence of MO nanoparticles without any deterioration of fundamental electrochemical properties of both components.12−15 Paek et al. obtained SnO2/GNS exhibiting a reversible capacity of 810 mAh/g. After 30 cycles, the charge capacity still remained 570 mAh/g.16 The same effect has been elucidated in a recent report on Mn3O4/graphene17 and Co3O4/graphene18 based anodes. In theory, research focuses on the mechanism of intercalation and deintercalation of Li on MOs.19 Lithiated titania has been studied penetratingly by computational method,20,21 and some new phases of highly lithiated titania were predicted using ab initio calculations, and possible formation mechanics were given.22,23 Yildirim et al.24 studied the effect of concentration on the energetics and dynamics of Li ion transport in anatase and amorphous TiO2 by molecular dynamics and density functional theory (DFT). Furthermore, toward understanding the effects of carbon- and nitrogen-doped carbon coating on © 2012 American Chemical Society

the electrochemical performance of Li4Ti5O12 (LTO) in LIBs, Ding et al.25 studied the atomistic model for the interface between the lithium transition-metal oxide and carbon coating layers by first-principles calculations. They found that electron transfer from the LTO surface to graphene greatly improves the electric conductivity of the interface. In the MO/GNS system, graphene-containing carbonaceous materials afford not only a buffer to relieve the cracking of MOs but a good ion and/or electron transport medium in the electrode.26 In the obtained MO/GNS, not only GNS but also MO nanoparticles could play a role as electrode materials to get a synergetic effect, resulting in the superior cyclic performances and eventually higher reversible capacities in comparison with the ordinary MO.27 However, previous research on MO/GNS as anode materials of LIBs are mainly experimental studies. The synergistic mechanism of the MO/GNS and its influence on intercalation and deintercalation of Li, which are still not well understood, are lacking of systematic theoretical study. So, this paper uses first-principles calculations to analyze performance and mechanism of Li adsorption on MO/GNS compositea case study of SnO2/graphene hybrid structure. We will study the structural stability, electronic, and Li diffusion performance of the SnO2 and graphene hybrid structure.

■

METHOD AND MODEL The first-principles calculations are performed using the VASP code,28 in the framework of DFT.29,30 We use the projector augmented waves31 method and the generalized gradient approximation32 exchange-correlation function developed by Perdew, Burke, and Ernzerh.33 A plane wave cutoff of 400 eV Received: July 3, 2012 Revised: November 11, 2012 Published: December 13, 2012 23

dx.doi.org/10.1021/jp306572c | J. Phys. Chem. C 2013, 117, 23−27

The Journal of Physical Chemistry C

Article

and a k-point mesh of 2 × 2 × 1 in the Monkhorst Pack34 sampling scheme are used. The structural relaxation is performed using the conjugated gradient minimization method with the maximum force on each atom less than 0.05 eV/Å. To model the experimental SnO2/GNS (SnO2 nanoparticles with diameter of about 10−50 nm and graphene nanosheets of a few layers with about 10−1000 μm size) composite,19 the interface structure of graphene and the SnO2 surface is constructed. The synergistic mechanism is believed to be mainly contributed from the interaction at the interface, while the rest of the graphene and SnO2 far away from the interface have little effect and remain their respective Li intercalation and deintercalation performance. The (110) surface is the most stable surface of SnO2,35 and the (110) plane of rutile structure is clearly observed in the experiment,15,16 when SnO2/GNS is used for the LIB electrode. There are three kinds of different SnO2(110) surfaces: two O-terminated surfaces (O1 surface and O2 surface) and one Sn- and O-terminated surface (Sn−O surface) (Figure 1a). So, in this study, we correspondingly

Figure 2. (a) Side view of the G@O1 hybrid structure; (b) detailed view of interface; (c) charge transfer of bonded C and O atoms.

SnO2 surface, and the hybrid structure, respectively. The single bond energy Esingle is given by Esingle = Eb/n, where n is the number of C−O bonds. The binding energies of the former two structures are about 0.05 eV, which means there is weak van der Waals interaction between the graphene and SnO2. While in the G@O1 structure, the single C−O bond energy is 1.26 eV, indicating this structure is more stable than the other two. The result of the binding energy agrees with the geometric structure analyses well, which shows that there is a strong connection between graphene and SnO2 in the G@O1 structure but not in G@O2 or G@Sn−O. The strong connection could prevent the agglomeration of SnO2 nanoparticles, which would cause the rapid capacity decay in the experiment. So, the G@O1 structure is more persuasive to describe the experimental situation, and we will mainly discuss it. To express the cycling stability of hybrid structure, a stress− strain curve and Young’s modulus are calculated (Figure 3).

Figure 1. (a) Three kinds of SnO2(110) surfaces: O1 surface, O2 surface, and Sn−O surface. (b) Top view of the G@O1 hybrid structure.

investigate three kinds of SnO2/graphene hybrid structures, with graphene placed on the above three surfaces, and denote these structures as G@O1, G@O2, and G@Sn−O, respectively. We model the SnO2(110) surface square lattice with a 3 × 2 supercell (9.73 Å × 13.64 Å) and the graphene square lattice with a 3 × 3 supercell (9.84 Å × 12.28 Å), including 36 C atoms. The lattice mismatch is less than 10%. For precise results, the thickness of the vacuum and SnO2(110) layer is checked in our study.

■

RESULTS AND DISCUSSION We begin with the stability of three interface structures (G@ O2, G@Sn−O, and G@O1). In the G@O2 and G@Sn−O structures, the distances between the SnO2 surface and graphene are 3.41 and 3.33 Å, respectively. Besides, graphene in these two structures remains very flat, and the SnO2 surface has little change compared to the original SnO2(110) surface. However in the G@O1 structure, six C atoms of graphene bond to the surface oxygen atoms of the SnO2 substrate with sp3 hybridization (Figure 2a), and both the surfaces of the substrate and graphene are distorted. The average of the nearest C−O distance is 1.42 Å (see Figure 2b), which is very close to the length of the sp3 C−O bond in the CH3−CH2−OH molecule (1.43 Å), indicating oxygen−carbon covalent bonding. It is also proved by the charge transfer between the bonded C and O atoms as shown in Figure 2c. We also study the binding energy Eb = Egra + ESnO2 − Etot. Here, Egra, ESnO2, and Etot are the total energy of graphene, the

Figure 3. Stress−strain curve of graphene, the hybrid structure, and SnO2.

Young’s modulus, also known as the tensile modulus, is a measure of the stiffness of an elastic material and is a quantity used to characterize materials. It is defined as the ratio of the uniaxial stress over the uniaxial strain in the range of stress in which Hooke’s law holds. Here, strain is defined as ε ≡ L/L0 − 1, and stress σ is the Cauchy stress. Generally, material has an isotropic in-plane elastic response at small strain, and the slope of the stress−strain curve is the Young’s modulus of the material. It depends on the bond strength of the material and can be used to describe the degree of deformation of anode material during charge/discharge here. Our calculated Young’s moduli of graphene (1058 GPa) and SnO2 (40 GPa) are very similar with Liu’s result36 and Huang’s 24

dx.doi.org/10.1021/jp306572c | J. Phys. Chem. C 2013, 117, 23−27

The Journal of Physical Chemistry C

Article

result,37 respectively, in keeping with their cycling performance as the LIB anode.8,13 The Young’s modulus of the hybrid structure is about 134 GPa, much higher (about 300%) than that of SnO2. To determine the cycling stability accurately, we also consider the Young’s modulus of the hybrid structure in different Li concentrations of 12.5%, 25%, 37.5%, and 50% without considering surface adsorption. As shown in Figure 3, the moduli, about 130, 125, 120, and 114 GPa of each of the Li concentrations, respectively, decrease as the concentrations increase. This Young’s modulus is still much higher than that of SnO2, although there is a certain degree of reduction. Volume changes of the hybrid structure with Li intercalation are estimated by comparing the original structure’s cell volume without any stress. The results are 1.8%, 3.7%, 5.6%, and 7.5% of each of the Li concentrations, respectively, which shows an incremental relationship with the Li concentration and corresponds to the change of Young’s modulus. LIBs would perform with pretty good cycling stability for this small volume change, because bulk SnO2 shows a very large volume change of about 300% during the full charge/discharge process. It indicates that the interaction with graphene could greatly muffle the destructive deformation of the SnO2 anode during the cycling process. It can be predicted that the hybrid structure would perform much better than SnO2 in cycle dynamics, which is expected for application in LIBs. The density of states (DOS) is calculated to measure the electronic conductivity of structure. For comparison, the total DOS of surface and bulk SnO2 are shown in Figure 4a. The

and the particular band structure leads to very high conductivity.11 To discuss why the interaction between graphene and SnO2 not reduces the conductivity performance of graphene, we compare the DOS of graphene with −OH revised graphene (G−OH), whose bonded C atoms are saturated by −OH as shown in Figure 4(c). After saturated by −OH, the DOS of G−OH around Fermi level is much bigger than 0, indicating that few regular C−O covalent bonds make a little impact on the electronic conductivity of graphene. As we known, the electronic conductivity of material is extremely correlative with its band structure and DOS. Especially the energy band gap around fermi level could almost determine the conductivity property. Although DFT calculated band gap at 0 K would change a litter at room temperature, qualitative conductivity property of material could be inferred according the calculated value. According to our analysis of DOS, conductivity of bulk SnO2 is slightly lower than that of SnO2 (110) surface, which are both semiconductive and not good enough as electrode for those big band gaps. The nonzero DOS peak around fermi level of hybrid structure also shows that the extremely high conductivity of graphene is not obviously undermined by the interaction between graphene and SnO2, and the hybrid structure is still quite conductive. Here, graphene provides the electronic conductive channel for hybrid structure to improve the electrochemical performances, which agrees with Paek’s analysis.16 The much higher conductivity of hybrid structure than SnO2, indicates that the internal resistance of hybrid structure anode would be much lower, which is expected in application of LIBs. We next consider Li adsorption on hybrid structure. The stability of Li adsorption can be measured by the binding energy, given as Ead=Eh+ELi―Et. Et, Eh and ELi are the total energies of total structure after Li adsorption, hybrid structure and isolated Li atom. The different adsorption site of Li atom in the hybrid structure is shown in Figure 5(a), and the binding energy Ead and charge of Li atom QLi at each site are shown in Figure 5(b). The binding energies at T3, T4, and T5 sites in the center of the hybrid structure are all about 2.7 eV, close to that of bulk SnO2 (2.49 eV), and the binding energy at the G site (1.21 eV) is also close to the case of isolated graphene (1.2 eV). The higher binding energy at the T6 site (3.13 eV) could be

Figure 4. (a) DOS of SnO2(110) surface and bulk; (b) DOS of the hybrid structure; (c) DOS of graphene and G−OH, inset is −OH revised graphene.

calculated band gap of bulk SnO2 is 2.29 eV, narrower than the experimental value (3.6 eV),38 for DFT always underestimating the band gap. The narrower band gap of the SnO2(110) surface (1.3 eV) is due to the DOS peak slightly below the Fermi level from the surface O atoms. Both the calculated band gaps of bulk SnO2 and surface are very close to other DFT values (1.4 and 2.25 eV).39 It is noteworthy that surface and bulk have a quite wide band gap, which is not conducive to electron transport, showing a semiconductor behavior. The total DOS of hybrid structure in Figure 4(b) with no band gap around the Fermi level shows a conductor behavior, obviously superior to the case of SnO2, which is important for the application in LIBs. As well-known, graphene is semimetal,

Figure 5. (a) adsorption sites of Li atoms in hybrid structure; (b) binding energy and charge of Li atom at each site. 25

dx.doi.org/10.1021/jp306572c | J. Phys. Chem. C 2013, 117, 23−27

The Journal of Physical Chemistry C

Article

Figure 6. (a) Li diffusion paths in bulk SnO2; (b) energy barrier of lithium diffusion in bulk SnO2; (c) diffusion paths at the interface of the hybrid structure; (d) energy barrier of lithium diffusion in the hybrid structure.

explained by the surface adsorption.40 Moreover, the binding energies at T1 and T2 sites are even up to 4.28 and 3.90 eV, which is attributed to the synergetic interaction of the SnO2 surface and graphene. When the anode of the hybrid structure is charged, it is most probable that Li tends to occupy the interface sites first, according to the calculated binding energy at each site. By using Mulliken population analysis,41 the charge of the Li atom QLi is calculated as shown in Figure 5b. It can be found that the electron transfer from the adsorbed Li atom to the hybrid structure shows a similar relationship with a corresponding binding energy Ead as the site of absorption. In particular, the value of QLi at the T1 site (0.76e) and the G site (0.41e) are the maximum and minimum electron transfer of the Li atom in all sites, respectively. To investigate the deintercalation performance of Li with high binding energy near the interface, the Li diffusion barrier is calculated. We first calculate the Li diffusion in the bulk SnO2, as shown in Figure 6a for comparison. Path x is equivalent to path z, for the square symmetry of the bulk SnO2 structure. However, due to the reduced symmetry of the hybrid structure from the interaction with graphene, there are three possible nonequivalent pathways (path x′, path y′, and path z′) of lithium diffusion near the interface of the hybrid structure (see Figure 6c). In bulk SnO2, the energy barriers of paths x and y are 2.23 and 0.49 eV, respectively. It means that path y is the main Li diffusion pathway with better Li diffusion performance, whose barrier is close to the case of isolated graphene.42 We can get the reason by analyzing the detail structure of bulk SnO2. In the path x, the Li atom has to pass through a five-membered ring from one adsorption site to another one (see inset of Figure 6b). In contrast, path y is a broad square channel with relatively weaker interaction of Li with SnO2. In the hybrid structure, the energy barriers of paths x′ and y′ are 0.85 and 0.46 eV, respectively, which are correspondingly lower than those of paths x and y. It should be noted that the energy barrier of path x′ is evidently reduced. Therefore, the rapid Li diffusion along

both paths x′ and y′ of the hybrid structure is possible. As the inset of Figure 6d shows, the old O−Sn bonds breaking and the new O−C bonds forming make the five-membered ring along path x opened to larger space near the interface. While the interaction with graphene has small influence on the energy barrier of path z′ (about 2.2 eV). The Arrhenius equation D = D0exp(−Q/RT) gives the relationship between diffusion coefficient D and diffusion energy barrier Q, and here, D0, R, and T are the scale factor, Boltzmann constant, and temperature, respectively. Accordingly, we calculated the proportional diffusion coefficient at room temperature (T = 300 K) by obtained barriers, as D0 is approximately equal of each situation. In the bulk SnO2 case, the result Dy/Dx,z = 3.6 × 1029 indicates that it is almost impossible for Li diffusion along paths x and z, and only the (001) direction (path y) of SnO2 is conducive to diffusion of lithium, impeding the rapid charge and discharge of the LIB anode. Meanwhile at the interface of graphene and SnO2, we obtain that Dx′/Dx is almost to 1.1 × 1023, which means the incorporation of graphene markedly improves the diffusion performance of the x direction. The diffusion coefficient of paths x′ and y′ (Dy′/Dx′ = 2.2 × 105) is close at the order of magnitude, and path x′ also can be a good Li diffusion pathway. It is worth noting that the isolated Li ion diffusion can be very different from even low concentration regimes in such a system. The analysis of Yildirim et al.24 of Li ion diffusion in anatase TiO2 with Li concentrations ranging 10−50% indicates that Li ions become more diffusive with increasing concentration. They pointed out a surprise and an interesting result that an increase in Li ion concentration will lead to a decrease in effective barriers. This effect extremely likely occurs in our system for the similar structure of SnO2 and anatase TiO2, which needs more works to confirm in the future.

■

CONCLUSIONS In summary, the structural stability, electronic, and Li diffusion performance of the SnO2 and graphene hybrid structure have been studied. It was found that (1) the structural stability and 26

dx.doi.org/10.1021/jp306572c | J. Phys. Chem. C 2013, 117, 23−27

The Journal of Physical Chemistry C

Article

(23) Koudriachova, M. V. Chem. Phys. Lett. 2008, 458, 108−112. (24) Yildirim, H.; Greeley, J.; Sankaranarayanan, S. K. R. S. J. Phys. Chem. C 2011, 115, 15661−15673. (25) Ding, Z.; Zhao, L.; Suo, L.; Jiao, Y.; Meng, S.; Hu, Y. S.; Wang, Z.; Chen, L. Phys. Chem. Chem. Phys. 2011, 13, 15127−15133. (26) Zhu, X.; Zhu, Y.; Murali, S.; Stoller, M. D.; Ruoff, R. S. ACS Nano 2011, 5, 3333−3338. (27) Yu, G.; Hu, L.; Vosgueritchian, M.; Wang, H.; Xie, X.; McDonough, J. R.; Cui, X.; Cui, Y.; Bao, Z. Nano Lett. 2011, 11, 2905−2911. (28) Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251−14269. (29) Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, B864−B871. (30) Kohn, W.; Sham, L. Phys. Rev. 1965, 140, A1133−A1138. (31) Blöchl, P. E. Phys. Rev. B 1994, 50, 17953−17979. (32) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh Carlos Fiolhais, D. J. Phys. Rev. B 1992, 46, 6671−6687. (33) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 2006, 77, 3865−3868. (34) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188−5192. (35) Mulheran, P. A.; Hardingt, J. H. Modell. Simul. Mater. Sci. Eng. 1992, 1, 39−43. (36) Liu, F.; Ming, P.; Li, J. Phys. Rev. B 2007, 76, 064120−1− 064120−7. (37) Huang, J. Y.; Zhong, L.; Wang, C. M.; Sullivan, J. P.; Xu, W.; Zhang, L. Q.; Mao, S. X.; Hudak, N. S.; Liu, X. H.; Subramanian, A.; et al. Science 2010, 330, 1515−1520. (38) Dawar, A. L.; Joshi, J. C. J. Mater. Sci. 1984, 19, 1−23. (39) Mäki-Jaskari, M. A.; Rantala, T. T. Phys. Rev. B 2001, 64, 075407−1−075407−7. (40) Zhang, Q.; Zhang, W.; Wan, W.; Cui, Y.; Wang., E. Nano Lett. 2010, 10, 3243−3249. (41) Mulliken, R. S. J. Chem. Phys. 1995, 23, 1833−1840. (42) Uthaisar, C.; Barone., V. Nano Lett. 2010, 10, 2838−2842.

electronic conductivity of the hybrid structure are superior to that of SnO2, for the introduction of graphene; (2) the sites near the interface are more suitable for Li insert than other sites at the surface of graphene and inside of SnO2; (3) the Li diffusion performance along the SnO2(110) direction have been evidently improved. Our results are beneficial to understanding the microscopic origin of the synergistic mechanism of SnO2 and graphene as the anode of LIBs with high cycle stability and charge/discharge rate.

■

AUTHOR INFORMATION

Corresponding Author

*Phone/fax: +86-027-87544472; e-mail: [email protected]. cn. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work is supported by the National Natural Science Foundation of China (grant nos. 61172003 and 61006051). REFERENCES

(1) Arico, A. S.; Bruce, P.; Scrosati, B.; Tarascon, J. M.; Schalkwijk, W. V. Nat. Mater. 2005, 4, 366−377. (2) Bruce, P. G.; Scrosati, B.; Tarascon, J. M. Angew. Chem., Int. Ed. 2008, 47, 2930−2946. (3) Poizot, P.; Laruelle, S.; Grugeon, S.; Dupont, L.; J. Tarascon, M. Nature 2000, 407, 496−499. (4) Armand, M.; J. Tarascon, M. Nature 2008, 451, 652−657. (5) Gao, X. P.; Bao, J. L.; Pan, G. L.; Zhu, H. Y.; Huang, P. X.; Wu, F.; Song, D. Y. J. Phys. Chem. B 2004, 108, 5547−5551. (6) Yuan, S. M.; Li, J. X.; Yang, L. T.; Su, L. W.; Liu, L.; Zhou, Z. ACS Appl. Mater. Interfaces 2011, 3, 705−709. (7) Liu, L.; Li, Y.; Yuan, S. M.; Ge, M.; Ren, M. M.; Sun, C. S.; Zhou, Z. J. Phys. Chem. C 2010, 114, 251−255. (8) Su, J.; Cao, M.; Ren, L.; Hu, C. J. Phys. Chem. C 2011, 115, 14469−14477. (9) Poizot, P.; Laruelle, S.; Grugeon, S.; Dupont, L.; Tarascon, J. M. J. Power Sources 2001, 235, 97−98. (10) Needham, S. A.; Wang, G. X.; Liu, H. K. J. Power Sources 2006, 159, 254−257. (11) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Science 2004, 306, 666−669. (12) Sun, Y.; Hu, X.; Luo, W.; Huang, Y. ACS Nano 2011, 5, 7100− 7107. (13) Chen, D.; Ji, G.; Ma, Y.; Lee, J. Y.; Lu, J. ACS Appl. Mater. Interfaces 2011, 3, 3078−3083. (14) Wang, X.; Zhou, X.; Yao, K.; Zhang, J.; Liu, Z. Carbon 2011, 49, 133−139. (15) Li, Y.; Lv, X.; Lu, J.; Li, J. J. Phys. Chem. C 2010, 114, 21770− 21774. (16) Paek, S. M.; Yoo, E.; Honma, I. Nano Lett. 2009, 9, 72−75. (17) Wang, H.; Cui, L. F.; Yang, Y.; Sanchez Casalongue, H.; Robinson, J. T.; Liang, Y.; Cui, Y.; Dai, H. J. Am. Chem. Soc. 2010, 132, 13978−13980. (18) Wu, Z. S.; Ren, W.; Wen, L.; Gao, L.; Zhao, J.; Chen, Z.; Zhou, G.; Li, F.; Cheng, H. M. ACS Nano 2010, 4, 3187−3194. (19) Sushko, M. L.; Rosso, K. M.; Liu, J. J. Phys. Chem. C 2010, 114, 20277−20283. (20) Ouyang, C. Y.; Zhong, Z. Y.; Lei, M. S. Electrochem. Commun. 2007, 9, 1107−1112. (21) Zhong, Z.; Ouyang, C.; Shi, S.; Lei, M. Chem. Phys. Chem. 2008, 9, 2104−2018. (22) Koudriachova, M. V.; de Leeuw, S. W.; Harrison, N. M. Chem. Phys. Lett. 2003, 371, 150−156. 27

dx.doi.org/10.1021/jp306572c | J. Phys. Chem. C 2013, 117, 23−27