ARTICLE pubs.acs.org/JPCC
First-Principles Calculations on the LiMSO4F/MSO4F (M = Fe, Co, and Ni) Systems Yongmao Cai,† Gang Chen,‡ Xiaoguang Xu,§ Fei Du,‡ Zhe Li,† Xing Meng,‡ Chunzhong Wang,‡ and Yingjin Wei*,‡ †
College of Materials Science and Engineering, Jilin University, Changchun 130012, People’s Republic of China College of Physics/State Key Laboratory of Superhard Materials, Jilin University, Changchun 130012, People’s Republic of China § School of Materials Science and Engineering, University of Science and Technology Beijing, Beijing 100083, China ‡
ABSTRACT: Systematic first-principles calculations based on the density functional theory are carried out to discuss the crystal and electronic structures of the LiMSO4F/MSO4F (M = Fe, Co, and Ni) systems. It is shown that all of the LiMSO4F compounds are in a high spin antiferromagnetic ground state. However, they transform to different ground states with Liþ extraction. LiFeSO4F is a typical Mott-Hubbard insulator and then transforms to a charge-transfer insulator with Liþ extraction. The theoretical intercalation voltages of LiMSO4F are 3.54 (Fe), 4.73 (Co), and 5.16 V (Ni), respectively, which are close to corresponding LiMPO4 phosphates. First-principles calculations show that a significant amount of electron-charge transfer takes place on the oxygen anions with Liþ extraction, especially for LiCoSO4F and LiNiSO4F. This will lead to significant loss of oxygen from the material lattice which is an intrinsic drawback of LiMSO4F due to the concerns of structural and thermal stabilities.
1. INTRODUCTION Lithium ion batteries have some specific advantages such as high energy density and long cycle life. They have been intensively studied both experimentally and theoretically since 1970s. Cathode materials undergo an oxidation/reduction process when Liþ ions are extracted/inserted. So they are typical transition metal compounds, such as LiCoO2,1 LiMn2O4,2 LiFePO4,3 and Li2FeSiO4.4 These cathode materials have been reviewed by several groups.57 It is now widely accepted that LiFePO4 is a promising cathode material for large-scale lithium ion batteries due to its low price, high safety and nontoxic properties. However, LiFePO4 shows intrinsic low electronic conductivity and lithium ion diffusivity, necessitating nanosizing and/or carbon coating. Recently, by introducing a fluorine atom and replacing the [PO4]3- group of LiFePO4 with [SO4]2-, a new fluorosulfate cathode material LiFeSO4F was synthesized by Recham et al.8 X-ray diffraction revealed that LiFeSO4F has a triclinic structure with space group of P-1. The material shows a voltage plateau at 3.6 V vs Li/Liþ when cycled between 2.5 and 4.2 V, with a reversible specific capacity of 130 mAh g1. It was found that the ionic conductivity of LiFeSO4F is about 103 times higher than that of LiFePO4. This makes it possible to reach high rate capabilities without nanosizing. The discovery of LiFeSO4F not only provides a strong competitor of LiFePO4 but also suggests a new class of fluoror 2011 American Chemical Society
oxyanion cathode material for lithium ion batteries. Recently, Barpanda et al. prepared LiCoSO4F and LiNiSO4F using a low temperature ionothermal method.9 Unfortunately these materials did not show any electrochemical activity. Previous studies showed that first-principles method based on the density functional theory (DFT) is a powerful tool to study the electronic structures of cathode materials.10 Using first-principles calculations, Liu et al. obtained a theoretical intercalation voltage of 3.7 V for LiFeSO4F, which fits well with the experimental results.11 In a recent study, Ramzan et al. predicted the crystal structures of LiMSO4F (M = Co and Ni) via first-principles calculations.12 They reported that the compounds share a similar crystal structure with LiFeSO4F. However, the electronic structures of these compounds still need detailed study. As we know, at microscopic scale the electrochemical and physical properties of electrode materials are strongly correlated to their electronic structures. Therefore, we purpose to calculate the crystal and electronic structures of LiMSO4F (M = Fe, Co, and Ni) as well as their delithiated forms using first-principles calculations. In this paper, we present a deep understanding on the electrochemical and physical properties of the LiMSO4F/MSO4F systems. Received: November 29, 2010 Revised: February 24, 2011 Published: March 21, 2011 7032
dx.doi.org/10.1021/jp111310g | J. Phys. Chem. C 2011, 115, 7032–7037
The Journal of Physical Chemistry C
ARTICLE
Table 1. Relative Single-Point Energies and Magnetic Moments of LiMSO4F/MSO4F Calculated from GGA þ U Approximationa LiFeSO4F FeSO4F LiCoSO4F CoSO4F LiNiSO4F NiSO4F AFM
0
0
0
0
0
0
FM
31
115
18
80
10
51
4.63
5.04
3.65
2.57
2.02
192
NM M (μB)
0
a
The single-point energies of the AFM state are set as 0 meV. The magnetic moments are calculated based on the AFM states.
Figure 1. Crystal structure of LiMSO4F. The two different M sites are labeled as M(1) and M(2), respectively.
2. COMPUTATIONAL DETAILS All the first-principles calculations were performed with the CASTEP code in the Material Studio 5.0 package,13 which uses the plane-wave pseudopotential (PW-PP) approach. The number of plane waves was determined by a cutoff energy of 500 eV. The electronion interaction was described with the Vanderbilt ultrasoft pseudopotentials.14 The exchange-correlation functional proposed by Perdew, Burke, and Ernzerhof (PBE)15 was used. Generalized gradient approximation (GGA) and GGA þ U approximation16 dealing with the 3d electronelectron correlation were performed to determine the single-point energies and the electronic structures. A single effective coulomb energy U0 took the place of the on-site coulomb energy U and the exchange interactions J (U0 = U J). The effective parameter U0 is an experienced parameter, which was set to 3.0, 4.0, and 5.0 eV for the Fe, Co, and Ni system, respectively. It will show below that the theoretical intercalation voltages based on these settings are coinciding well with experimental results, which indicates that these U0 values were reliable for calculations. The special k-points sampling integration over the Brillouin zone was employed by using the Monkhorst-Pack method17 with a k-mesh of 5 5 4. The lattice parameters and the internal coordinates of LiMSO4F and MSO4F were optimized using the BFGS algorithm18 with a convergence tolerance of 5.0 104 Å maximum displacement in the structure and 5.0 106 eV/atom in the total energy. 3. RESULTS AND DISCUSSION Before calculating the electronic structures, the crystal structures of LiMSO4F (M = Fe, Co, and Ni) were optimized based on the experimental data.8,9 All the three compounds have a triclinic cell with space group of P-1, which is isostructural with that of LiMgSO4F19 (as shown in Figure 1). There are two types of M sites in the materials. M(1) and M(2) are the next and remote sites to lithium, respectively. Both M(1) and M(2) are surrounded by four oxygen and two fluorine atoms forming a distorted MO4F2 octahedron. There are also two Li sites each of which has site occupancy of 0.5. For simplicity we deleted one Li-site and set up the occupancy of the other as 1.0. This approximation was also adopted by Ramzan et al. in their recent work.12 It should be pointed out that the atomic coordinates of LiNiSO4F have not been experimentally reported yet. In this study, they were optimized based on the experimental data of LiCoSO4F. The basic structure of the material was unchanged after full extraction of Liþ from LiFeSO4F.8 The crystal structures of MSO4F were therefore obtained by optimizing the
Table 2. Calculated Crystal Lattice Parameters of LiMSO4F and MSO4F (M = Fe, Co, and Ni) Together with Experimental Data a (Å)
b (Å)
c (Å)
V (Å3) R (deg) β (deg) γ (deg)
LiFeSO4F 5.1340 5.1747 FeSO4F 5.1023 5.0735 LiCoSO4F 5.2240 5.1721 CoSO4F 5.0395 LiNiSO4F 5.1947 5.1430 NiSO4F 5.2162
5.3087 5.4943 5.1113 5.0816 5.3633 5.4219 5.1304 5.4481 5.3232 5.0635
7.1717 7.2224 6.9790 7.3363 7.2834 7.1842 7.0227 7.3400 7.1404 7.1662
171.240 182.559 159.849 163.640 180.038 177.804 159.197 182.500 172.560 165.881
compound
107.824 106.522 106.965 110.975 108.658 106.859 107.313 107.461 106.802 108.266
107.692 107.210 106.991 111.189 107.3 107.788 108.413 107.946 107.512 109.317
98.147 97.791 99.816 88.157 96.002 97.986 97.673 97.758 98.395 94.787
a b a b a c a a c a
a
This work. b Experimental data from ref 8. c Experimental data from ref 9.
theoretical structure after all Liþ ions were removed from the corresponding compound. The single-point energies of LiMSO4F and MSO4F were calculated with different magnetic alignments including nonmagnetic (NM), ferromagnetic (FM) and antiferromagnetic (AFM), respectively (as listed in Table 1). The calculations clearly predict a high spin AFM ground state for all M ions in LiMSO4F, which are expressed as t2g(v)3eg(v)2t2g(V)1 for Fe2þ, t2g(v)3eg(v)2t2g(V)2 for Co2þ, and t2g(v)3eg(v)2t2g(V)3 for Ni2þ. As for the delithated form, the Fe3þ ion in FeSO4F still shows a high spin AFM state, i.e., t2g(v)3eg(v)2. A low spin FM state with a electron configration of t2g(v)3t2g(V)3eg(v)1 is obtained for the Ni3þ ion in NiSO4F. However, the nonspin polarized t2 g(v)3t2 g(V)3 state is predicted as the most stable for the Co3þ ion in CoSO4F. The change in the spin state of M (M = Ni and Co) ions after Liþ extraction is due to the large crystal field splitting caused by the shrunk of material lattice as shown below. The M atoms in the MOSOM chain are rather remote, therefore the magnetic interaction should be mainly from the superexchange effect through the fluorine atom. It is noticed from Table 1 that the magnetic moments of M ions in LiMSO4F and MSO4F are much smaller than those of calculated from the ionic model. This indicates the mixed covalent/ionic bonding of the compounds. The crystal structure of LiMSO4F and MSO4F were optimized by GGA method. The theoretical lattice parameters and the selected bond lengths are compared with experimental data as shown in Tables 2 and 3, respectively. Our calculation results fit well with those in available literatures.8,9 One can see that the theoretical lattice parameters of LiFeSO4F are slightly smaller than the experimental data, but the overall errors are still reasonable. On the contrary, the theoretical lattice parameters of LiCoSO4F and LiNiSO4F are slightly larger than the 7033
dx.doi.org/10.1021/jp111310g |J. Phys. Chem. C 2011, 115, 7032–7037
The Journal of Physical Chemistry C
ARTICLE
Table 3. Bond Lengths of MO, MF, and SO of LiMSO4F and MSO4F LMO (Å)
LMF (Å)
LSO (Å)
LiFeSO4F
Fe(1)
2.01 2, 2.03 2
1.96 2
1.48 2, 1.49 2,
FeSO4F
Fe(2) Fe(1)
2.00 2, 2.02 2 1.93 4
1.95 2 1.88 2
1.50 4 1.49 8
Fe(2)
1.94 2, 1.93 2
1.88 2
Co(1)
2.04 2, 2.07 2
1.98 2
Co(2)
2.04 2, 2.10 2
1.97 2
LiCoSO4F CoSO4F
Co(1)
1.95 4
1.91 2
Co(2)
1.94 2, 1.95 2
1.90 2
1.49 4,1.50 4 1.49 8
LiNiSO4F
Ni(1)
2.10 2, 2.12 2
2.01 2
NiSO4F
Ni(2) Ni(1)
2.07 2, 2.13 2 1.97 2, 1.91 2
2.00 2 1.96 2
1.48 2,1.50 6 1.47 2,1.49 4,
Ni(2)
2.10 2, 1.94 2
1.88 2
1.51 2
Table 4. Calculated Average Intercalation Voltages, Vavg, of the Li/LiMSO4F Cell compound
method
LiFeSO4F
GGA
2.90
GGAþU 8
expt.
Vavg (V)
3.54 3.60
LiCoSO4F
GGA
3.56
LiNiSO4F
GGAþU GGA
4.73 4.23
GGAþU
5.16
experimental data. There are two different M sites in the compounds. Each M has two equatorial MF bond lengths, while the four oxygen atoms display two different MO bond lengths. Removal of lithium from LiMSO4F induces general shrink of the MO and MF bonds. As a result, the Fe(1)O and Co(1)O bonds tend to be more uniform, while the NiO4F2 octahedral of NiSO4F are still strongly distorted. This is because that the high spin Fe3þ and nonspin Co3þ ions are JahnTeller inactive while the low spin Ni3þ ions are still JahnTeller active. In contrast to the significant changes of MO and MF, the SO bonds of the compounds keep almost unchanged before/ after Liþ removal. This implies that the SO42- structure is rather stable which is good for the material to maintain structural stability during chargedischarge cycling. The theoretical intercalation voltage of a Li/LiMSO4F cell was calculated by a well developed method.20,21 Considering the following electrochemical reaction LiMSO4 F T MSO4 F þ Liþ þ e
ð1Þ
the average intercalation voltage, Vavg, can be determined by Vavg ¼
Figure 2. Total density of states (TDOS) and partial density of states (PDOS) of LiFeSO4F (left) and FeSO4F (right) computed with GGA þ U approximation. The black/red lines denote the spin-up/down DOS.
ΔG Δx
ð2Þ
where Δx refers to the number of Liþ ions transferred, here Δx = 1.0 for eq 1, ΔG is the difference in Gibbs free energy for the intercalation reaction. If we neglect the small changes in volume and entropy, ΔG can be approximated as the difference in the total energies between the LiMSO4F and the sum of MSO4F and metallic Li. The calculated average intercalation voltages of
LiMSO4F are shown in Table 4. One can see that GGA approximation underestimated the Vavg (2.90 V) of LiFeSO4F, while GGA þ U approximation gained a value of 3.54 V that coinciding well with the experimental result (3.6 V).8 This validates the strong correlated effects of the LiFeSO4F/FeSO4F system, as most of intercalation materials.22 Using GGA þ U approximation we obtained that the average intercalation voltages of LiCoSO4F and LiNiSO4F are 4.73 and 5.16 V, respectively. These voltages are much larger than that of LiFeSO4F. In addition, the calculation results show that the intercalation voltages of LiMSO4F are very close to those of LiMPO4 (M = Fe, Co, and Ni) phosphates.23 Barpanda et al. studied the electrochemical properties of LiCoSO4F and LiNiSO4F recently.9 They did not obtain any electrochemical activity in the voltage window of 2.54.2 V. Our work shows that the Liþ ions in LiCoSO4F and LiNiSO4F can be extracted by using a higher upper cutoff voltage. The left panel of Figure 2 plots the total density of states (TDOS) and partial density of states (PDOS) of LiFeSO4F. It has shown that the Fe2þ ion in LiFeSO4F has an electron configuration of t2g(v)3eg(v)2t2g(V)1. In the PDOS of Fe-3d, the valence bands between 7.8 and 1.0 eV are assigned to the majority spin t2g(v)3eg(v)2 electrons. They are hybridized with the O-2p bands. The hybridization between Fe-3d and F-2p is relatively weak perhaps due to the strong electronegativity of F. Since the Fe ion in LiFeSO4F has a minority spin t2g(V) electron, an additional Fe-3d band is filled, which can be seen between 0.5 and 0 eV. This Fe-3d band is known as the lower Hubbard band. The Fe-3d bands in the conduction band are found between 2.2 and 4.0 eV, which form the upper Hubbard bands. These features show that LiFeSO4F is most likely a MottHubbard (MH) insulator. The right panel of Figure 2 plots the TDOS and PDOS of FeSO4F. The changes in electronic structure with Liþ extraction can be clearly seen. The Fe-3d bands in the valence band shift to lower energies and now are found between 8.2 and 5.75 eV. The valence states near the Fermi level are now dominated by the O-2p states. This indicates that FeSO4F appears to be a charge7034
dx.doi.org/10.1021/jp111310g |J. Phys. Chem. C 2011, 115, 7032–7037
The Journal of Physical Chemistry C
ARTICLE
Figure 4. TDOS and PDOS of LiCoSO4F (left) and CoSO4F (right) computed with GGA þ U approximation. The black/red lines denote the spin-up/down DOS.
Figure 5. TDOS and PDOS of LiNiSO4F (left) and NiSO4F (right) computed with GGA þ U approximation. The black/red lines denote the spin-up/down DOS.
Figure 3. Electron clouds of FeSO4F between 6.45 and 5.75 eV. The three highly localized t2g orbitals of Fe can be clearly seen.
transfer (CT) insulator. Figure 3 shows the electron clouds of the Fe3d valence bands, from which one can see that the Fe-3d electrons
are highly localized. This will result in a strong coulomb interaction between the electrons. Recently, Kinyanjui et al. predicted a similar MH to CT transformation in the LiFePO4/FePO4 system by ab initio calculations, and then confirmed it by EELS experiments.24 Wu et al. also predicted by first-principles calculations that another Febased polyanion cathode material, Li2FeSiO4, transforms from MH insulator to CT insulator with Li extraction.25 This kind of MH to CT transition is of great interest for potential applications. But at present the physical mechanism of this transformation is still unclear, which is worth of study in future. The TDOS and PDOS of LiCoSO4F/CoSO4F and LiNiSO4F/NiSO4F are plotted in Figures 4 and 5, respectively. One can see that the Co-3d and Ni-3d majority spin bands of LiCoSO4F and LiNiSO4F are fully occupied, highlighting the 7035
dx.doi.org/10.1021/jp111310g |J. Phys. Chem. C 2011, 115, 7032–7037
The Journal of Physical Chemistry C
ARTICLE
Table 5. Values of Electron-Charge Transfer and Their Percentages (in Bracket) of the M-3d, M-4p, S-3p, O-2p, F-2p, and Li-2s States upon Liþ Extraction from LiMSO4F M-3d
M-4p
S-3p
O-2p
F-2p
Li-2s
LiFeSO4F/FeSO4F
0.430 (39%)
0.099 (9%)
0.044 (4%)
0.078 4 (28%)
0.131 (12%)
0.093 (8%)
LiCoSO4F/CoSO4F LiNiSO4F/NiSO4F
0.027 (3%) 0.053 (5%)
0.122 (13%) 0.117 (11%)
0.031 (3%) 0.038 (4%)
0.129 4 (56%) 0.152 4 (57%)
0.191 (21%) 0.163 (15%)
0.089 (10%) 0.084 (8%)
high spin state of Co2þ and Ni2þ. As for the delithiated compounds, the Ni-3d majority spin band of NiSO4F is partially occupied, while the Co-3d spin-up and spin-down subbands of CoSO4F are equivalently occupied. These features fit well with those of low spin Ni3þ and non spin Co3þ, respectively. The transformation from MH insultor to CT insulator that happens in LiFeSO4F/FeSO4F is not observed in the LiCoSO4F/CoSO4F and LiNiSO4F/NiSO4F systems. In order to explore the charge transfer mechanism accompanying with lithium extraction, we compare the values of electroncharge transfer of the M-3d, M-4p, S-3p, O-2p, F-2p, and Li-2s states upon Liþ extraction. The values of electron-charge transfer were obtained by calculating the differences of the integrated PDOS below the Fermi level between LiMSO4F and MSO4F. The results are listed in Table 5. The M-3d electron occupation numbers (not shown in the table) of LiMSO4F and MSO4F are both larger than those of ideal ionic M2þ and M3þ. This is a typical fact of transition metal compound for theoretical calculations, which always find some covalent contribution. Lithium is almost fully ionized so that its extraction effectively carries off an electron from the host material. For LiFeSO4F, about 39% of the electrons are donated by the Fe-3d state, 28% of the rest electrons are from oxygen, and others are from the S and F anions. This charge transfer mechanism is similar to the report of Ceder et al. that a part of charge transfer takes place on the oxygen anions in some layered cathode materials such as LiCoO2 and LiNiO2.20 The LiNiSO4F/NiSO4F system shows some differences comparing to LiFeSO4F/FeSO4F. Only 5% and 11% of electron-charge transfer takes place on the Ni-3d and Ni-4p bands. LiCoSO4F/CoSO4F can be regarded as an extreme case because all of the charge transfer takes place on the Co-4p bands instead of Co-3d. The valuse of electron-charge transfer taking place on the oxygen anions are 56% and 57% for LiCoSO4F/CoSO4F and LiNiSO4F/NiSO4F respectively, which are much larger than that of LiFeSO4/FeSO4F. The removal of such a lot of electrons from the O-2p band will result in a large amount of O anions. This process is accompanied with the following peroxide formation which leads to ultimate loss of oxygen from the lattice surface.26 2ðO2 Þ2 ¼ 2O2 þ O2 v
ð3Þ
The loss of oxygen is harmful to the cathode materials which may cause lattice collapse and safety problems.
4. CONCLUSIONS First-principles calculations showed that all of the LiMSO4F (M = Fe, Co, and Ni) compounds are in the high spin AFM ground state. After removal of Liþ from the compounds, Fe3þ still keeps the high spin state, while Ni3þ and Co3þ change to the low spin and nonspin states, respectively. The Fe-3d electrons of FeSO4F are highly localized, which results in a strong coulomb interaction between the 3d electrons. LiFeSO4F transforms from
Mott-Hubbard insulator to charge-transfer insulator with Liþ extraction. This phenomenon has been observed for some other Fe-based polyanion cathode materials such as LiFePO4 and Li2FeSiO4. The Liþ extraction induces general shrink of the MO and MF bonds. But the SO bonds keep almost unchanged. This helps the materials to maintain good structural stability during chargedischarge cycling. The calculation based on GGA þ U approximation obtained a theoretical intercalation voltage of 3.54 V for LiFeSO4F, which fits well with the experimental result. The theoretical intercalation voltages of LiNiSO4F and LiCoSO4F are much higher than that of LiFeSO4F. This indicates that they are potential high voltage cathode materials for lithium ion batteries. First-principles calculations showed a significant amount of electron-charge transfer takes place on the oxygen anions with Liþ extraction, especially for LiCoSO4F and LiNiSO4F. This will lead to ultimate loss of oxygen from the material lattice, which is not good for their applications as cathode materials due to the concerns of structural and thermal stabilities.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Phone and Fax: þ86 431 85155126.
’ ACKNOWLEDGMENT This work was sponsored by the Major State Basic Research Project of China (No. 2009CB220104) and the Program of New Century Excellent Talents (NCET-07-0366). It was also supported by the National Natural Science Foundation of China (No.11004073), Jilin Province (No.201101058), and Basic Research Fund of Jilin University(No.200903323). The authors thank Prof. Y. Ueda of the University of Tokyo for useful discussions. ’ REFERENCES (1) Mitzushima, K.; Jones, P. C.; Wiseman, P. J.; Goodenough, J. B. Mater. Res. Bull. 1980, 15, 783–789. (2) Thackeray, M. M.; Davida, W. I. F.; Brucea, P. G.; Goodenough, J. B. Mater. Res. Bull. 1983, 18, 461–472. (3) Padhi, A. K.; Nanjundaswamy, K. S.; Goodenough, J. B. J. Electrochem. Soc. 1997, 144, 1188–1194. (4) Nishimura, S. I.; Hayase, S.; Kanno, R.; Yashima, M.; Nakayama, N.; Yamada, A. J. Am. Chem. Soc. 2008, 130, 13212–13213. (5) Ellis, B. L.; Lee, K. T.; Nazar, L. F. Chem. Mater. 2010, 22, 691–714. (6) Whittingham, M. S. Chem. Rev. 2004, 104, 4271–4302. (7) Fergus, J. W. J. Power Sources 2010, 195, 939–954. (8) Recham, N.; Chotard, J. N.; Dupont, L.; Delacourt, C.; Walker, W.; Armand, M.; Tarascon, J. M. Nat. Mater. 2010, 9, 68–74. (9) Barpanda, P.; Recham, N.; Chotard, J. N.; Djellab, K.; Walker, W.; Armand, M.; Tarascon, J. M. J. Mater. Chem. 2010, 20, 1659–1668. (10) Meng, Y. S.; Dompablo, M. E. A. Energy Environ. Sci. 2009, 2, 589–609. 7036
dx.doi.org/10.1021/jp111310g |J. Phys. Chem. C 2011, 115, 7032–7037
The Journal of Physical Chemistry C
ARTICLE
(11) Liu, Z.; Huang, X. Solid State Ionics 2010, 181, 1209–1213. (12) Ramzan, M.; Lebegue, S.; Ahuja, R. Phys. Rev. B 2010, 82, 125101–15. (13) Clark, S. J.; Segall, M. D.; Pickard, C. J.; Hasnip, P. J.; Probert, M. J.; Refson, K.; Payne, M. C. Z. Kristallogr. 2005, 220, 567–570. (14) Vanderbilt, D. Phys. Rev. B 1990, 41, 7892–7895. (15) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865–14. (16) Cococcioni, M.; de Gironcoli, S. Phys. Rev. B 2005, 71, 035105–116. (17) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188–5192. (18) Pfrommer, B. G.; Cote, M.; Louie, S. G.; Cohen, M. L. J. Comput. Phys. 1997, 131, 233–240. (19) Sebastian, L.; Gopalakrishnan, J.; Piffard, Y. J. Mater. Chem. 2002, 12, 374–377. (20) Aydinol, M. K.; Kohan, A. F.; Ceder, G.; Cho, K.; Joannopoulos, J. Phys. Rev. B 1997, 56, 1354–1365. (21) Courtney, I. A.; Tse, J. S.; Mao, O.; Hafner, J.; Dahn, J. R. Phys. Rev. B 1998, 58, 15583–15588. (22) Zhou, F.; Cococcioni, M.; Marianetti, C. A.; Morgan, D.; Ceder, G. Phys. Rev. B 2004, 70, 235121–18. (23) Zhou, F.; Cococcioni, M.; Kang, K.; Ceder, G. Electrochem. Commun. 2004, 6, 1144–1148. (24) Kinyanjui, M. K.; Axmann, P.; Wohlfahrt-Mehrens, M.; Moreau, P.; Boucher, F.; Kaiser, U. J. Phys.: Condens. Matter 2010, 22, 275501–18. (25) Wu, S. Q.; Zhu, Z. Z.; Yang, Y.; Hou, Z. F. Comput. Mater. Sci. 2009, 44, 1243–1251. (26) Goodenough, J. B.; Kim, Y. Chem. Mater. 2010, 22, 587–603.
7037
dx.doi.org/10.1021/jp111310g |J. Phys. Chem. C 2011, 115, 7032–7037