First principle study of LiXS2 (X = Ga, In) as cathode materials for Li ion batteries
Rao Feng-Ya1, Ning Fang-Hua1, Jiang Li-Wei1, Zeng Xiang-Ming2, Wu Mu-Sheng1, Xu Bo1, Ouyang Chu-Ying1, †,
Department of Physics, Laboratory of Computational Materials Physics, Jiangxi Normal University, Nanchang 330022, China
School of New Energy Science and Engineering, Xinyu University, Xinyu 338004, China


† Corresponding author. E-mail:

Project supported by the National High Technology and Development Key Program, China (Grant No. 2015AA034201), the National Natural Science Foundation of China (Grant Nos. 11234013 and 11264014), the Natural Science Foundation of Jiangxi Province, China (Grant Nos. 20133ACB21010, 20142BAB212002, and 20132BAB212005), and the Foundation of Jiangxi Provincial Education Committee, China (Grant Nos. GJJ14254 and KJLD14024).


From first principle calculations, we demonstrate that LiXS2 (X = Ga, In) compounds have potential applications as cathode materials for Li ion batteries. It is shown that Li can be extracted from the LiXS2 lattice with relatively small volume change and the XS4 tetrahedron structure framework remains stable upon delithiation. The theoretical capacity and average intercalation potential of the LiGaS2 (LiInS2) cathode are 190.4 (144.2) mAh/g and 3.50 V (3.53 V). The electronic structures of the LiXS2 are insulating with band gaps of 2.88 eV and 1.99 eV for X = Ga and In, respectively. However, Li vacancies, which are formed through delithiation, change the electronic structure substantially from insulating to metallic structure, indicating that the electrical conductivities of the LiXS2 compounds should be good during cycling. Li ion migration energy barriers are also calculated, and the results show that Li ion diffusions in the LiXS2 compounds can be as good as those in the currently widely used electrode materials.

1. Introduction

It is now universally recognized that lithium ion batteries (LIBs), with advantages like high energy density and long cycling life, are one of the most important energy storage sources and being widely used in various consumer electronics industries. The most promising application of the LIBs in the near future is in the power supplies for electrical vehicles (EVs).[13] However, the comprehensive performance of the currently commercialized LIBs is difficult to fully satisfy the market demands of the EV industry in terms of the energy density, cycling life and safety. Many efforts, either experimentally or theoretically, have been made to improve the comprehensive performance of the LIBs.[47] As for the safety problem, the inflammable organic liquid electrolyte is (at least partly) responsible for most of the fire or explosion accidents occurring in either the consumer electronic and power LIBs. One alternative is to develop the all solid state LIBs, in which non-inflammable solid state electrolytes are used to replace the liquid electrolyte and therefore the safety of the battery system is enhanced.[8] In addition to the better safety, Li metal can be directly used as anode for all solid state LIBs, which is beneficial to enhancing the energy density of the battery system. On the other hand, Li metal anode allows us to select the cathode with more flexibilities, because Li anode can be used as the Li source and thus the cathode material does not necessarily contain Li element. Furthermore, all-solid-state Li battery has a wider electrochemical window and the solid/solid interface is electrochemically more stable than the solid/liquid interface. These features are beneficial to obtaining full utility of the cathode’s capacity.[9,10]

In order to accelerate the development of all-solid-state LIBs, it is important to discover or design novel solid state electrolyte materials that possess both good electrochemical stability and high Li ion conductivity. There are three types of solid state electrolyte materials that are reported in the literature. The most widely studied type is the inorganic crystalline solid state electrolyte, which includes the perovskite structured materials,[11] the garnet structured materials,[12,13] the NASICON,[14] and LISICON[15] structured materials, etc. Generally speaking, these inorganic solid state electrolytes are electrochemically very stable with high electrochemical window while the Li ion conductivity is not good enough to satisfy the industrial application, particularly the Li diffusion across the solid electrode/electrolyte interface leads to large interfacial impedance. Polymer electrolyte (PEO) is another type of solid state electrolyte and has been widely studied in the literature in the past years.[16,17] Unlike the inorganic solid state electrolyte, the plasticity of PEO is very good and therefore the electrode/electrolyte contact is good with PEO electrolyte. However, the Li ionic conductivity in PEO ranges from 108 S/cm to 10−4 S/cm at temperatures between 40 °C and 100 °C,[18] and thus the allsolidstate LIBs with using PEO as electrolyte must be operated under high temperatures.[19] The third type of solid state electrolyte is non-crystalline glassy sulfide material that contains Li2S and other sulfide as the main component. The non-crystalline sulfide electrolyte has higher Li ion conductivity than the crystalline oxide electrolyte in general, while this type of material is very reactive to H2O molecule in air and therefore becomes very unstable in air.[20,21]

Recently, Kamaya et al.[22] reported a novel crystalline solid state electrolyte Li10GeP2S12 (LGPS), with an electrochemical window width of 5 V and Li ion conductivity reaches 1.2 × 10−2 S/cm under room temperature. This value is the highest in Li ion conductivities ever reported so far in the literature for solid state electrolytes. As a promising solid state electrolyte, the lithium ion diffusion mechanism,[23,24] the phase and structural stability,[25] the mechanical properties,[26] and the electrochemical performances of LGPS[27] have been studied recently.

Although the LGPS has very good Li ionic conductivity, there is still a long way to go before it can be widely used in the all-solid-state LIBs. The interfacial problem at the electrode and the electrolyte boundary occurs for all kinds of sulfide solid state electrolytes.[28] The electrode/electrolyte interface in an Li/LGPS/Li half battery reduces the Li ion conductivity by more than one order of magnitude, indicated from the ion conductivity measurement of the bulk material.[27] More importantly, from density functional theory calculations, it is shown that the LGPS is electrochemically unstable at the electrode/electrolyte interface.[29] The interfacial reaction decomposes the LGPS in the interface area, and therefore the electrochemical performance of the battery becomes deteriorated. Technically, a thin buffer layer (several nanometers thick) between the oxide cathode and the sulfide solid state electrolyte can reduce the interfacial resistance and hence the rate capacity can be increased. Those buffer layers include LiNbO3[30,31] and Li2O-SiO2 glasses,[32] and Li4Ti5O12,[33] while the oxide cathodes include LiCoO2 and LiMn2O4.[34] The improved electrochemical performance via introducing the thin buffer layer is attributed to the lowering of the space-charge potential in the interface area.

As discussed above, there are two major problems concerning the electrode/electrolyte interface, which are recognized in the literature, namely, the reaction of sulfide solid electrolyte with the electrode, and the formation of space-charge layer in the interface region. These problems block the way of rapidly developing the all-solid-state LIB with using sulfide electrolytes. Although coating can lower the space-charge potential and prevent the reaction between the sulfide electrolyte and the cathode, it makes the manufacture procedure more complicated and the cost increased. Another strategy is to find novel cathode materials that are more compatible with the sulfide electrolyte. If we can use Li sulfide compound cathode rather than oxide cathode, the interfacial reaction and the space-charge induced Schottky barrier can be substantially reduced.

With the development of the computer simulation techniques, it is possible to screen novel materials using high-throughput computational technology. Materials Genome Initiative (MGI) combines the high-throughput computational materials sciences with available experimental data to accelerate the development of novel materials. Using the main idea of the MGI, Hautier et al. identified two novel cathode materials with capacity values higher than 200 mAh/g.[35] In the present paper, from high-throughput first principle calculations, we search for all possible Li sulfide compounds. Selection criteria include those for the theoretical capacity, the intercalation voltage, the electronic and Li ion conductivity, and the structural stability.[36] We identify two Li sulfide compounds that have potential application as cathode material for all-solid-state LIB with sulfide electrolyte.

2. Screening criteria and computational parameters

In order to use the MGI idea to fast screen the physical and chemical properties of the Li compounds using the high-throughput computational techniques, we develop a computational platform that supports the first principle calculation and the output result analysis spontaneously. The architecture of the platform and the details on the data collection and data mining are introduced in Ref. [37]. The screening procedure and criteria for Li ion battery materials are introduced in Ref. [36]. In the present work, we select all Li sulfide compounds as candidate from the inorganic crystal structure database (ICSD). The theoretical Li storage capacity and the average intercalation potential criterion are larger than 140 mAh/g and higher than 3.0 V. Then the structural stability is evaluated through analyzing the structural change and calculating the volume change upon Li extraction and insertion. If these criteria are satisfied, the Li ion migration energy barriers and the electronic structures are calculated to evaluate the Li ion diffusion and electrical conduction. Finally, we identify that two compounds, namely the LiGaS2 and LiInS2 satisfy all criteria and have potential applications as cathode materials for allsolidstate LIBs.

First principle calculations are performed using the Vienna ab initio simulation package (VASP) within the projector augmented-wave (PAW) approach.[38] The ground state of the electronic structure is described with density functional theory (DFT) using the generalized gradient approximation (GGA) with PW91 exchange correlation functional.[39] The wave functions are expanded into plane waves and the cutoff energy is selected to be 520 eV. The calculation of density of states (DOS) is smeared by the Gaussian smearing method with a smearing width of 0.02 eV. The Li migration pathway and energy barrier are optimized by the nudged elastic band (NEB) method.[40]

The average lithium extraction potential is calculated from[4143]

where ΔG is the variation of Gibbs free energy of the electrochemical charge/discharge reaction, F is the Faraday constant, and n is the mole number of Li ion participating in the reaction. In addition to the inner energy, the Gibbs free energy also includes contributions from the variations of volume and entropy, which are negligibly small and not included in the present work. Therefore, the average intercalation potential can be approximately obtained by calculating the total energy change from

where ΔE is the total energy difference.

3. Properties and performances of the LiGaS2 and LiInS2 compounds
3.1. Energetics and structures of the LiXS2 (X = Ga, In)

From our first principle calculation, we identify that all Li ions can be delithiated from the LiXS2 lattice and therefore the theoretical Li storage capacities of the two materials are calculated to be 190.4 mAh/g and 144.2 mAh/g for LiGaS2 and LiInS2, respectively. On the other hand, the average intercalation potentials are calculated to be 3.50 V and 3.53 V versus Li/Li+ for the LiGaS2 cathode and LiInS2 cathode, respectively. Therefore, the corresponding energy densities of the two cathode materials are 665.6 Wh/kg and 468.7 Wh/kg, which are comparable to those of the currently widely used oxide cathodes like spinel structured LiMn2O4 material (475 Wh/kg) and the olivine structured LiFePO4 material (586 Wh/kg).

The LiGaS2 and LiInS2 compounds were synthesized several decades ago.[44,45] The crystal structures of compounds LiXS2 (X = Ga, In) both belong to the orthogonal crystal system with space group of Pna21[45,46] as shown in Fig. 1. All the atoms occupy 4a sites, and the Li atoms and In atoms are all located in the centers of the tetrahedrons while the S atoms at the vertex sites. These orthorhombic crystals are known as good infrared nonlinear optical crystals with unique optical properties such as high thermal conductivities, wide band gaps, and low anisotropies of linear thermal expansion coefficient and wide opportunities of phase-matching conditions.[46,47] Recently, the piezoelectric, elastic and dielectric properties of these materials have been measured, in which piezoelectric crystal is considered to be applicable to pressure sensors, ultrasonic transducers for medical and sonar imaging, etc.[4850] In the present work, it is the first time to report that they can be used as the electrode materials for Li ion batteries.

Fig. 1. Optimized structures of the LiXS2 (X = Ga, In) compounds before ((a), (c)) and after ((b), (d)) delithiation.

In Table 1 summarized are the theoretically optimized lattice parameters before and after Li extraction from a 2 × 2 × 2 supercell. They keep their orthorhombic structures and the XS4 tetrahedrons remain stable upon relaxation after Li atoms have been fully delithiated, indicating that the delithiated structure is at least a metastable structure and can be stable upon electrochemical delithiation. The volume changes of the two compounds from lithiated state LiXS2 to the delithiated state Li0XS2 are −20.6% and −17.22% for X = Ga and In, respectively. Comparing with the volume changes of other cathode materials upon lithiation, the volume change of the LiXS2 cathode is not high, indicating that the structural stability of the LiXS2 cathode is good upon charge/discharge cycling. On the other hand, unlike the lattice expansion upon delithiation for the LiCoO2 cathode, the volume decreases upon delithiation (charging of the battery) for the case of LiXS2 cathode. This is beneficial to the battery design of the all-solid- state Li battery system in which Li metal is generally used as anode material, whose volume expands in the charging process. Therefore, the volume decrease at the cathode side compensates for the volume expansion at the anode side.

Table 1.

Theoretical lattice parameters for a 2 × 2 × 2 supercell of the LiXS2 (X = Ga, In).

3.2. Electronic structures of the LiXS2 (X = Ga, In)

To achieve high charge and discharge speed, the electrical conductivity values of the electrode materials of the Li ion batteries should be high. In the present work, we calculate the electronic structures of the LiXS2 cathodes (including their delithiated states Li0XS2) and the electrical conductivity values can be evaluated qualitatively from the analyses of the electronic structures. The calculated total density of states (TDOS) is given in Fig. 2. As shown in the figure, the LiXS2 compounds have insulating electronic structures with band gaps of 2.88 eV and 1.99 eV for X = Ga and In, respectively. Those values are smaller than the experimental measured values of 4.15 eV (LiGaS2) and 3.57 eV (LiInS2),[46] as DFT always underestimates the band gaps of insulators. On the other hand, our predicted band gap is in agreement with other theoretical result. Ma et al.[51] calculated the electronic structure of the LiGaS2 compound from the GGA-PBE level and they obtained a band gap of 3.12 eV. From LDA + U calculations, Atuchin et al. obtained the band gap of the LiGaS2 to be about 3.28 eV.[52] Those DFT calculated band gaps of the LiGaS2 from the literature are close to the one obtained in the present work, while much smaller than the experimentally observed value.

Fig. 2. The TDOSs of the LiGaS2 (a) and LiInS2 (b) cathode materials before (solid black lines) and after (dashed blue lines) delithiation. The Fermi level is set to be 0 eV.

The insulating electronic structure indicates the poor electrical conductivity, which is not good for the rate performance. However, this is a common problem that has already existed for most of the other cathode materials. For example, it is shown that the electronic structures of the LiCoO2 and LiMn2O4 cathodes with band gaps of 2.7 eV[53] and 1.2 eV,[54] respectively are also insulating. Furthermore, the band gap of the olivine structured LiFePO4 cathode is even as high as 3.7 eV.[55]

Technically, defects can change the electronic structures of these insulating materials, thus helping to improve their electrical conductivities. In the present work, we show that Li vacancies, which are created in the delithiation process, can change the electronic structures of the LiXS2 compounds substantially from insulating to metallic structure. As shown in Fig. 2, the electronic structures of the fully delithiated states Li0XS2 are metallic. Therefore, it is expected that the electrical conductivity can be enhanced in the delithiation process. Further analysis of the projected density of states (PDOS) shows that the p-states of the S atoms contribute to electronic states near the Fermi level as shown in Fig. 3. Here we mention that the small contributions of the electronic states near the Fermi level from the Ga/In atoms presented in Fig. 3 are due to the technical errors when we calculate the PDOSs, which are obtained by integrating the wave functions in a sphere with a specific radius. In order to have an accurate description of the electronic states near the Fermi level, we calculate the band decomposed charge density within an energy range from −0.5 eV to 0.5 eV around the Fermi level, which is the wave function in a specific energy range. As shown in the insets of Fig. 3, the states near the Fermi level all originate from S atoms and have dumbbell shapes, indicating that they are typical p states.

Fig. 3. Projected density of states of the delithiated states of Li0GaS2 (a) and Li0InS2 (b). The insets show the band decomposed charge density contours, in which the small (yellow) and large (light and dark brown) spheres denote S and Ga (In) atoms, respectively. The charge density contour is in blue color with the shape of a dumbbell. The isosurface value is 0.025 e/Å3.
3.3. Lithium ion migration

The rate performance of an electrode material is also strongly related to the Li ion diffusion behavior in the bulk material. From the analyses of the Li ion distribution and the symmetry of the crystalline structure, we figured out two independent Li diffusion pathways (path-I and path-II in Fig. 4) in the lattice, which interconnect the Li sites in different planes and thus a three-dimensional diffusion network can be formed by the two pathways in the bulk material. Path-I is within the ab-plane while Path-II is within the bc-plane of the lattice. The Li ion migration pathways are then optimized by the NEB method and the calculated energy barriers are given in Fig. 4. As shown in the figure, the energy barriers for Li ion migration along path-I are 0.325 eV and 0.478 eV for LiGaS2 and LiInS2, respectively. On the other hand, Li ion migration energy barrier along path-II is higher than that along path-I. The corresponding energy barriers are 0.559 eV and 0.998 eV along path-II for LiGaS2 and LiInS2, respectively. By comparing the Li diffusion pathways and the energy barrier values with those of the currently widely used electrode materials, we can conclude that the Li ion diffusivity values in the LiXS2 cathode are acceptable. For example, the Li ion diffusion in the LiCoO2 cathode with layered structure is two-dimensional and the energy barriers are reported to be in a range from 0.39 eV[56] to 0.83 eV.[57] On the other hand, Li diffusion in LiFePO4 is one-dimensional, and the energy barrier is reported to be 0.6 eV.[58] Li ion diffusion pathways in spinel structures are three-dimensional and the energy barrier of Li ion migration in LiMn2O4 cathode and Li4Ti5O12 anode materials are 0.23[59] and 0.70 eV,[60] respectively. It is shown that the Li ion migration energy barriers of the LiXS2 cathodes are comparable to those of the commercially used electrode materials. Therefore, we believe that the Li ion diffusivity values of the LiXS2 cathodes can satisfy the requirements for commercial applications.

Fig. 4. Energy profiles for Li ion migration along the two independent pathways in (a) LiInS2-path-I, (b) LiInS2-path-II, (c) LiGaS2-path-I, and (d) LiGaS2-path-II. The optimized pathways are also presented schematically in the insets with ball and stick models. The large spheres (brown), middle spheres (yellow), and small spheres (purples) represent In, S, and Li atoms, respectively. The Li pathways are given by green spheres (intermediate states) and red (initial and final states) spheres.
4. Summary and conclusions

In this study, two compounds LiXS2 (X = Ga, In) are identified as potential cathode materials for Li ion batteries. The first principle calculations show that their basic physical and chemical properties, including the structural stabilities, the electrical and Li ion conductivities, the intercalation potentials, are suitable for use as cathode materials. As the LiXS2 are sulfide, they are particularly suitable for all-solid-state LIBs with using sulfide solid state electrolyte. From the obtained theoretical Li storage capacities and the average intercalation potential, the corresponding energy densities of the two cathode materials are calculated to be 665.6 Wh/kg and 468.7 Wh/kg, which are comparable to those of currently widely used oxide cathodes. Although the electronic structures of the LiXS2 compounds are insulating, their electrical conductivities can be improved substantially upon delithiation, as the electronic structure changes from insulating to metallic structure when Li vacancies are induced. Furthermore, from the calculated Li ion migration energy barriers in the LiXS2 compounds, we identify a three-dimensional Li diffusion network in each of these materials with energy barriers that are comparable to those of the currently widely used electrode materials, indicating that the Li ion migration dynamic performances of the LiXS2 compounds are acceptable for the cathode materials.

1Goodenough J BKim Y S 2010 Chem. Mater. 22 587
2Tarascon J MArmand M 2001 Nature 414 359
3Huang X J 2015 Physics 44 1 (in Chinese)
4Ouyang C YWang D YShi S QWang Z XLi HHuang X JChen L Q 2006 Chin. Phys. Lett. 23 61
5Wang Z PLiu PWang L F 2013 Chin. Phys. B 22 088801
6Li J Ru Q Hu S J Guo L Y 2014 Acta Phys. Sin. 63 168201 (in Chinese)
7Zhou XChen JGu LMiao L 2015 Chin. Phys. Lett. 32 26102
8Huang W WZhu Z QWang L JWang S WLi HTao Z LShi J FGuan L HChen J 2013 Angew. Chem. Int. Ed. 52 9162
9Tian JJin YGuan Y BSu Y FBao L YChen SWu F 2014 Chin. Sci. Bull. 59 1950
10Baek S WLee J MKim T YSong M SPark Y 2014 J. Power Sources 249 197
11Zhao Y SDaemen L L 2012 J. Am. Chem. Soc. 134 15042
12Murugan RThangadurai VWeppner W 2007 Angew. Chem. Int. Ed. 46 7778
13Cao YLi Y QGuo X X 2013 Chin. Phys. B 22 078201
14Feng J KYan B GLiu J CLai M OLi L 2013 Mater. Tech. 28 276
15Kanno RMurayama M 2001 J. Electrochem. Soc. 148 A742
16Stephan A M 2006 European Polymer Journal 42 21
17Fergus J W 2010 J. Power Sources 195 4554
18Song J JWang YWan C C 1999 J. Power Sources 77 183
19Fontenella J JWintergill M CCalame J PAndeen C G 1983 Solid State Ionics 8 333
20Mizuno FHayashi ATadanaga KMasahiro T 2006 Solid State Ionics 177 2721
21Liu Z QTang Y FWang Y MHuang F Q 2014 J. Power Sources 260 264
22Kamaya NHomma KYamakawa YHirayama MKanno RYonemura MKamiyama TKato YHama SKawamoto KMitsui A 2011 Nat. Mater. 10 682
23Adams SRao R P 2012 J. Mater. Chem. 22 7687
24Hu C HWang Z QSun Z YOuyang C Y 2014 Chem. Phys. Lett. 591 16
25Ong S PMo Y FRichards W D 2013 Energy Environ. Sci. 6 148
26Wang Z QWu M SLiu GLei X LXu BOuyang C Y2014Int. J. Electrochem. Sci.9562
27Wang Y MLiu Z QZhu X LTang Y FHuang F Q 2013 J. Power Sources 224 225
28Kobayashi TYamada AKanno R 2008 Electrochim. Acta 53 5045
29Mo Y FOng S PCeder G 2012 Chem. Mater. 24 15
30Ohta NTakada KSakaguchi IZhang L QMa R ZFukuda KOsada MSasaki T 2007 Electrochem. Commun. 9 1486
31Ogawa MKanda RYoshida KUemura THarada K 2012 J. Power Sources 205 487
32Sakuda AKitaura HHayashi ATadanaga KTatsumisago M 2008 Electrochem. Solid-State Lett. 11 A1
33Ohta NTakada KZhang L QMa R ZOsada MSasaki T 2006 Adv. Mater. 18 2226
34Takada KOhta NZhang L QXu X XHang B TOhnishi TOsada MSasaki T 2012 Solid State Ionics 225 594
35Hautier GJain AChen H LMoore COng S PCeder G 2011 J. Mater. Chem. 21 17147
36Ouyang C YChen L Q 2013 Sci. Chin.-Phys. Mech. & Astron. 56 2278
37Xiao J MWang HOuyang C Y20152015 J. Jiangxi Norm. Univ. (Nat. Sci. Ed.) 3973(in Chinese)
38Blochl P E 1994 Phys. Rev. B 50 17953
39Wang YPerdew J P 1991 Phys. Rev. B 44 13298
40Henkelman GJónsson H 2000 J. Chem. Phys. 113 9978
41Aydinol M KKohn A FCeder GCho KJoannopoulos J 1997 Phys. Rev. B 56 1354
42Ouyang C YZhong Z YLei M S 2007 Electrochem. Commun. 9 1107
43Zhong Z YOuyang C YShi S QLei M S 2008 Chem. Phys. Chem. 9 2104
44Leal-Gonzalez JMelibary S SSmith A J 1990 Acta Cryst. C 46 2017
45Boyd G DKasper H MMcFee J H 1973 J. Appl. Phys. 44 2809
46Isaenko LVasilyeva IMerkulov AYelisseyev ALobanov S 2005 J. Cryst. Growth 275 217
47Isaenko L IVasilyeva I G 2008 J. Cryst. Growth 310 1954
48Lagoun BBentria TBentria B 2013 Comput. Mater. Sci. 68 379
49Badikov V VChizhikov V IEfimenko V V 2003 Opt. Mater. 23 575
50Ma T HZhuang Z PRen Y L2012Acta Phys. Sin.61208401(in Chinese)
51Ma T HYan C HXie YSun LLv W QWang RRen Y L 2010 Physica B 405 363
52Atuchin V VIsaenko L IKesler V GLobanov SHuang HLin Z S 2009 Solid State Commun. 149 572
53Elp van JWieland J LEskes HKuiper PSawatzky G AGroot de F M FTurner T S 1991 Phys. Rev. B 44 6090
54Ouyang C YDeng H DYe Z QLei M SChen L Q 2006 Thin Solid Films 503 268
55Zhou FKang K SMaxisch TCeder GMorgan D 2004 Solid State Commun. 132 181
56Ning F HLi SXu BOuyang C Y 2014 Solid State Ionics 263 46
57Van der Ven ACeder GAsta MTepesch P D 2001 Phys. Rev. B 64 184307
58Ouyang C YShi S QWang Z XHuang X JChen L Q 2004 Phys. Rev. B 69 104303
59Ouyang C YShi S QWang Z XHuang X JChen L Q 2004 Europhys. Lett. 67 28
60Chen Y COuyang C YSong L JSun Z L 2011 Electrochim Acta 56 6084