Article pubs.acs.org/JPCC
Comparative Study of Tavorite and Triplite LiFeSO4F as Cathode Materials for Lithium Ion Batteries: Structure, Defect Chemistry, and Lithium Conduction Properties from Atomistic Simulation Sanghun Lee*,† and Sung Soo Park*,‡ †
Department of BioNano Technology, Gachon University, Seongnam, Gyunggido 443-803, Republic of Korea Corporate R&D Center, Samsung SDI Co. Ltd., Yongin, Gyunggido 446-577, Republic of Korea
‡
S Supporting Information *
ABSTRACT: To explore the possibility of LiFeSO4F with two polymorphs (tavorite and triplite) as the cathode material for lithium ion batteries, structure, defect chemistry, and Li+ ion conduction properties are studied by atomistic simulation with empirical potential parameters. We investigate the correct structure of tavorite LiFeSO4F, which was newly determined. The concentration of intrinsic defects in the tavorite form is very low in comparison with the triplite form. Configurations of FeO4F2 octahedra in the triplite form are in favor of cornersharing connections over edge-sharing ones. Even though calculated migration energies prove that both isomorphs are Li+ ion conductors, the Li+ ions in the triplite LiFeSO4F move in the restricted migration paths (one- or two-dimensional), whereas the tavorite isomorph has a continuous three-dimensional Li+ ion conducting network.
1. INTRODUCTION For the last two decades, layered lithium transition metal oxides (LiMO2, M: transition metal, that is, Co, Ni, Mn, etc.) have been commonly used for cathode materials in lithium ion batteries (LIBs) for portable electronic devices. However, as the usage of LIBs has increased in other areas, such as, transport applications and energy storage systems for renewable energy, increasing demands for cheaper, safer, and more environmentally benign LIBs have prompted the development of new cathode materials. Several new materials have been explored as potential alternatives, including those with polyoxyanions of (XO4)x− (X: S, P, Si, etc.). Owing to the inductive effect of the polyoxyanions, the polyoxyanion materials exhibit relatively high redox potential as well as good thermal stability. In particular, olivine-structured LiFePO4 with redox potential 3.4 V (vs Li), has been largely investigated for the past decade as one of the most promising cathode materials.1−3 However, LiFePO4, which has 1D channels for lithium ion diffusion, shows intrinsically low ionic and electronic conductivity, necessitating nanostructural modification or conductive coating. Recently, this problem has been partly improved by synthesizing tavorite LiFeSO4F, which contains electron-withdrawing sulfate groups and fluorine atoms.4,5 This tavorite material exhibits a slightly higher voltage of 3.6 V and suppresses the need for nanosizing and conductive coating. Meanwhile, Barpanda et al. reported that its manganese homologue LiMnSO4F, which adopts triplite structure, does not show any electrochemical activity.6 In this triplite structure, Li and Mn are almost randomly distributed over two crystallographically different sites, which could be considered as a © 2014 American Chemical Society
possible origin of inactivity of LiMnSO4F. However, this triplite structure is maintained on Fe substitution up to 80−90% and the materials show good redox activity at the highest potential ever reported for the Fe3+/Fe2+ redox couple (3.9 V).6 This indicates that the electrochemical inactivity of LiMnSO4F may not originate from the lithium diffusion kinetics in the disordered structure of the triplite form but from the high operating voltage of Mn3+/Mn2+ redox couple. A similar observation, electrochemically active for Fe compound and inactive for Mn one, was reported in the study of pyrophosphate materials of Li2FeP2O7 and Li2MnP2O7.7 Following Barpanda et al.’s study,6 various synthetic strategies have been suggested to control the structure-type of products and to obtain pure triplite LiFeSO4F.8−11 In order to determine the origin of the high voltage of the triplite over the tavorite form of LiFeSO4F, a few first-principles calculations were recently performed.12,13 In Chung et al.’s study, it was reported that the difference in voltages is mainly due to the difference in stabilities of the delithiated stages (FeSO4F), that is, the triplite FeSO4F is less stable than the tavorite, which is due to strong Fe3+···Fe3+ repulsion from the edge-sharing geometry of FeO4F2 octahedra.12 Meanwhile, Yahia et al. suggested that the difference of destabilizing effect between the two polymorphs is due to the configurational difference of fluorine atoms between the delithiated triplite (cis F−F) and tavorite (trans F−F) forms.13 In addition, from an Received: March 18, 2014 Revised: May 17, 2014 Published: May 23, 2014 12642
dx.doi.org/10.1021/jp502672k | J. Phys. Chem. C 2014, 118, 12642−12648
The Journal of Physical Chemistry C
Article
experimental calorimetric study, Radha et al. reported that the tavorite is energetically more stable than the triplite, however, the transformation of the tavorite to triplite is thermodynamically favored at the high temperatures because of entropy effect.14 To understand the factors influencing the electrochemical behavior of cathode materials, it is desirable to investigate the structure, defect chemistry, and transport properties at the atomic level with molecular simulation. Particularly, the classical atomistic simulation has an advantage of accessibility to comparatively larger systems than the first-principles calculation. Therefore, from the classical atomistic simulation, many cathode materials including polyoxyanions for LIBs, such as sulfates,15 phosphates,16−21 and silicates,22−24 have been successfully simulated. Recently, the tavorite LiFeSO4F and NaFeSO4F were investigated from the classical atomistic simulation. In the study of Tripathi et al., they calculated activation energies of many possible migration paths of alkaline atoms and found that the tavorite LiFeSO4F is a 3D lithium ion conductor with an activation energy of ∼0.4 eV for long-range diffusion.15 However, it has been found that the investigated structure of LiFeSO4F in their study is inaccurate,25 therefore, the current study treats the corrected structure and extends the work to comparative study of the structures and Li+ ion migration between the tavorite and triplite polymorphs of LiFeSO4F.
Figure 1. Unit-cell structure of (a) tavorite and (b) triplite LiFeSO4F.
corner-sharing type connection is most stable in enthalpy and well-matched with the experimental unit-cell parameters. More details are shown in Results and Discussion. The energies of defects and lithium ion migrations were calculated by the Mott-Littleton method.30 This method is to partition the crystal lattice surrounding defect into two spherical regions. In the inner sphere, the ions are largely affected by the defect so that the interactions are explicitly treated and the ions are allowed to fully relax. In contrast, the ions in the outer region are implicitly treated as a dielectric continuum. All calculations were performed by the GULP code.31
2. SIMULATION METHODS The simulation methodology employed in this work has been applied to a wide range of inorganic solids, including cathode materials for LIBs.15−24 Hence, only a brief description is given here and, for further information, the reader is directed to other references.26,27 Interactions between ions are composed of long-range Coulombic and short-range nonbonded interaction components. The short-range interactions were modeled using the Buckingham potential function. Additionally, the Morse potential was also supplemented for the interactions between oxygen and sulfur atoms. To account for the bending interactions of O−S−O angles in the SO4 units, the threebody interaction term was included. The electronic polarization of Fe3+ and F− was modeled by the core−shell model.26,28 The potential parameters in this study are provided in Supporting Information as originally published in Gardiner’s thesis.29 Using these parameters, Tripathi et al.15 successfully reproduced structural properties of the tavorite LiFeSO4F in good agreement with the structure proposed by the X-ray powder diffraction,5 in which two different sites are occupied by Li+ ions with half-occupancy. However, last, a more accurate structure, in which the Li+ ions occupy in one single site with full-occupancy, was revealed from the high resolution neutron diffraction.25 Therefore, we investigated the corrected structure in the present study (Figure 1a). In the unit cell of the triplite LiFeSO4F, 8 Fe3+ and 8 Li+ ions are randomly distributed over two independent 8f Wyckoff positions (denoted as sites 1 and 2, Figure 1b). Li+ content in site 2 is slightly larger than that in site 1 (∼57%).10 The fully disordered triplite LiFeSO4F corresponds to a mixing of 4900 configurations, which is almost impossible to compute. Hence, Yahia et al. chose a set of core-distributions that can be considered as representative configurations of triplite in terms of connection types of FeSO4F octahedra (i.e., corner-sharing and edge-sharing types, Figure 2) in their first-principles calculation study.13 From our calculation, the structure of pure
3. RESULTS AND DISCUSSION Tavorite LiFeSO4F. To determine the unit-cell parameters, the structure of tavorite LiFeSO4F was optimized under the constant pressure conditions (0 Pa) with the P1̅ symmetry constraint. The lattice parameters and ion positions were allowed to relax. The calculated unit-cell parameters, which are highly comparable with the experimental values,5 are shown in Table 1. The successful reproduction of the complex structure of tavorite LiFeSO4F (within ∼1.5%) supports that the potential models and parameters used in this study are sufficiently reliable to characterize various properties of LiFeSO4F in our interests. As shown in Figure 1a, FeO4F2 octahedra coner-shared through F atoms run along the c-axis, while the four oxygen atoms in each octahedron are bonded to sulfur atoms, therefore, the SO4 tetrahedra have a role of crosslinking of FeO4F2 octahedra in the other directions. Note that the 3D Li+ ion channels in the supercells are well-secured by several combinations of Li+ ion paths (Figure 3). The details of Li+ ion migration are described in a later part of Li+ ion migration. Understanding of defect chemistry is necessary due to the fact that defects in electrode materials have influence on electrochemical behaviors. In addition, it is also useful for developing new entropy-driven synthesis methods, in which the defects in disordered phase promote atomic diffusion.9 To determine Frenkel defect formation energies, isolated defect 12643
dx.doi.org/10.1021/jp502672k | J. Phys. Chem. C 2014, 118, 12642−12648
The Journal of Physical Chemistry C
Article
Figure 2. Connection type of FeSO4F octahedral: (a) corner-sharing and (b) edge-sharing.
Table 1. Calculated and Experimental Unit-Cell Parameters of Tavorite LiFeSO4Fa
Table 2. Formation Energies of Frenkel and Antisite Defects in Tavorite LiFeSO4F
lattice parameter
calcd
exp5
exp25
defect
energy (eV)
a (Å) b (Å) c (Å) α (deg) β (deg) γ (deg)
5.2524 5.5304 7.3127 106.585 107.161 96.082
5.1751 (0.0773) 5.4915 (0.0389) 7.2211 (0.0916) 106.506 (0.079) 107.178 (−0.017) 97.865 (−1.783)
5.1800 (0.0724) 5.4917 (0.0387) 7.2289 (0.0838) 106.486 (0.099) 107.186 (−0.025) 97.910 (−1.828)
Li Frenkel Fe Frenkel F Frenkel Li/Fe antisite
2.16 6.57 3.39 2.26
pair defects, which involve the exchange of a Li+ ion with a Fe2+ ion. The Li/Fe antisite defect is given by
a
The values in parentheses are the differences between the calculation in this study and the each experiment.
X X Li/Fe antisite: Li Li + Fe Fe → Fe•Li + Li′Fe
(4) 17
Compared with the olivine-structured LiFePO4 (1.13 eV), the Li/Fe antisite defect energy of this materials (2.26 eV) is considerably high. This result, as noted in the previous study,15 indicates that conduction blocking by Fe2+ ions on Li+ ion sites is hardly probable. Employing the Mott-Littleton method,30 systematic calculation of the activation energy for diffusion, that is, migration energy, of the Li+ ion has been performed by a simple vacancy hopping mechanism. In this calculation, the position of the highest potential energy along the migration path corresponding to the migration energy, that is, transition state position (stationary point with a single negative eigenvalue for the Hessian), is searched by rational functional optimization (RFO) method32 implemented in the GULP code. We calculated nine paths, in which the distances between adjacent two Li+ ions are less than 6 Å. Only the available paths, which have relatively low migration energies, are exhibited in Figure 3. As shown in Table 3, the low migration energies (90% of corner- and
