Article pubs.acs.org/JPCC
Cite This: J. Phys. Chem. C XXXX, XXX, XXX−XXX
Structures, Electronic States, and Reactions at Interfaces between LiNi0.5Mn1.5O4 Cathode and Ethylene Carbonate Electrolyte: A FirstPrinciples Study Yukihiro Okuno,*,†,‡ Keisuke Ushirogata,†,‡ Keitaro Sodeyama,§,∥,⊥ Ganes Shukri,‡,∥ and Yoshitaka Tateyama*,‡,§,∥
Downloaded via UNIV DE BARCELONA on January 9, 2019 at 03:07:20 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
†
Research and Development Management Headquarters, FUJIFILM Corporation, 210 Nakanuma, Minamiashigara, Kanagawa 250-0193, Japan ‡ Center for Green Research on Energy and Environmental Materials (GREEN) and International Center for Materials Nanoarchitectonics (MANA), National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan § Center for Materials Research Information Integration (cMI2), Research and Services Division of Materials Data and Integrated System (MaDIS), NIMS, 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan ∥ Elements Strategy Initiative for Catalysts & Batteries, Kyoto University, 1-30 Goryo-Ohara, Nishikyo-ku, Kyoto 615-8245, Japan ⊥ PRESTO, Japan Science and Technology Agency (JST), 4-1-8 Honcho, Kawaguchi, Saitama 333-0012, Japan S Supporting Information *
ABSTRACT: Electrolyte decomposition on cathode surfaces of lithium-ion batteries has attracted considerable attention because it leads to battery degradation and formation of a cathode solid−electrolyte interphase. In this study, we used density functional theory (DFT) calculations to investigate the distribution of the adsorption modes of ethylene carbonate (EC) electrolyte molecules and EC decomposition reactions on the (100) surfaces of lithiated (pristine) and delithiated forms of spinel-type LiNi0.5Mn1.5O4 (LNMO) as model cathode surfaces. DFT molecular dynamics (MD) simulations indicated that EC molecules have two characteristic adsorption modes. These two modes can satisfactorily explain the experimental observations and suggest a new feature of electrolyte−cathode interfaces in terms of the control of interfacial dipoles. On the basis of the DFT-MD results, we examined several possible pathways of EC decomposition on the LNMO (100) surfaces and estimated their activation barriers. We then found that the pristine LNMO (100) surface was inert with respect to the EC decomposition, whereas on delithiated surfaces, twofold-coordinated surface oxygen atoms generated by the delithiation process served as active sites for nucleophilic attack on the carbonyl carbon and the methylene group of adsorbed EC molecules. The induction of ring opening of the EC molecule by the former attack, and hydrogen abstraction from the methylene group and subsequent CO2 generation by the latter were consistent with experimental observations.
1. INTRODUCTION
processes are poorly understood compared to the well-studied reductive reactions at the anode−electrolyte interfaces.10−15 Generally, elucidation of chemical reactions at liquid−solid interfaces requires a molecular-scale understanding of the interface structures and relevant processes. Such an understanding, however, is lacking in the LIB field because of the complexity of the interfaces. Liu et al. have attempted to remedy this situation by using sum frequency generation (SFG) vibrational spectroscopy.16,17 They evaluated the preferred adsorption modes of solvent moleculesincluding ethylene carbonate (EC), diethyl carbonate (DEC), and
Lithium-ion batteries (LIBs) have achieved great commercial success in the portable power source market. Because of their characteristically high energy density, they have attracted considerable attention for use in larger power sources in, for example, electric vehicles. However, if LIBs are to be used for such purposes, their durability and safety must be improved.1−3 Degradation of the cathode−electrolyte interface is one mechanism by which LIBs deteriorate.4,5 For instance, it is widely accepted that dissolution of transition metals (TMs) from the cathode material into the electrolyte induces cathode interface degradation.6−9 The durability of LIBs is governed by oxidative decomposition of liquid electrolyte molecules at the cathode−electrolyte interfaces. However, these microscopic © XXXX American Chemical Society
Received: October 31, 2018 Revised: December 17, 2018
A
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
between a liquid electrolyte and a solid cathode and we analyzed electrolyte decomposition processes in light of the adsorption modes. Specifically, we used EC molecules16,17 adsorbed onto LNMO cathode surfaces as a representative model system. EC is the most commonly used electrolyte, and LNMO cathodes have attracted considerable attention for use in next-generation high-voltage (5 V) LIBs;40−44 the electrolyte durability at 5 V is crucial. We began by estimating the adsorption energies of single EC molecules in various modes of adsorption to elucidate the predominant modes under vacuum surface conditions. To understand the equilibrium structures of the interfacial electrolyte molecules, we then investigated the electrolyte− cathode interface by means of DFT-based molecular dynamics (MD) simulations with explicit treatment of EC molecules in the liquid electrolyte. Elucidation of the predominant modes of adsorption of EC on the cathode surface provided insight into the initial steps of the chemical reactions of EC on the surface. In addition, we compared our calculation results to those of the SFG experiments.16,17 To investigate the chemical reactions of EC, we focused on the nucleophilic attack of the cathode surface oxygens on electrolyte molecules. This attack may lower the electrolyte stability against oxidation.45,46 To elucidate the reaction pathways of the electrolyte, we treated the surface oxygen atoms of LNMO as chemically active sites and investigated their role in hydrogen abstraction in detail. Note that the pathways associated with O2 loss will be examined in a future study. We used the CO2 gas evolution experimentally observed to evaluate the electrolyte decomposition pathways. CO2 is one of the main products of electrolyte oxidation at the cathode surface, as indicated by differential electrochemical mass spectrometry.21−26 Evolution of CO2 has been detected not only for LNMO26 but also for various other cathode materials, such as LiCoO2,23,24 LiMn2O4, and LiNiCoAl. Although there are discrepant results regarding the evolution process of CO222,24 and which pairs of electrolyte and cathode materials cause the evolution,23,24 the chemical reactions that generates CO2 are surely the main reactions of the electrolyte on the cathode surface. From an industrial point of view, CO2 generation is dangerous because of the internal pressure buildup in the LIBs, and thus, the elucidation of the CO2 generation mechanism is quite important.
dimethyl carbonate (DMC)on a typical LiCoO2 cathode surface. The results suggested that EC molecules adsorb on the surface in two modes. The predominant mode is one in which the carbonyl group points toward the LiCoO2 surface; in the other mode, the carbonyl group points away from the surface. However, this interpretation of their results is still a matter of debate. In addition to the interfacial adsorption, interphase film formation via electrolyte decomposition on the cathode surfaces is also essential to the degradation. A cathode− electrolyte interphase (CEI) with a composition similar to that of the solid−electrolyte interphase that forms at the anode has been observed to evolve on the surface of cathode materials such as LixMn2O4 and LixCoO2.18−26 Besides electrolyte decomposition products such as LiF,18 other compounds detected in the film include acetone,19 an aldehyde,20 CO2,21−26 and various unidentified organic27 species. These results suggest that a variety of chemical reactions occur at cathode−electrolyte interfaces. However, their atomic-scale origins have not been fully elucidated. Oxidative reactions of the electrolyte molecules themselves are also problematic. The oxidation potentials of the typical electrolytes used in LIBs are usually above the range of the operating voltage. For instance, the oxidation potential of EC has been calculated to be approximately 5.8 V versus Li/Li+, whereas the maximum operating voltage of a LixMn2O4 cathode is reported to be 4.3 V.27−29 Theoretical considerations of electrolyte anions such as PF6− suggest that these anions decrease the electrolyte oxidation potential (e.g., the value for the EC/PF6− pair is 4.9 V vs Li/Li+30,31), although the potential remains within the electrochemically stable region for the electrolyte. This result is in sharp contrast to the potential on the anode side, where the operating voltage is below the thermodynamic reduction stability limit of the organic electrolytes. Therefore, the cathode−electrolyte interface is likely to play a crucial role in the electrolyte reaction and decomposition by acting as a catalyst. For instance, attack of oxygen radicals consisting of the interfacial lattice oxygen has been recently proposed as a possible cause of reactions.32,33 Thus, understanding the electrolyte reactions on the cathode interfaces on the atomic scale is of great importance. Density functional theory (DFT) is a powerful tool, commonly used to understand the atomistic properties of materials. However, there have been only a few DFT-based studies of the adsorption modes and reactions of EC molecules on the surface of cathode materials. Examples include studies of LiMn2O4,34,35 LiNi0.5Mn1.5O4 (LNMO),36 Li(Ni,Mn,Co)O2,37 and LiCoO2.38,39 Although these studies have provided some information about electrolyte reactions on cathode surfaces, the information is incomplete and, in some cases, controversial. For instance, various studies have given conflicting results with regard to abstraction of a hydrogen atom from intact physisorbed EC.34−36 Specifically, Kumar et al. have estimated hydrogen abstraction to be endothermic,35 whereas Borodin et al. have suggested that it occurs spontaneously on the LNMO surface.36 Resolving these discrepancies will require computational analyses that are more comprehensive. Furthermore, the results of these studies do not directly correspond to the electrolyte decomposition products (gases) of the electrolyte, detected in the experiments. In the present study, we computationally investigated the adsorption modes of electrolyte molecules at the interface
2. CALCULATION In constructing the LNMO surfaces, we used the ordered P4332 state with nickel ordered on the 4b sites of LNMO because this bulk crystal state has lower energy than the disordered (Fd3m) one. We selected the (100) surface because it has been the focus of many theoretical studies on spinel cathode materials.47−53 Although a recent computational study has shown that the (111) surface is the most stable, the calculated surface energy of the (100) surface is close to that of the (111) surface,48,53 which led to the appearance of the (100) surface in the Wulff construction.54,55 In fact, the (100) surface is experimentally observed even after LNMO is annealed at 800 °C for 100 h.56 We therefore consider that the present model can capture the processes happening at the LNMO/EC surfaces reasonably well. There have been many computational studies on the spinel (100) surfaces, and comparison of our results with those studies can provide valuable insights. We constructed the LNMO (100) surface B
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
To calculate the barriers to EC decomposition on the LNMO surface, we used the nudged elastic band (NEB)61 method. The convergence tolerance of the force in the geometrical optimization and in the NEB barrier calculation was set to 10−3 a.u. We performed Γ point calculations for the k-points. The dimensions of the simulation cell used at this stage were 11.55 × 11.55 × 28.17 Å3 for the LNMO surfaces, for which we used LixNi8Mn24O64 slab models with x values of 16, 8, and 0, which correspond to different states of charge. All electronic structure calculations were performed with Quantum Espresso code58 via a spin-dependent DFT plus Hubbard U (DFT + U) method62 for the d orbitals of TMs. Ultrasoft pseudopotentials63 were used for treatment of the core electrons of the atoms. For the +U augmented treatment of the nickel and manganese 3d orbitals, we set the Hubbard U values to 5.96 and 4.5 eV for Ni and Mn, respectively.51,52,64 Many previous studies have provided sufficient justification for the U values used in this work. Those studies gave reasonable results for the electronic state of cathode materials and reproduced the battery-related quantities such as lithium insertion potentials. We should also mention that our simulations did not directly involve the voltage; instead, we simulated the effects of applied voltage by varying the lithium content in the LNMO slab model. To understand the voltage effect on the adsorption behavior of EC molecules, more sophisticated approaches, such as those proposed by Leung and Leenheer65 and Otani and Sugino,66 will be necessary.
with a lithium-terminated surface layer. Half of the lithium ions on the surface layer were moved to another side of the surface slab to avoid an artificial net dipole in the supercell. The resulting layer stacking was 1/2(Li2); M4O8; Li2; ...; M4O8; 1/ 2(Li2) (where M = nickel or manganese).48 Although LNMO is known to have an antiferromagnetic state, the energy difference between the ferromagnetic and antiferromagnetic spin-ordering states was as small as 0.02 eV in our calculation. Therefore, we assumed that the influence of differences in the magnetic structure on the surface reaction mechanism was negligible, and we used only the ferromagnetic spin-ordering state for simplicity. The electronic state analyses are depicted in Figures S1−S2 in the Supporting Information. To account for the dynamical and statistical properties of the EC molecules on the LNMO surfaces, we carried out DFTMD simulations in the framework of Car−Parrinello molecular dynamics,57 implemented in the Quantum Espresso code.58 A fictitious electronic mass of 400 a.u. and a time step of 5 a.u. (ca. 0.12 fs) were chosen. The system temperature was controlled with a Nosé thermostat and the target temperature is set at 353 K, which can keep the liquid state of the EC electrolyte as well as allow the enhancement of sampling. After equilibration for at least 5 ps, the statistical averages were determined from trajectories of 5 ps in length. To control the adiabaticity of the electron dynamics, we used an additional thermostat for the electron component59 by setting the average kinetic energy of an electron to 0.65 a.u. We used an 11.55 × 11.55 × 37.89 Å3 supercell consisting of a Li16Ni8Mn24O64 slab and 30 EC molecules; the spacing of the EC molecules in the supercell reproduced an EC density of 1.32 g/cm3.60 Figure 1 depicts the structure of the EC molecule with the labels used in this paper, along with a schematic of the
3. RESULTS AND DISCUSSION 3.1. Adsorption of a Single EC Molecule on the LNMO Surface. We first calculated the adsorption energies of a single EC molecule on LNMO (100) surfaces in a vacuum (Table 1). Table 1. Adsorption Energies (eV) of a Single EC Molecule in Various Modes on LNMO (100) Surfaces under Vacuum Conditions (the Modes are Depicted in Figure 2)a Li16Ni8Mn24O64 Li10Ni8Mn24O64 Ni8Mn24O64
Li−OC
Li−OE
Mn−OC
Ni−OC
0.61 (2.0) 0.60(2.0)
0.36(2.1) 0.40 (2.1)
0.39(2.3) 0.36(2.4) 0.54 (2.1)
0.38(2.5) 0.38(2.8) 0.52 (2.4)
a
Values in parentheses are binding distances (Å) between the atoms on the LNMO surface and in the EC molecule.
We considered adsorption via binding interactions between four pairs of LNMO and EC atoms: Li−OC, Li−OE, Mn−OC, and Ni−OC (Figure 2). Note that our surface model did not treat the cathode voltage directly; instead, we accounted for the applied voltage by varying the lithium content of the LNMO surface slab (i.e., by introducing lithium vacancies). A positive adsorption energy indicated a stable adsorption mode. The Li−OC binding mode was the most stable adsorption
Figure 1. (a) Structure of EC, with the atom labels used throughout this paper. (b) Top view of the (100) surface of a Li16Ni8Mn24O64 slab model. (c) Side view of a supercell (11.55 × 11.55 × 37.89 Å3) consisting of a (100) LNMO slab and 30 EC molecules.
supercell used for the DFT-MD simulations. To determine the effect of the size of the surface slab, we estimated the adsorption energies of EC molecules with typical adsorption modes in a large supercell containing a Li32Ni16Mn48O128 slab, and we compared the estimated energies with those for a supercell containing a Li16Ni8Mn24O64 slab (see Table S1 in the Supporting Information). This comparison confirmed that the difference in the adsorption energies of the two slabs was small and thus that the Li16Ni8Mn24O64 slab was large enough to estimate the adsorption modes of the EC molecules.
Figure 2. Binding modes for a single EC molecule on the LNMO (100) surface. Silver, blue, purple, red, gray, and white spheres denote Li, Ni, Mn, O, C, and H atoms, respectively. C
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C mode. Furthermore, for x = 16 and 8, the adsorption energy of the Li−OE binding mode was comparable to the energies of the transition-metal binding modes (Ni−OC and Mn−OC). In the Li−OC and Li−OE binding modes, the distance between the lithium ion and OC and that between the lithium ion and OE were about 2.0 and 2.1 Å, respectively, regardless of the lithium content. The difference between these two distances reflected differences in adsorption strength between the Li−OC and Li−OE binding modes. The Coulombic attraction between the lithium ion and oxygen stabilized the EC molecules adsorbed via Li−O binding. The adsorption energies for the transition-metal-bound modes tended to increase with decreasing lithium content because of the oxidation of the TMs. Bader charge analysis of Li16Ni8Mn24O64 and fully delithiated Ni8Mn24O64 surfaces indicated that the average charges on nickel were 1.29 and 1.31 and that those on manganese were 1.75 and 1.77, respectively. The average Bader charges of the TM atoms in bulk Li 1 6 Ni 8 Mn 24 O 6 4 (Ni8Mn24O64) are 1.30 (1.34) for Ni and 1.85 (1.85) for Mn, respectively. Thus, the surface TMs are slightly reduced compared to the bulk ones, in particular for Mn atoms. When we started the geometry optimization from the Mn− OE adsorption mode, the adsorption mode changed to the Mn−OC or Li−OC mode during the optimization. This result suggests that the binding energies for adsorption modes involving one of the TMs and one of the OE atoms were smaller than those listed in Table 1. We also examined an initial state in which a hydrogen atom of the methylene group in EC faced toward a surface oxygen atom of LNMO, but in that case, movement of the EC molecule away from the surface during the geometry optimization indicated that adsorption modes involving the methylene hydrogen and the surface oxygen were unlikely under vacuum conditions. Here, we discuss the van der Waals (vdW) interaction effect on the adsorption energies. We estimated the adsorption energies of some binding modes of a single EC molecule using the vdW functional,67 and we found that the adsorption energies of Li−OC and Li−OE binding modes to fully lithiated Li16Ni8Mn24O64 were 1.04 and 0.73 eV, respectively. Although the vdW interaction increased the adsorption energies, the difference of the adsorption energies was almost the same as the result calculated without the vdW interaction. Therefore, the present calculations without vdW interaction will be sufficient to assess the relative stability among the adsorption modes. Finally, we discuss features of the surface Li on the LNMO surface. Li in the bulk LNMO has tetrahedral coordination of oxygen atoms with average Li−O distance of 1.94 Å, while Li on the surface in our (100) LNMO slab model is likely to have a distorted rectangular coordination with 2.15 Å of the average distance due to the surface reconstruction. These results implied weaker Li−O bonding around the surface region. On the other hand, the EC adsorption changes the number of the coordination of adsorbed Li to 3, and this Li has the distances of 1.89 and 1.98 Å to O in the LNMO subsurface and OC of the adsorbed EC molecule, respectively. Such a change of lithium−oxygen distances upon adsorption suggested that the surface Li can be stabilized by the EC adsorption. Thus, the surface Li adsorbed by the EC molecule may survive up to the higher voltages during the delithiation process than the notadsorbed Li on the LNMO surface. 3.2. DFT-MD Calculation of the EC Liquid Electrolyte/ LNMO Cathode Interface. We next investigated the
equilibrium interfacial structures of the EC liquid electrolyte−LNMO cathode system by DFT-MD sampling with explicit consideration of many-body effects and the dynamics of EC. Our supercell consisted of 30 EC molecules, 8 of which were in direct contact with the LNMO surfaces. The LNMO slab model had fully lithiated stoichiometry (Li16Ni8Mn24O64). We used DFT-MD simulations to estimate the θ and z distributions, where θ is the angle between the outgoing vector normal to the surface and the dipole moment (from oxygen to carbon) of the EC carbonyl bond, and z is the vertical distance of atoms in the EC molecules from the LNMO surface (Figure 3a).
Figure 3. (a) Definition of θ and z, which describe the orientation and position of the EC molecule relative to the LNMO surface. (b) Snapshot of interfacial EC molecules on the LNMO (100) surface in the equilibrium trajectory of the DFT-MD simulations. (c) Angle distribution profile for EC molecules adsorbed on the LNMO (100) surface.
In Figure 3b, we show a snapshot of the EC molecules near the LNMO surfaces during a DFT-MD run. On the surface are four EC molecules, two of which are adsorbed in the Li−OC binding mode and two of which are oriented with the carbonyl group pointing away from the surface. The θ distribution has two peaks: one at around 60° and another at around 120° (Figure 3c). The peak around 60° originated from EC molecules oriented with their carbonyl groups pointing downward toward the LNMO surface (hereafter designated as the Li−OC downward adsorption mode). The calculations for a single adsorbed EC molecule under vacuum conditions indicate that the Li−OC downward mode predominated and the exposed lithium ions on the LNMO surface preferentially interacted with the carbonyl oxygen of EC. Such a Li−OC downward mode has also been reported in previous theoretical studies.35,37 The peak at around 120° corresponded to EC molecules with the methylene group oriented downward toward the LNMO surface (hereafter designated as the upward adsorption mode); these EC molecules were not bound to a lithium atom on the LNMO surface. Interestingly, the upward adsorption mode was unstable under vacuum conditions. The presence of the upward mode in the solution environment was the result of many-body effects of EC molecules on the LNMO surface. The upward mode was probably stabilized by dipole interactions with molecules in the Li−O C downward adsorption mode. We confirmed that the presence of the EC molecules in the upward mode did not depend on the initial D
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
indicates that the peak of the hydrogen atom distribution at around 1 Å originated from the molecules in the upward mode. The sharp peak at around 2 Å in the OC distribution of Figure 4c confirms that Li−OC binding tightly pinned molecules in the Li−OC downward mode to the LNMO surface. In contrast, the broad peak of the z distribution of OC in the upward mode at around 4.5 Å indicates that EC molecules fluctuated more in the upward mode than that in the downward mode. The OE distribution (Figure 4d) showed that EC molecules on the surface fluctuated to a certain extent in both modes. For molecules in the downward mode, the fluctuation of the z position of OE was caused by rotation around the axis of the carbonyl bond pinned to the surface via Li−OC binding. These results of the z distribution for OE (Figure 4d) suggest that EC molecules in both modes may be bound to the surface via OE. Figure 5 shows the radial distribution functions (RDFs) of the distance between the atoms of LNMO and the hydrogen
state used for DFT-MD sampling. That is, the upward mode emerged regardless of the choice of initial states; if we used an initial state in which all surface EC molecules were oriented with the carbonyl group pointing downward toward the LNMO surface, the surface EC molecules that did not adopt the Li−OC downward mode quickly adopted the upward mode. Our observation of EC molecules in the upward mode near the cathode surface in the DFT-MD simulations is consistent with the results of SFG experiments.16,17 The observation that the peaks in the θ distribution were not sharp indicates that the orientations of the EC molecules in the two adsorption modes were not fixed; that is, the orientations could fluctuate to some extent on the LNMO surface. In particular, the orientations of molecules adsorbed in the upward mode fluctuated considerably because, unlike molecules in the Li−OC downward mode, they were not bound tightly to the LNMO surface. Figure 4a shows the distributions
Figure 4. Distributions of z (distance from the LNMO surface) for (a) all atoms of all EC molecules and (b) H atoms, (c) OC atoms, and (d) OE atoms of the surface EC molecules during 5 ps DFT-MD sampling. “Down EC” and “up EC” denote the Li−OC downward adsorption mode and the upward adsorption mode, respectively.
Figure 5. RDFs for H atoms of EC molecules in (a) the downward adsorption mode and (b) the upward adsorption mode relative to LNMO atoms. RDFs from CC atoms of EC molecules in the (c) downward and (d) upward adsorption modes relative to LNMO atoms.
of distances from the LNMO surface to all atoms in all EC molecules. The strong peak for OC, the carbonyl oxygen, at around 2 Å indicates the presence of the EC molecules in the Li−OC downward mode. The peak of the distance distribution for the hydrogen atoms of the methylene group at about 1 Å from the LNMO surface suggests that the surface oxygen could abstract hydrogen atoms from EC, as discussed below. Here, we discuss the size effect of our slab model. The density of the monolayer of EC molecules on the cathode surface is about 0.9 g/cm3, which is smaller than the bulk density of EC (1.32 g/cm3). Furthermore, the structure of the EC downward adsorption mode at the LNMO/EC interface was almost the same as that of the single EC molecule Li−OC adsorption mode in the vacuum model. These results indicate that EC molecules were not confined in a small space in our DFT-MD calculation and our slab model had sufficient size to capture the essence of interfacial structures of the EC liquid electrolyte. We also show the z distributions for the hydrogen, OC, and OE atoms of EC molecules in the downward and upward adsorption modes in Figure 4b−d, respectively. Figure 4b
and CC atoms of EC molecules in the Li−OC downward adsorption mode and the upward adsorption mode; the hydrogen and CC are potential sites for attack by the surface oxygen atoms of LNMO. The RDF for the distance between the hydrogen atoms of molecules in the downward mode and the LNMO surface oxygen atoms shows a small peak at around 2.3 Å (Figure 5a). The presence of this peak indicates that the hydrogen atoms of the EC methylene groups can react with the oxygen atoms, as discussed in the next section. In addition, the RDF of the distance between the hydrogen atoms of EC molecules in the upward adsorption mode and the LNMO oxygen atoms shows a broad peak rising at 2 Å (Figure 5b). This distribution indicates that the methylene hydrogen atoms of EC molecules in the upward mode weakly interacted with the LNMO oxygen atoms via hydrogen bonds. In the upward adsorption mode, there is a possibility that the EC methylene hydrogen can react with the LNMO surface oxygen. In the RDF of the distance from CC of EC molecules in the downward adsorption mode relative to LNMO oxygen, the peak at around 2.9 Å indicates the possibility of nucleophilic E
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C attack of the oxygen atoms on CC, a possibility we discuss in section 3.3. 3.3. Analysis of EC Decomposition on the LNMO Surface by NEB Barrier Calculation. In this section, we describe our calculations of the activation barriers of EC decomposition on the LNMO (100) surfaces in various adsorption modes. Here, we focus mainly on a partially delithiated (100) slab (Li10Ni8Mn24O64). The results for the other stoichiometries are presented in the Supporting Information. To examine whether spontaneous oxidation of EC molecule occurs, we first show the electronic density of states (DOS) of the LNMO (100) surface with adsorption of one EC molecule in Figure 6. For the adsorption mode, we adopted the Li−OC
Figure 7. Energy profile for nucleophilic attack of twofoldcoordinated surface oxygen on the carbonyl carbon of EC on the (100) surface of Li10Ni8Mn24O48.
The results shown in Figure 7 are consistent with those in a previous study34 of an EC molecule on the (100) surface of LixMn2O4; that previous study also involved nucleophilic attack of surface oxygen on the carbonyl carbon of EC, and the calculated activation barrier (∼0.24 eV) was similar to the present value. Another similar value (∼0.29 eV) was also suggested in a recent study of EC decomposition on Li(Ni,Mn,Co)O2 under a vacuum condition.37 These results have suggested that the product of nucleophilic attack of surface oxygen on the CC atom of EC is a ring-opened molecule formed by cleavage of one of the CC−OE bonds. Although the present calculations did not reveal a clear ring opening, there was an indication of EC ring cleavage on the partially delithiated (100) Li10Ni8Mn24O48 surface. When we calculated the energy profile on the fully delithiated Ni8Mn24O48 (Figure S3 in the Supporting Information), a ring-opened EC molecule was surely observed as a product. These results indicate that a ring-opened EC is a likely product of surface oxygen attack on CC of EC molecules under delithiation conditions. On the other hand, the adsorption structure of the carbonyl carbon CC of EC makes it difficult to produce CO2 gas from this surface reaction. Therefore, we should consider other reaction pathways for CO2 gas evolution from the EC decomposition products. We then investigated abstraction of a hydrogen atom from EC by nucleophilic attack of surface oxygen. Figure 8 shows the results of the NEB barrier calculations for the hydrogen abstraction. We examined surface oxygen atoms with threefold coordination (Figure 8c) on the pristine surface and with twofold coordination (Figure 8d) on the surface of the delitiated cathode. For the initial configuration of the EC molecule on the LNMO surface, we used the Li−O C downward adsorption mode, which is predominant according to our DFT-MD calculations. The dissociation of a hydrogen atom from an EC methylene group and formation of a covalent bond with a surface oxygen atom led to the formation of a hydroxyl group on the LNMO surface. The activation and reaction energies for hydrogen abstraction from an adsorbed EC molecule by a threefoldcoordinated surface oxygen atom were 0.63 and 0.46 eV, respectively (Figure 8a). That is, hydrogen abstraction by three-fold-coordinated oxygen was endothermic and therefore unlikely to occur. In contrast, the pathway via nucleophilic attack by twofold-coordinated oxygen has the activation and reaction energies of 0.41 and −0.46 eV, respectively, indicating that this reaction was thermodynamically exothermic and therefore possible (Figure 8b). Because the Fermi level of the system was −6.09 eV and the singly occupied molecular orbital of the neutral hydrogen-abstracted EC molecule was located at
Figure 6. DOSs of the Li−OC downward EC adsorption mode to the (001) surface of (a) fully lithiated Li16Ni8Mn24O64 and (b) partially delithiated Li10Ni8Mn24O64. The vertical broken line denotes the Fermi level. The blue and red lines show the projected DOSs of the adsorbed EC molecule and the LNMO surface slab, respectively.
downward EC mode as it is the most stable adsorption structure of an EC molecule, as shown in Table 1. Figure 6 indicates that both highest occupied and the lowest unoccupied states consist of the LNMO electronic states and the Fermi level is located between the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital of the EC molecule. In fact, the HOMO levels in fully lithiated Li 1 6 Ni 8 Mn 2 4 O 6 4 and partially delithiated Li10Ni8Mn24O64 lie at about 3.2 and 2.4 eV, respectively, below the Fermi level. Thus, spontaneous oxidation of the adsorbed EC molecule is unlikely to occur in the surfaces examined here. This suggests that the reaction of the EC molecule on the LNMO cathode surface is caused by the catalytic process of the LNMO surface and the surface oxygen can be a promising candidate for the reaction site. First, we investigated nucleophilic attack of surface oxygen atoms on the carbonyl carbon (CC in Figure 1) of EC. For the initial adsorption mode, we used EC in the Li−OC downward adsorption mode on the (100) surface of partially delithiated LNMO (Li10Ni8Mn24O48). Figure 7 shows the energy profile for the nucleophilic attack of twofold-coordinated surface oxygen on CC. In the transient state of this reaction, a bond between CC and the LNMO surface oxygen (Os) was formed. In the final structure, the CC−Os bond length was 1.44 Å and one of the CC−OE bonds of EC was lengthened to 1.45 Å (compared with 1.37 Å in the isolated neutral EC molecule). This bond lengthening can be regarded as an indication of ring opening. The reaction and activation energies were about −0.07 and 0.21 eV, respectively. Because the Li−O C downward adsorption mode is the dominant adsorption mode and has a low activation barrier, the present reaction is a major reaction of EC molecules in this adsorption mode. F
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
Figure 9. Energy profile for two-hydrogen abstraction from an EC molecule by nucleophilic attack of two different twofold-coordinated surface oxygen atoms on the Li8Ni8Mn24O64 surface.
of the EC molecule. If this condition is satisfied, a twohydrogen abstraction is feasible. We next investigated the processes with CO2 generation subsequent to hydrogen abstraction. Generation of CO2 from the hydrogen-abstracted EC molecule in the Li−OC downward adsorption mode is difficult because of the strong pinning of the EC carbonyl oxygen to a surface lithium atom on the LNMO surface. However, as shown in DFT-MD study of the LNMO/EC surface, the downward adsorption mode is not the only possible mode. The orientation of EC molecules in the upward adsorption mode fluctuates on the LNMO surface, and the molecules are likely to have a chance to be bound to the surface via the OE oxygen. To evaluate such processes after hydrogen abstraction, we assumed that an EC molecule first adsorbed via the Mn−OE interaction. We then considered the following two steps; a hydrogen atom is abstracted by a twofold-coordinated surface oxygen (step 1) and the hydrogen-abstracted EC molecule decomposes to generate CO2 (step 2). Figure 10 displays the NEB calculation results for the two-step process. The calculated activation and reaction energies of hydrogen abstraction were 0.75 and −0.31 eV, respectively (Figure 10a), whereas those in step 2 were about +0.38 and −1.0 eV (Figure 10b), respectively. Overall, the conversion of EC into CO2 and a remaining fragment was highly exothermic (by more than 1.3
Figure 8. Energy profiles for hydrogen abstraction from EC on the (100) surface of partially delithiated Li10Ni8Mn24O64 by nucleophilic attack of (a) threefold-coordinated and (b) twofold-coordinated surface oxygen atoms. Panels (c,d) show the threefold- and twofoldcoordinated oxygen atoms, respectively.
−4.82 eV, an electron was subsequently transferred from the hydrogen-abstracted EC molecule to the LNMO surface. The indication was that the EC molecule on the LNMO surface was oxidized. We estimated the total charge of the hydrogenabstracted EC molecules to be 0.58 e (e > 0) by means of Bader charge analysis. This charge surely indicated that the EC molecule existed as a cation. Because of the oxidized state of the hydrogen-abstracted EC molecule, the distance between lithium and OC of the EC molecule increased to 2.3 Å from the original Li−OC length of 2.0 Å. These results suggest that the hydrogen was abstracted to the LNMO surface, from which it could easily desorb. Leung34 and Kumer et al.35 have also used NEB calculations to study hydrogen abstraction from an EC molecule on a cathode surface and its subsequent reaction. Their results have shown that the hydrogen abstraction is endothermic and involves threefold-coordinated oxygen, as also indicated in this study. Borodin et al.36 have investigated the abstraction of hydrogen from EC by using structure relaxation calculations and have found that one or two hydrogen atoms are abstracted from EC, depending on the initial configuration of the adsorbed EC molecule. These investigations used a fully delithiated LNMO surface (Ni0.5Mn1.5O4), and their surface oxygen atoms were twofold-coordinated. We investigated abstraction of two hydrogen atoms from EC by using two different twofold-coordinated oxygen atoms on a Li8Ni8Mn24O64 surface. Figure 9 shows the energy profiles calculated with the NEB method. The abstractions of the two hydrogen were sequential rather than simultaneous, and the structure with one hydrogen atom removed was transient. The overall activation and reaction energies for the two-hydrogen abstraction were 0.27 and −2.50 eV, respectively. The low activation barrier is consistent with that reported in a previous study36 of spontaneous abstraction of two hydrogen atoms from EC by surface oxygen of the Ni0.5Mn1.5O4 slab. Our calculations indicate that the reaction is highly exothermic. For this reaction to occur, however, twofold-coordinated surface oxygens must be close to the two methylene hydrogen atoms
Figure 10. Energy profiles for (a) hydrogen abstraction from an EC molecule in the Mn−OE adsorption mode (step 1) and (b) subsequent EC decomposition accompanied by CO2 generation (step 2). G
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C eV relative to the EC molecule in the Mn−OE adsorption mode). This thermodynamic preference of the decomposition reaction was attributed to the binding of the C2OH3 moiety of the EC molecule to the LNMO surface. In the Supporting Information, we provide an energy diagram for CO 2 production from cationic hydrogen-abstracted EC molecules in the solution phase; the diagram indicates that the reaction is endothermic (Figure S4 in the Supporting Information). At the transition state of the CO2 generation process (see Figure 10b), the bond length of OE (attached to Mn atom) and CC in EC molecules is 2.12 Å and that of OE and CE is 2.09 Å. Hence, these two bonds seem to concertedly dissociate to release the CO2 moiety. In the case of the Li10Ni8Mn24O48 surface, the activation barrier of the entire reaction was about 0.75 eV, whereas that for CO2 generation after hydrogen abstraction was only 0.35 eV. This activation barrier permits a reaction rate of 10 μs−1 if we assume a typical molecular vibrational prefactor of 1012 s−1. Therefore, it is highly probable that CO2 generation occurs at the typical battery charging/discharging rate. We analyzed Bader charges of the hydrogen-abstracted EC molecule at the CO2 generation step to clarify the electron transfer between the EC molecule and the LNMO surface. The initial state of the Mn−OE adsorbed hydrogen-abstracted EC has 0.45 e, positive charge as we discussed above, while at the transition state of the CO2 gas generation process, the charge of hydrogen-abstracted EC becomes 0.18 e, indicating the electron transfer from the LNMO surface to the hydrogen-abstracted EC. The charge of the product state in the C2OH3 moiety on the LNMO surface is −0.06 e, almost neutral in the end. Here, we point out that this CO2-generation process was not limited to Mn−OE binding. We also observed hydrogen abstraction and CO2 generation for molecules with Li−OE binding (Figure S5 in the Supporting Information). It is thus crucial that the carbonyl carbon (OC) of EC is not bound to a TM (manganese or nickel) or lithium atom on the LNMO surface. Generation of CO2 gas is not unique to LNMO surfaces. The process depends on the positioning of the twofoldcoordinated oxygen atoms and the adsorption mode of the hydrogen-abstracted EC molecule. We also examined EC decomposition on a LiCoO2 (LCO) (110) surface; the energy profiles for hydrogen abstraction and subsequent generation of CO2 are shown in Figures S6 and S7, respectively, in the Supporting Information. The resultant profiles show that CO2 generation from EC molecules is possible if a twofoldcoordinated surface oxygen is present and if the adsorption mode of the hydrogen-abstracted EC molecules is properly located relative to twofold-coordinated surface oxygen so that hydrogen abstraction occurs. 3.4. Features of the Electrolyte−Cathode Interface. These results provide important insights about the characteristic configurations at the liquid electrolyte−cathode interfaces in LIBs and the EC decomposition triggered by chemically active surface oxygen sites on the cathode surfaces. In particular, the coexistence of a pinned Li−OC downward adsorption mode and a flexible upward adsorption mode is crucial to EC decomposition. Such multiple adsorption modes play a role in controlling the interfacial dipole. Figure 11 shows a schematic of a surface EC monolayer on an LNMO surface obtained in this study. The EC molecules in the downward adsorption mode are strongly pinned to the LNMO surface unless they react and decompose, whereas those in the upward adsorption mode can
Figure 11. Schematic illustration of the surface EC molecules in downward and upward adsorption modes on the cathode surface: the labels M, M′, and M″ denote metal species on the cathode surface. (a) EC molecules in the upward adsorption mode may fluctuate on the cathode surface and (b) the fluctuated upward EC molecule can change to the M′ (TM or Li)−OE adsorption mode.
suppress unfavorable interfacial net dipoles by dipole−dipole interactions with the EC molecules in the downward mode (Figure 11a). In addition, the upward mode molecules can rotate relatively freely on the surface and quickly respond to the changes of the lithium content of the cathode surface. When a voltage is applied and the surface becomes positively charged during delithiation, some of the EC molecules in the upward mode can rotate into the downward mode for screening of the induced interfacial dipole by polarization of the EC molecule. These rotations of the upward molecules are expected to induce orientation changes of EC molecules in the second and further layers, namely, response of the electric double layer at the electrolyte−electrode interface during the charging process. Note that the same mechanisms can be applied for the discharge process. Our results indicate that CO2 is not generated from EC molecules in the most stable adsorption mode (the Li−OC downward mode). However, EC molecules are not confined to that mode. The orientation of EC molecules in the upward adsorption mode can fluctuate and thereby produce absorption structures with Li−OE or TM−OE binding in response to changes of the dipole on the cathode surface (Figure 11b). In addition, when lithium ions, some of which are bound to the OC atoms of EC molecules in the downward adsorption mode, are removed from the cathode surface by delithiation, EC molecules adsorbed in the downward mode also move away from the cathode along with the lithium ions. As a result, the orientation of nearby EC molecules can fluctuate and the molecules thus have a chance to adopt an adsorption structure that does not involve Li−OC binding. The reorientation of EC in the upward state or readsorption of EC in the downward state does not seem to have large activation free energy, based on our observations of large fluctuation of the EC molecules in our DFT-MD samplings. Furthermore, oxidized hydrogen-abstracted EC molecules are more weakly adsorbed on the LNMO surface than neutral EC molecules, and the former can desorb from the surface and be readsorbed in a different mode. Hydrogen-abstracted EC molecules may bind to atoms of the cathode surface via the methylene carbons from which a hydrogen atom is abstracted. Figure S7 in the Supporting Information shows that hydrogenabstracted EC molecules can bind in this type of structure and that CO2 can be generated. H
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
Our calculations also suggested several possible adsorption structures during fluctuation, such as Li−OE or M−OE binding in the upward mode. We used NEB barrier calculations to demonstrate that the twofold-coordinated oxygen atoms on the delithiated LNMO surfaces served as nucleophilic active sites for EC decomposition by attacking the carbonyl carbon as well as the hydrogen atoms of the methylene groups of EC. The nucleophilic attack on CC of EC in the Li−OC downward mode leads to the EC ring-opening reaction. The attack on the methylene hydrogen atoms and subsequent hydrogen abstraction were also probable. During hydrogen abstraction, an electron was simultaneously transferred to the LNMO surface from an EC molecule. This reaction is one of a class of reactions called proton-coupled electron transfer.68 We also demonstrated that the EC decomposition accompanied by CO2 generation is possible after hydrogen abstraction. In contrast, the threefold-coordinated surface oxygen present on the pristine LNMO surface did not attack EC molecules in such ways.
Hydrogen abstraction and CO2 generation can be expected to occur to a considerable extent under voltage application because the number of twofold-coordinated oxygen atoms on the cathode surface increases as a result of delithiation; this prediction is consistent with the results of experimental studies.21−26 Generation of CO2 has been discussed in a recent theoretical study by Tebbe et al.,39 who investigated EC decomposition on an LCO (101̅4) surface. The pathways of CO2 generation from EC in their study actually differed from the pathway we have suggested, and their activation energy was higher (about 1.81 eV from an EC molecule with the LCO surface) than ours (0.75 eV, including the hydrogenabstraction process in the LNMO case; 1.27 or 0.85 eV, depending on the structure of the EC adsorbed on the LCO (110) surface; see Figure S7 and S8 in the Supporting Information). Their high activation energy is attributable to their initial EC adsorption mode before the reaction, which is in the Li−OC downward mode. Therefore, the pathways demonstrated in the present study are more plausible for CO2 generation from EC molecules on LIB cathode surfaces. As we have shown, the presence of the LNMO surface promotes the EC decomposition reactions that lead to ring opening or CO2 generation. The decomposed products can be strongly bound to the cathode surface. These products are sources of the organic components of the CEI film recently discussed. In this study, we considered only a solvent consisting of cyclic molecules like EC. However, acyclic molecules, such as DMC or DEC, and the counter ions of salts may cause the solvent deterioration. In fact, the salt anions such as PF6− have been shown to affect EC deterioration30,31,39 and to lower the activation barrier of the hydrogen-abstraction reaction.30,31 Such effects of the salts on the electrolyte chemical reaction will be the subjects of future studies.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.8b10625. DOSs of LNMO bulk and surface; cluster-boundary condition DFT analysis of the CO2 gas generation reaction of EC without the cathode surface; and NEB analysis of chemical reactions on LNMO (100) and LCO (110) surfaces (PDF)
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail: yukihiro.okuno@fujifilm.com (Y.O.). *E-mail:
[email protected] (Y.T.). ORCID
4. CONCLUSIONS In this study, we used DFT calculations to elucidate the adsorption modes and chemical reactions of EC electrolyte molecules adsorbed on LNMO (100) surfaces. By calculating the adsorption energies of a single EC molecule on LNMO surfaces under a vacuum condition, we showed that the most stable adsorption mode is the one in which the OC atom of EC is bound to a surface lithium atom (the Li−OC downward adsorption mode). We also examined liquid−solid interface structures at the LNMO/EC interface by using DFT-MD calculations, and we found that an EC upward adsorption mode in which the methylene groups of EC were oriented toward the LNMO surface is present in addition to EC molecules in the Li−OC downward adsorption mode. The multiple adsorption modes demonstrated by our calculations explain the results of SFG experiments.16,17 The geometric distribution functions calculated by the DFTMD samplings demonstrated the possible rotations and fluctuating orientations of the surface EC molecules in the Li−OC downward adsorption mode and upward adsorption mode. The motion flexibility as well as an alternate alignment of the two modes (Figure 11a) suggests that there is suppression and control of excess interfacial dipoles (electric double layer) by the monolayer EC molecules on both lithiated pristine and delithiated LNMO surfaces. These results provide new insights into the nature of the liquid electrolyte− solid cathode interfaces in LIBs.
Yoshitaka Tateyama: 0000-0002-5532-6134 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Y.O. and K.U. acknowledge the support of M. Suzuki, Dr. H. Watanabe, and Dr. K. Furuya at the Fuji Film Corporation. This work was supported by the Elements Strategy Initiative for Catalysts & Batteries (ESICB) of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), JSPS KAKENHI Grant-in-Aid for Scientific Research (C) (no. JP15K05138), and the “Priority Issue (No. 5) on Post K Computer” project of MEXT. The calculations were carried out on the supercomputers at NIMS, ISSP, and ITC of The University of Tokyo and Kyushu University, as well as on the K computer at the RIKEN Advanced Institute for Computational Science (AICS), partly through the HPCI System Research Projects (Project IDs: hp150090, hp170122, hp170169, hp170241, hp170292, and hp180091).
■
REFERENCES
(1) Job, T. R.; Xu, K.; Borodin, O.; Ue, M. Electrolytes for Lithium and Lithium-Ion Batteries (Modern Aspects of Electrochemistry); Springer: Berlin, Germany, 2014. (2) Xu, K. Electrolytes and Interphases in Li-ion Batteries and Beyond. Chem. Rev. 2014, 114, 11503−11618. I
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C (3) Goodenough, J. B. Evolution of Strategies for Modern Rechargeable Batteries. Acc. Chem. Res. 2012, 46, 1053−1061. (4) Aurbach, D.; Markovsky, B.; Rodkin, A.; Levi, E.; Cohen, Y. S.; Kim, H.-J.; Schmidt, M. On the capacity fading of LiCoO 2 intercalation electrodes. Electrochim. Acta 2002, 47, 4291−4306. (5) Edström, K.; Gustafsson, T.; Thomas, J. O. The cathode− electrolyte interface in the Li-ion battery. Electrochim. Acta 2004, 50, 397−403. (6) Hunter, J. C. Preparation of a new crystal form of manganese dioxide: λ-MnO2. J. Solid State Chem. 1981, 39, 142−147. (7) Gummow, R. J.; de Kock, A.; Thackeray, M. M. Improved capacity retention in rechargeable 4 V lithium/lithium-manganese oxide (spinel) cells. Solid State Ionics 1994, 69, 59−67. (8) Thackeray, M. M. Spinel electrodes for lithium batteries. J. Am. Ceram. Soc. 1999, 82, 3347−3354. (9) Blyr, A.; Sigala, C.; Amatucci, G. G.; Guyomard, D.; Chabre, Y.; Tarascon, J. M. Self-Discharge of LiMn2O4/C Li-Ion Cells in Their Discharged State. J. Electrochem. Soc. 1998, 145, 194−209. (10) Lithium-Ion Batteries: Solid-Electrolyte Interphase; Wang, Y.; Balbuena, P. B., Eds.; Imperial College: London, 2004. (11) Leung, K.; Budzien, J. L. Ab initio molecular dynamics simulations of the initial stages of solid-electrolyte interphase formation on lithium ion battery graphitic anodes. Phys. Chem. Chem. Phys. 2010, 12, 6583−6586. (12) Leung, K. Electronic Structure Modeling of Electrochemical Reactions at Electrode/Electrolyte Interfaces in Lithium Ion Batteries. J. Phys. Chem. C 2012, 117, 1539−1547. (13) Ushirogata, K.; Sodeyama, K.; Okuno, Y.; Tateyama, Y. Additive Effect on Reductive Decomposition and Binding of Carbonate-Based Solvent toward Solid Electrolyte Interphase Formation in Lithium-Ion Battery. J. Am. Chem. Soc. 2013, 135, 11967−11974. (14) Ushirogata, K.; Sodeyama, K.; Futera, Z.; Tateyama, Y.; Okuno, Y. Near-shore aggregation mechanism of electrolyte decomposition products to explain solid electrolyte interphase formation. J. Electrochem. Soc. 2015, 162, A2670−A2678. (15) Okuno, Y.; Ushirogata, K.; Sodeyama, K.; Tateyama, Y. Decomposition of the fluoroethylene carbonate additive and the glue effect of lithium fluoride products for the solid electrolyte interphase: an ab initio study. Phys. Chem. Chem. Phys. 2016, 18, 8643−8653. (16) Liu, H.; Tong, Y.; Kuwata, N.; Osawa, M.; Kawamura, J.; Ye, S. Adsorption of Propylene Carbonate (PC) on the LiCoO2 Surface Investigated by Nonlinear Vibrational Spectroscopy. J. Phys. Chem. C 2009, 113, 20531−20534. (17) Yu, L.; Liu, H.; Wang, Y.; Kuwata, N.; Osawa, M.; Kawamura, J.; Ye, S. Preferential Adsorption of Solvents on the Cathode Surface of Lithium Ion Batteries. Angew. Chem., Int. Ed. 2013, 52, 5753−5756. (18) Aurbach, D.; Gamolsky, K.; Markovsky, B.; Salitra, G.; Gofer, Y.; Heider, U.; Oesten, R.; Schmidt, M. The Study of Surface Phenomena Related to Electrochemical Lithium Intercalation into LixMOy Host Materials (M = Ni, Mn). J. Electrochem. Soc. 2000, 147, 1322−1331. (19) Ufheil, J.; Würsig, A.; Schneider, O. D.; Novák, P. Acetone as oxidative decomposition product in propylene carbonate containing battery electrolyte. Electrochem. Commun. 2005, 7, 1380−1384. (20) Moshkovich, M.; Cojocaru, M.; Gottlieb, H. E.; Aurbach, D. The study of the anodic stability of alkyl carbonate solutions by in situ FTIR spectroscopy, EQCM, NMR and MS. J. Electroanal. Chem. 2001, 497, 84−96. (21) Takami, N.; Ohsaki, T.; Hasebe, H.; Yamamoto, M. Laminated Thin Li-Ion Batteries Using a Liquid Electrolyte. J. Electrochem. Soc. 2002, 149, A9−A12. (22) Imhof, R.; Novak, P. Oxidative Electrolyte Solvent Degradation in Lithium-Ion Batteries: An In Situ Differential Electrochemical Mass Spectrometry Investigation. J. Electrochem. Soc. 1999, 146, 1702− 1706. (23) Wuersig, A.; Scheifele, W.; Novák, P. CO2 Gas Evolution on Cathode Materials for Lithium-Ion Batteries. J. Electrochem. Soc. 2007, 154, A449−A454.
(24) Wang, H.; Rus, E.; Sakuraba, T.; Kikuchi, J.; Kiya, Y.; Abruña, H. D. CO2 and O2 Evolution at High Voltage Cathode Materials of Li-Ion Batteries: A Differential Electrochemical Mass Spectrometry Study. Anal. Chem. 2014, 86, 6197−6201. (25) Vetter, J.; Holzapfel, M.; Wuersig, A.; Scheifele, W.; Ufheil, J.; Novák, P. In situ study on CO2 evolution at lithium-ion battery cathodes. J. Power Sources 2006, 159, 277−281. (26) Michalak, B.; Berkes, B. B.; Sommer, H.; Bergfeldt, T.; Brezesinski, T.; Janek, J. Gas Evolution in LiNi0.5Mn1.5O4/Graphite Cells Studied In Operando by a Combination of Differential Electrochemical Mass Spectrometry, Neutron Imaging, and Pressure Measurements. J. Anal. Chem. 2016, 88, 2877−2883. (27) Eriksson, T.; Andersson, A. M.; Bishop, A. G.; Gejke, C.; Gustafsson, T.; Thomas, J. O. Surface Analysis of LiMn2O4 Electrodes in Carbonate-Based Electrolytes. J. Electrochem. Soc. 2002, 149, A69− A78. (28) Matsuta, S.; Kato, Y.; Ota, T.; Kurokawa, H.; Yoshimura, S.; Fujitani, S. Electron-Spin-Resonance Study of the Reaction of Electrolytic Solutions on the Positive Electrode for Lithium-Ion Secondary Batteries. J. Electrochem. Soc. 2001, 148, A7−A10. (29) Jang, D. H.; Oh, S. M. Electrolyte Effects on Spinel Dissolution and Cathodic Capacity Losses in 4 V Li/LixMn2O4 Rechargeable Cells. J. Electrochem. Soc. 1997, 144, 3342−3348. (30) Xing, L.; Borodin, O.; Smith, G. D.; Li, W. Density Functional Theory Study of the Role of Anions on the Oxidative Decomposition Reaction of Propylene Carbonate. J. Phys. Chem. A 2011, 115, 13896−13905. (31) Borodin, O.; Jow, T. R. Quantum Chemistry Studies of the Oxidative Stability of Carbonate, Sulfone and Sulfonate-Based Electrolytes Doped with BF4‑, PF6‑ Anions. ECS Trans. 2011, 33, 77−84. (32) Jung, R.; Metzger, M.; Maglia, F.; Stinner, C.; Gasteiger, H. A. Oxygen Release and Its Effect on the Cycling Stability of LiNixMnyCozO2(NMC) Cathode Materials for Li-Ion Batteries. J. Electrochem. Soc. 2017, 164, A1361−A1377. (33) Jung, R.; Metzger, M.; Maglia, F.; Stinner, C.; Gasteiger, H. A. Chemical versus Electrochemical Electrolyte Oxidation on NMC111, NMC622, NMC811, LNMO, and Conductive Carbon. J. Phys. Chem. Lett. 2017, 8, 4820−4825. (34) Leung, K. First-Principles Modeling of the Initial Stages of Organic Solvent Decomposition on LixMn2O4(100) Surfaces. J. Phys. Chem. C 2012, 116, 9852−9861. (35) Kumar, N.; Leung, K.; Siegel, D. J. Crystal Surface and State of Charge Dependencies of Electrolyte Decomposition on LiMn2O4 Cathode. J. Electrochem. Soc. 2014, 161, E3059−E3065. (36) Borodin, O.; Olguin, M.; Spear, C. E.; Leiter, K. W.; Knap, J. Towards high throughput screening of electrochemical stability of battery electrolytes. Nanotechnology 2015, 26, 354003. (37) Shenzhen, X.; Luo, G.; Jacobs, R.; Fang, S.; Mahanthappa, M. K.; Hamers, R. J.; Morgan, D. Ab Initio Modeling of Electrolyte Molecule Ethylene Carbonate Decomposition Reaction on Li(Ni,Mn,Co)O2 Cathode Surface. ACS Appl. Mater. Interfaces 2017, 9, 20545−20553. (38) Tamura, T.; Kohyama, M.; Ogata, S. Combination of FirstPrinciples Molecular Dynamics and XANES Simulations for LiCoO2Electrolyte Interfacial Reactions in a Lithium-ion Battery. Phys. Rev. B 2017, 96, 035107. (39) Tebbe, J. L.; Fuerst, T. F.; Musgrave, C. B. Degradation of Ethylene Carbonate Electrolytes of Lithium Ion Batteries via Ring Opening Activated by LiCoO2 Cathode Surfaces and Electrolyte Species. ACS Appl. Mater. Interfaces 2016, 8, 26664−26674. (40) Amdouni, N.; Zaghib, K.; Gendron, F.; Mauger, A.; Julien, C. M. Structure and insertion properties of disordered and ordered LiNi0.5Mn1.5O4 spinels prepared by wet chemistry. Ionics 2006, 12, 117−126. (41) Chen, Z.; Zhu, H.; Ji, S.; Linkov, V.; Zhang, J.; Zhu, W. Performance of LiNi0.5Mn1.5O4 prepared by solid-state reaction. J. Power Sources 2009, 189, 507−510. J
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C (42) Fan, Y.; Wang, J.; Ye, X.; Zhang, J. Physical properties and electrochemical performance of LiNi0.5Mn1.5O4 cathode material prepared by a coprecipitation method. Mater. Chem. Phys. 2007, 103, 19−23. (43) Fang, H.-s.; Wang, Z.-x.; Li, X.-h.; Guo, H.-j.; Peng, W.-j. Exploration of high capacity LiNi0.5Mn1.5O4 synthesized by solid-state reaction. J. Power Sources 2006, 153, 174−176. (44) Liu, G. Q.; Wen, L.; Liu, Y. M. Spinel LiNi0.5Mn1.5O4 and its derivatives as cathodes for high-voltage Li-ion batteries. J. Solid State Electrochem. 2010, 14, 2191−2202. (45) Martha, S. K.; Sclar, H.; Szmuk Framowitz, Z.; Kovacheva, D.; Saliyski, N.; Gofer, Y.; Sharon, P.; Golik, E.; Markovsky, B.; Aurbach, D. A comparative study of electrodes comprising nanometric and submicron particles of LiNi0.50Mn0.50O2, LiNi0.33Mn0.33Co0.33O2, and LiNi0.40Mn0.40Co0.20O2 layered compounds. J. Power Sources 2009, 189, 248−255. (46) Martha, S. K.; Markevich, E.; Burgel, V.; Salitra, G.; Zinigrad, E.; Markovsky, B.; Sclar, H.; Pramovich, Z.; Heik, O.; Aurbach, D.; Exnar, I.; Buqa, H.; Drezen, T.; Semrau, G.; Schmidt, M.; Kovacheva, D.; Saliyski, N. A short review on surface chemical aspects of Li batteries: a key for a good performance. J. Power Sources 2009, 189, 288−296. (47) Thackeray, M. M.; Johnson, P. J.; de Picciotto, L. A.; Bruce, P. G.; Goodenough, J. B. Electrochemical extraction of lithium from LiMn2O4. Mater. Res. Bull. 1984, 19, 179−187. (48) Benedek, R.; Thackeray, M. M. Simulation of the surface structure of lithium manganese oxide spinel. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 195439. (49) Ouyang, C. Y.; Zeng, X. M.; Š ljivancanin, Ž .; Baldereschi, A. Oxidation States of Mn Atoms at Clean and Al2O3-Covered LiMn2O4(001) Surfaces. J. Phys. Chem. C 2010, 114, 4756−4759. (50) Benedek, R.; Thackeray, M. M.; Low, J.; Bučko, T. Simulation of Aqueous Dissolution of Lithium Manganate Spinel from First Principles. J. Phys. Chem. C 2012, 116, 4050−4059. (51) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. Electron-energy-loss spectra and the structural stability of nickel oxide: An LSDA+U study. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 57, 1505−1509. (52) Karim, A.; Fosse, S.; Persson, K. A. Surface structure and equilibrium particle shape of the LiMn2O4 spinel from first-principles calculations. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 87, 075322. (53) Lee, E.; Persson, K. A. First-principles study of the nano-scaling effect on the electrochemical behavior in LiNi0.5Mn1.5O4. Nanotechnology 2013, 24, 424007. (54) Chemelewski, K. R.; Shin, D. W.; Li, W.; Manthiram, A. Octahedral and truncated high-voltage spinel cathodes: the role of morphology and surface planes in electrochemical properties. J. Mater. Chem. A 2013, 1, 3347−3354. (55) Huang, J.; Liu, H.; Zhou, N.; An, K.; Meng, Y. S.; Luo, J. Enhancing the Ion Transport in LiMn1.5Ni0.5O4 by Altering the Particle Wulff Shape via Anisotropic Surface Segregation. ACS Appl. Mater. Interfaces 2017, 9, 36745−36754. (56) Huang, M.-R.; Lin, C.-W.; Lu, H.-Y. Crystallographic facetting in solid-state reacted LiMn2O4 spinel powder. Appl. Surf. Sci. 2001, 177, 103−113. (57) Car, R.; Parrinello, M. Unified Approach for Molecular Dynamics and Density-Functional Theory. Phys. Rev. Lett. 1985, 55, 2471−2474. (58) Giannozi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I.; Dal Corso, A.; Fabris, S.; de Gironcoli, S.; Fratesi, G.; Gebauer, R.; Gerstmann, U.; Gougoussis, C.; Kokalj, A.; Lazzeri, M.; MartinSamos, L.; Marzari, N.; Mauri, F.; Mazzarello, R.; Paolini, S.; Pasquarello, A.; Paulatto, L.; Sbraccia, C.; Scandolo, S.; Sclauzero, G.; Seitsonen, A. P.; Smogunov, A.; Umari, P.; Wentzcovich, R. M. QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials. J. Phys.: Condens. Matter 2009, 21, 395502.
(59) Blöchl, P. E.; Parrinello, M. Adiabaticity in first-principles molecular dynamics. Phys. Rev. B: Condens. Matter Mater. Phys. 1992, 45, 9413−9416. (60) Márquez, A. Molecular dynamics studies of combined carbon/ electrolyte/lithium-metal oxide interfaces. Mater. Chem. Phys. 2007, 104, 199−209. (61) Henkelman, G.; Jónsson, H. Improved tangent estimate in the nudged elastic band method for finding minimum energy paths and saddle points. J. Chem. Phys. 2000, 113, 9901−9904. (62) Anisimov, V. I.; Solovyev, I. V.; Korotin, M. A.; Czyżyk, M. T.; Sawatzky, G. A. Density-functional theory and NiO photoemission spectra. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 48, 16929. (63) Vanderrbilt, D. Soft self-consistent pseudopotentials in a generalized eigenvalue formalism. Phys. Rev. B: Condens. Matter Mater. Phys. 1990, 41, R7892. (64) Ma, X.; Kang, B.; Ceder, G. High Rate Micron-Sized Ordered LiNi0.5Mn1.5O4. J. Electrochem. Soc. 2010, 157, A925−A931. (65) Leung, K.; Leenheer, A. How Voltage Drops Are Manifested by Lithium Ion Configurations at Interfaces and in Thin Films on Battery Electrodes. J. Phys. Chem. C 2015, 119, 10234−10246. (66) Otani, M.; Sugino, O. First-principles calculations of charged surfaces and interfaces: A plane-wave nonrepeated slab approach. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 73, 115407. (67) Lee, K.; Murray, É . D.; Kong, L.; Lundqvist, B. I.; Langreth, C. D. Higher-accuracy van der Waals density functional. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, 081101. (68) Mayer, J. M. PROTON-COUPLED ELECTRON TRANSFER: A Reaction Chemist’s View. Annu. Rev. Phys. Chem. 2004, 55, 363−390.
K
DOI: 10.1021/acs.jpcc.8b10625 J. Phys. Chem. C XXXX, XXX, XXX−XXX