Computational Exploration of the Li-Electrode|Electrolyte Interface in

Sep 30, 2016 - ... becomes an important tool to understand the nanoscale processes and .... Direct monitoring of trace water in Li-ion batteries using...
5 downloads 4 Views 4MB Size
Download PDF
Article pubs.acs.org/accounts

Computational Exploration of the Li-Electrode|Electrolyte Interface in the Presence of a Nanometer Thick Solid-Electrolyte Interphase Layer Published as part of the Accounts of Chemical Research special issue “Nanoelectrochemistry”. Yunsong Li,† Kevin Leung,‡ and Yue Qi*,† †

Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, Michigan 48824, United States Sandia National Laboratories, Albuquerque, New Mexico 87185, United States

‡

CONSPECTUS: A nanometer thick passivation layer will spontaneously form on Li-metal in battery applications due to electrolyte reduction reactions. This passivation layer in rechargeable batteries must have “selective” transport properties: blocking electrons from attacking the electrolytes, while allowing Li+ ion to pass through so the electrochemical reactions can continue. The classical description of the electrochemical reaction, Li+ + e → Li0, occurring at the Limetal|electrolyte interface is now complicated by the passivation layer and will reply on the coupling of electronic and ionic degrees of freedom in the layer. This passivation layer is called “solid electrolyte interphase (SEI)” and is considered as “the most important but the least understood in rechargeable Liion batteries,” partly due to the lack of understanding of its structure−property relationship. Predictive modeling, starting from the ab initio level, becomes an important tool to understand the nanoscale processes and materials properties governing the interfacial charge transfer reaction at the Li-metal|SEI|electrolyte interface. Here, we demonstrate pristine Li-metal surfaces indeed dissolve in organic carbonate electrolytes without the SEI layer. Based on joint modeling and experimental results, we point out that the well-known two-layer structure of SEI also exhibits two different Li+ ion transport mechanisms. The SEI has a porous (organic) outer layer permeable to both Li+ and anions (dissolved in electrolyte), and a dense (inorganic) inner layer facilitate only Li+ transport. This two-layer/two-mechanism diffusion model suggests only the dense inorganic layer is effective at protecting Li-metal in electrolytes. This model suggests a strategy to deconvolute the structure−property relationships of the SEI by analyzing an idealized SEI composed of major components, such as Li2CO3, LiF, Li2O, and their mixtures. After sorting out the Li+ ion diffusion carriers and their diffusion pathways, we design methods to accelerate the Li+ ion conductivity by doping and by using heterogonous structure designs. We will predict the electron tunneling barriers and connect them with measurable first cycle irreversible capacity loss. Finally, we note that the SEI not only affects Li+ and e− transport, but it can also impose a potential drop near the Li-metal|SEI interface. Our challenge is to fully describe the electrochemical reactions at the Li-metal|SEI|electrolyte interface. This will be the subject of ongoing efforts. Li+ ion to pass through. Therefore, this passivation layer was called “solid electrolyte interphase (SEI)”.2,3 The idea of SEI was proposed by Peled in 1970s when studying the electrochemical behavior of the alkali metals in nonaqueous electrolytes for high energy battery systems.4 Due to the hazardous dendrite growth on Li-metal electrode, Limetal electrode was replaced with graphite, which made the commercial LIB possible in early 1990s, but at a cost of reduced capacity. The potential of lithiated graphite is still above the reduction potential of the electrolyte, therefore the success of LIB still relies on the quickly formed passivation SEI layer on graphite during the first charge. Additionally, the formation and continuous growth of SEI consume active Li+ ions and impede

1. INTRODUCTION Spontaneously formed passivation layers, as thin as nanometers, can kinetically block diffusion, hinder chemical reactions, and enable many important applications, such as stainless steel, aluminum boil kettle, and titanium ship in seawater. Compared to these examples, Li metal is much more active. It will react with O2, N2, and H2O in air and with the typical electrolytes used in Li-ion batteries (LIB). A passivation layer will spontaneously form on Li metal in battery applications, since the reduction potentials of the organic solvent in the electrolytes lie below the Fermi level referenced to Li+/Li(s) of Li-metal. Thus, upon contact, the electrolyte is reduced, decomposes, and forms a passivation layer.1 The passivation layer in a rechargeable battery must be “selective”. It should block electrons from attacking the electrolytes, while allowing © 2016 American Chemical Society

Received: July 15, 2016 Published: September 30, 2016 2363

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370

Article

Accounts of Chemical Research

Figure 1. AIMD snapshot of Li-metal|EC-liquid interface at (a) 0 ps and (b) 15 ps. (c) Orbital alignment at t = 0 ps. The EC decomposition is chemical, not due to long-range charge transfer, at this potential. Adapted from ref 23. Copyright 2013 American Chemical Society.

Li+ transport, thus critically impact the performance and life of LIBs. Over the past four decades, various studies have formed a general picture of the composition of SEI, which is a mixture of inorganic salts near the SEI|electrode interface, and organic salts near the SEI|electrolyte interface.2,5,6 Aurbach and coworkers,7,8 Xu,2,3 and Winter9 have provided extensive reviews on the topic. Despite these progresses, the basic transport properties and the structure−property relationship of SEI are still unclear and SEI is still claimed to be “the most important but the least understood in rechargeable Li-ion batteries.”9−11 Recent demands of vehicle electrification require a significant increase in the energy density compared to current LIB technologies in order to achieve an extended driving range at a lower cost. SEI design becomes critically needed for beyondgraphite anode materials, such as Li, Si, and Sn, where the SEI is especially unstable during repeated cycling. Recently, atomic layer deposition has been demonstrated as a promising technique to form conformal artificial SEI with precisely controlled distribution and thickness.12−14 How to design a chemically and mechanically stable multifunctional SEI coating with “selective” transport properties becomes the ultimate challenge. Li-metal gained renewed interests as it offers ultrahigh capacity, lowest reduction potential, and low density and is needed for high energy density and non-Li containing cathodes, such as Li−S, Li−air, and all-solid-Li-ion batteries.15 However, the fundamental transport steps at the Li|SEI|electrolyte interface are still not clear. Established electrochemical theory often considers the anodic reaction at a metal|electrolyte interface as Li+ + e− → Li0

starting from the ab initio level, becomes an important tool to address the fundamental questions and compute critical parameters of SEI. This Account will focus on our recent computational studies on the ionic and electronic transport properties of SEI components and our ongoing discussion on how SEI complicates the charge transfer reaction at a metal| liquid-electrolyte interface. Our strategy to deconvolute the structure−property relationships of SEI is to computationally interrogate the properties of idealized SEI inorganic components, such as Li2CO3,17,18 LiF,19,20 and their mixtures,21,22 and then correlate these properties with carefully designed physicalchemical and battery performance measurements.

2. DFT-BASED MULTISCALE MODELING Density functional theory (DFT) has been used to compute many important properties, such as defect formation energy,17,21 diffusion energy barrier,18 band structures, electron tunneling barrier, and reaction energy. DFT-based (“ab initio”) molecular dynamics (AIMD) has provided details of chemical and structural evolution for electrode|electrolyte interfaces.23 When DFT-based approach is limited by system size and simulation time, we have to apply other approximations. For example, instead of using more accurate constrained-DFT to predict electron transfer energetics, we used a less accurate but more efficient density functional tight binding (DFTB) method24,25 to include all the components of the Li|Li2CO3| liquid-EC interface. Furthermore, to make direct comparisons with or to better interpret experiments data, multiscale modeling is often required. We will show several examples of predicting experimentally measurable data. They are open circuit voltage of the electrode, isotope depth profile as a function of diffusion time, and the first cycle irreversible capacity vs the surface area of the electrode particles. Such multiscale modeling approach will be helpful to prove or disapprove fundamental mechanisms at the electrode|SEI| electrolyte interface. The goal is not only to understand but also to develop quantitative predictive models for each mechanism. Integration of verified quantitative battery failure models will eventually lead to battery life prediction without fitting parameters.

(1)

It has been found long ago that this reaction was insensitive to the concentration of Li+ ions in the solution,16 indicating it is oversimplified. This reaction is complicated by the SEI layer sandwiched between the electrode and electrolyte; therefore it depends on the coupling of electronic and ionic degrees of freedom in the SEI layer. For example, for a Li+ ion in the electrolyte to insert into the Li-metal, it will experience desolvation, transport through SEI, interfacial charge transfer, solid state diffusion, etc. If these steps are not optimized, metallic Li can nucleate and plate inside or outside of the SEI, rather than at the Li-metal|SEI interface, leading to low columbic efficiency and possible dendrite growth.4 Due to its thinness and the lack of in situ characterization tools to probe basic SEI properties, predictive modeling,

3. INDIVIDUAL PHASES AND INTERFACES 3.1. Reaction of Li with Typical Organic Electrolyte Used in LIB−Li|Electrolyte Interface

It can be readily demonstrated that, pristine Li-metal surfaces (no SEI film) dissolve in organic carbonate electrolytes.26 2364

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370

Article

Accounts of Chemical Research

is in coherent with the two-layer structure model of SEI, a porous (organic) outer layer permeable to both Li+ and anions (dissolved in electrolyte), and a dense (inorganic) inner layer facilitate only Li+ transport. A multiscale model was then developed and explained the remaining question: Why did the isotope 6Li+/7Li+ ratio peak at ∼5 nm beneath the outer surface? In this experiment, the dense SEI layer was predominantly Li2CO3 or Li2O. The dominant diffusion carrier in Li2CO3 on the negative electrode is excess interstitial Li+ (see subsection 3.3). DFT calculations predicted the energy barrier is 0.54 eV for the “direct-hopping” pathway (Li+ hopping through other empty interstitial sites without displacing lattice atoms) and 0.31 eV for the “knock-of f ” pathway (Li+ becoming a host lattice site by pushing a neighboring lattice Li+ into an adjacent interstitial site). Xu pointed out the “knock-of f ” pathway is similar to the “Grotthuss-like” proton conducting mechanism.3 Borodin also found this type of diffusion mechanism in Li2EDC.29 The knock-of f mechanism is favored since Li+ prefers to maintain a high O-coordination, rather than seeking the largest open channels in the Li2CO3 structure. We then formulated a onedimensional two-layer/two-mechanism diffusion model taking the DFT-predicted values and applied the initial and boundary conditions to mimic the isotope exchange experiments (Figure 2b). This model predicted the isotope ratio of 6Li+/7Li+ will saturate quickly at the outer layer of the SEI due to the fast pore diffusion assisted by the liquid electrolyte, but the 6 + 7 + Li / Li will keep increasing in the inner part of the SEI (Figure 2a), since only the knock-of f mechanism will allow 7Li+ already existed in the inorganic Li2CO3 to be replaced by 6Li+ in the electrolyte. Due to the lack of interstitial defects in Li2CO3, 6Li+ tends to accumulate on the surface of the Li2CO3 waiting to be knocked again by other interstitials. Therefore, the 6Li+/7Li+ peaks at the organic/inorganic interface.28 We would like to point out that not all the electrolyte decomposition products on the electrode surface will function as an “SEI” layer. If it is not dense, electrolyte can still diffuse through and be reduced, so it will not be fully effective. If it blocks ion transport, it shuts down further electrochemical reactions and should not be called “SEI” either. Therefore, although the electrochemically formed SEI can have complicated composition and structures, the densely packed inorganic layer should be the key component of an effective passivation layer to protect Li-metal in electrolytes. This brings us a much simpler model system, an anode covered by a dense inorganic layer, which will be referred as the “SEI” in the following discussions.

Figure 1 depicts AIMD simulations of 32 ethylene carbonate (EC) molecules on Li(100) surfaces. Periodic boundary conditions were used and no vacuum existed in the model. Within 15 ps, 12 EC molecules have received 2 electrons each and have decomposed.26 One CO32− and 11 OC2H4O2− were formed; the latter are expected to undergo further reactions to form SEI products.27 More significantly, 24 Li+ ions were released from the Li metal surface and bonded with the ionic products and/or solvent molecules. This happens despite the fact that the Li(100) surface is initially uncharged and is at a potential higher than Li+/Li0(s). The Li (uncharged bare surface) work function, Wf, is 2.9 eV, which translates into an anode potential of (2.9−1.4) = 1.5 V, with the 1.4 V being the conversion factor between electron in vacuum and the Li+/ Li(s) reference.23 This underscores the importance of surface passivation that blocks both chemical reactions and electron transfer to the liquid electrolyte. 3.2. Li Transport in SEI via a Two-Layer Model and the Knock-Off Mechanism in Li2CO3

Although SEI must allow Li+ ion to pass through, how this happens was not explicitly addressed until an intriguing isotope exchange experiment performed by Lu and Harris.28 They first formed SEI by using 7LiClO4 salt containing 90% isotope 7Li, rinsed it, and then soaked it in an electrolyte with a new 6LiBF4 salt. Time-of-flight secondary-ion-mass spectrometer (TOFSIMS) depth profiles showed that the anion BF4¯ only penetrated the top ∼5 nm of the SEI, but the cation 6Li+ has penetrated the ∼20 nm thick SEI and reached the SEI|currentcollector interface (Figure 2a). The size of BF4¯ (∼50 Å3) is comparable to an EC molecule (∼75 Å3); therefore, it is expected that the electrolyte can penetrate the top ∼5 nm. This

3.3. Increase the Ionic Conductivity in the Inorganic SEI Layer

If Li+ transport in the SEI is the rate-limiting step for the overall charge transfer reaction,4,28,30 how can we accelerate this process? The total Li+ ionic conductivity, σ, is the sum of contributions from each defect, such as interstitials, vacancies, Frenkel pairs, and Schottky pairs, etc. σ= Figure 2. (a) TOF-SIMS measured depth profiles of 6Li+/7Li+ and 11 + B for the SEI growing on a Cu substrate after 900 s of soaking and the calculated depth profiles of 6Li+/7Li+ (solid lines) after different soaking time. (b) Schematic of the two-layer/two-mechanism SEI diffusion model, pore diffusion in the porous organic layer and knockof f diffusion in the dense inorganic layer of SEI. Adapted from ref 17. Copyright 2012 American Chemical Society.

∑ σ(i , q) = ∑ i

i

F 2q2D(i , q) S(i , q) RT

(2)

where D(i,q) is the self-diffusion coefficients related to the diffusion energy barrier Em(i,q) and S(i,q) is the concentration related to the defect formation energy Ef(i,q). Since Em < Ef for Li+ interstitials in Li2CO3, the Li+ ion conduction is limited by the concentration of Li+ diffusing carriers. Consequently, 2365

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370

Article

Accounts of Chemical Research

Figure 3. Defect concentrations in (a) Li2CO3 (Adapted from ref 18. Copyright 2013 American Chemical Society) and (c) LiF (Adapted with permission from ref 19. Copyright 2015 American Physical Society); (b) possible defect reactions across the LiF|Li2CO3 interface (Adapted from ref 21. Copyright 2016 American Chemical Society).

thickness to eliminate electron leakage via mechanisms like electron tunneling.4 The following question is how thick does the dense inorganic layer need to be to block electron tunneling? This again requires DFT calculations to obtain the Wf and band structures for different SEI inorganic components and for Limetal. The electron tunneling barrier ΔEt is the energy from the Fermi level (εf) of Li-metal to the bottom of the conduction band of the SEI component (Figure 4a). The tunneling barrier on other anode can be shifted with voltage, for example, for LiC6, ΔEt will increase ∼0.1 eV. A critical thickness to block electron tunneling is therefore obtained by setting the tunneling probability extremely low in a 1D-WKB tunneling model. Only 2−3 nm thick LiF or Li2CO3 is needed to limit electron tunneling.20 One can assume the first cycle irreversible capacity loss, Cir, is the ratio of the Li+ consumed to form this passivation layer and those stored in the anode particle, therefore Cir can be a linear function of the surface area.20 The predicted slope (Figure 4b) shows good agreement with experimental results obtained on various carbon materials.34 This is a bit surprising, since no fitting parameters were used. This suggests that the initial SEI formation is likely to be controlled by the self-limiting electron tunneling property of the inorganic components; however, the continuous growth of SEI must be caused by other electron transport mechanisms, such as polaron, defects, interfaces, cracked SEI, and/or the instability of the outer SEI components, such as Li2EDC and other radicals.35−37

identifying the dominant diffusion carriers becomes a necessary step toward designing SEI with increased Li+ conductivity. S(i,q) is a function of the chemical potential of Li, μLi, which we assume has reached equilibrium between the SEI layer and the electrode it covers. So we predicted Ef(i,q) and S(i,q) for all possible point defects in Li2CO3 and LiF as a function of the open circuit voltage (OCV) of the electrode.18,19 Then defect concentrations were ranked over a voltage range (0 to 4.4 V) that includes typical anode and cathode materials (Figure 3a and c). We found Li+ interstitial is the main diffusion carrier in Li2CO3 at low OCV, such as on Li-metal; and Li+ vacancy becomes the main diffusion carrier in Li2CO3 at high OCV, such as on the cathode surface. In LiF, the dominating diffusion carrier is Li+ vacancies, which are balanced either by F− vacancies (as Schottky pairs) at low OCV or by electrons at high OCV. Increasing the diffusion carriers’ concentrations will efficiently increase Li+ ion conductivity. Thus, the computational results suggested doping Li2CO3 with PO43− replacing CO32− on an anode surface; while doping Li2CO3 with Ca2+ or/ and NO3− on the cathode surface. For LiF, it is well-known Mg2+ or Al3+ will increase Li+ vacancy concentration. Another way to increase diffusion carriers’ concentration is to design heterogeneous structures. Since both Li2CO3 and LiF are often found in SEI inorganic layer, we investigated all possible defect reactions across the LiF|Li2CO3 interface. We found it is energetically favorable to move Li+ from LiF lattice to the interstitial site in Li2CO3, causing accumulation of ionic carriers and depletion of electronic carriers near the LiF| Li2CO3. The reaction will be stopped by the interfacial space charge potential (Figure 3b), which is linked to the concentration of accumulated charge carriers through the Poison−Boltzmann relationship. By connecting the DFTpredicted defect reaction energies with the space-charging model, we predicted enhanced ionic conductivity in Li2CO3 and LiF nanocomposite SEI, which agreed with experimental observations.21,22

3.5. The Electrical Double Layer (EDL) and Potential Drop at the Li|SEI Interface

Established electrochemical theory describes EDL formation in the liquid near the surface of charged metal electrode. The SEI or passivation layer sandwiched between the Li-metal and liquid electrolyte raises the possibility that part of the potential drop occurs at the electrode|SEI interface or even within the SEI. Another example is the experimentally demonstrated Li storage capability at the interface between two materials, neither of which intercalates Li in the bulk state.38 This is accomplished by storing Li+ and e− separately at the two sides of the interface, leading to a finite surface density of electrostatic dipoles−in other words, EDL and potential drop. (Figure 5). Similar phenomena must occur on metal-electrode|SEI interfaces, such as Au(111)|Li2CO3(001)39 and Li(100)|Li2O-

3.4. Electron Tunneling Limited by the Inorganic Layer with a Critical Thickness

It has been proposed that SEI layers can block electrolyte reduction reactions either by blocking electron conduction4,20,31 or by limiting the solvent diffusion.32,33 The twolayer/two-mechanism diffusion model indicates that it is more critical for the inorganic layer to be dense and reach a critical 2366

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370

Article

Accounts of Chemical Research

(111). Here the true applied voltage, directly related to the electrode Fermi energy, must be matched to the Li chemical potential in that electrode to achieve equilibrium. Current DFT calculations are conducted at constant number of atoms, not constant chemical potential, and the equilibrium condition must be maintained manually in the slab geometry. In the literature, mostly coherent Li|electrolyte interfaces have been studied.40 We found that inserting more Li in an interlayer at the quasi-coherent Li(100)|Li2O(111) interface, coordinated to the O atoms of the oxide and Li atoms of the metal surface, renders the system more stable after accounting for Li chemical potential. Bader charge analysis reveals that the Li interlayer is all Li+, electrostatically compensated by a negative surface charge on the Li surface. The existence of Li+ right at the interface emphasizes that Li-insertion into SEI-coated electrodes is very different from metal plating at pristine liquid−solid interfaces.41 Electrons are not transferred to Li+ until the Li+ is transported through the oxide film. Note the above calculations were at the electrode|SEI interface. When liquid electrolyte is present outside the SEI layer, the proportion of the EDL residing in the solid and liquid must be determined by calculating the respective free energy costs. Finite temperature (T) effects are not yet included. T > 0 may be important for realistic electrochemical interfaces because the Debye screening length is proportional to T0.5. This will be examined using more coarse-grained modeling techniques including DFTB.

4. THE Li-METAL|Li2CO3|LIQUID-EC-ELECTROLYTE INTERFACE COMPLEX To form a more complete picture of the charge transfer reaction on an SEI-covered Li-electrode, we need to consider three steps: Li+ desolvation at the SEI|electrolyte interface, e− or Li+ diffusion through the interfacial layer, and annihilation of e− and Li+ (the charge transfer reaction). DFTB method24,25 was found to be an appropriate method to model this complex interface as it permits much longer simulation length and time scales compared to DFT. DFTB is a fast and efficient quantum mechanical simulation method based on a second-order expansion of the Kohn−Sham total energy in DFT with respect to charge density fluctuations. It balances the computational efficiency and accuracy required for this system.

Figure 4. (a) Aligning the Fermi level (εf), work function (Wf), and band gap (Eg) of the Li-metal and SEI to obtain the electron tunneling barrier (ΔEt), where εf of electrolyte means the small polaron level to accept one electron. (b) The predicted initial irreversible capacity loss due to SEI component formation. Adapted with permission from ref 20. Copyright 2016 Elsevier.

4.1. DFTB Parametrization and Model Details

The wave function confinement radius and the repulsive cutoffs in DFTB must be parametrized carefully to describe both metallic and ionic Li accurately. We have obtained the DFTB parameters by fitting to DFT results of a broad range of structures (BCC, FCC, and HPC, Li metal, Li2O, LiOH, LiH, LiCH3, and LiC6). The DFTB can accurately capture electronic structures, such as band structures of Li and Li2O and HOMO and LUMO of EC molecules. It can also reproduce the interface energies of Li|Li2CO3 as 0.50 J/m2 (0.64 J/m2 according to DFT) and Li+ solvation energy in EC as 5.4 eV (5.2−5.5 eV according to DFT23,42), in good agreement with reported DFT result (not included in the fitting). This set of parameters enable us to study Li stripping and plating at a Li2CO3 coated Li-metal in liquid EC electrolyte. We consider a Li(001)|Li2CO3(001)|EC slab model. The reduced state is represented by a perfect Li-metal slab, covered by 4 layers of Li2CO3 with 32 EC liquid molecules, as shown in Figure 6a. The optimized Li|Li2CO3 interface undergoes obvious atomic relaxation. One oxidized state is represented

Figure 5. Potential drop at (a) Li|electrolyte and (b) Li|SEI|electrolyte interface, where EDL is partially inside SEI and the vertical dashed line corresponds to the SEI/Electrolyte interface. Adapted from ref 39. Copyright 2015 American Chemical Society.

2367

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370

Article

Accounts of Chemical Research

Figure 6. Li|Li2CO3|EC slab model at (a) reduced state and (b) oxidized state. The cell size is 16.72 × 9.95 × 56.91 Å3.

by removing a Li+ from the Li-metal slab and putting it inside the electrolyte while leaving the electron on the Li-metal slab. All DFTB minimization calculations are performed after relaxing the structure with COMPASS force field. During DFTB optimization, only atomic positions were relaxed until a maximum force component of 10−4 eV/Å was achieved. The Brillouin zone was sampled with a 2 × 2 × 2 mesh including the Γ point. The Lennard-Jones dispersion model was used to correct van der Waals interactions.43 To locate the electron, we have computed atomic charges based on Mulliken population for an atomic-like basis. In the oxidized state, the Li+ in the electrolyte lost 0.62e and was surrounded by 5 EC molecules (highlighted in Figure 7b), agreeing with the Li ion solvation structure.30 The total charges on the Li-metal slab, the Li2CO3 layer, and the EC electrolyte showed that they have gained −0.47e, −0.05e, and −0.1e, respectively, indicating the electron is mainly located on the Li-metal slab. Additional DFTB simulations also confirmed that if there is no Li2CO3 or if the thickness of Li2CO3 is less than 4 layers, the electron will reside in the electrolyte and participate in electrolyte decomposition reactions after geometry optimization, as DFT predicted in subsection 3.1. Vacancy generation accompanies Li removal, which can occur from the surface (case 1) or from the bulk (case 2) of the Li-metal slab (Figure 6b). The calculated total energy of case 1 is 0.3 eV lower than that of case 2, suggesting Li atom will be stripped from the surface first. Therefore, case 1 shown in Figure 6b is chose as the model of the oxidized state.

Figure 7. (a) Thermodynamics cycle of charge transfer reaction. (b) Energetic coordinates for Li+ transport at Li(001)|Li2CO3(001)|EC interface.

Li n − 1(metal, with VLi) + x Li+(EC)m + xe−(on Li metal) ΔG

⎯→ ⎯ Li n(metal) + (EC)m

(3.1)

ΔG(x = 1) = −Ef (VLi) − Evaporization − E ionization + Wf + Esolvation(Li+)

(3.2)

To find the zero voltage of Li /Li -in-EC, several oxidized states with varying electron density xe− will be computed until ΔG = 0. This method has already been developed by Leung for LiC6|EC-electrolyte interface with DFT calculations.23 Here we present of the case of x = 1, which corresponds to an electron density of 0.002 e/Å2 on Li-metal surface (surface area is 166.27 Å2). The energy of this oxidized state is 1.42 eV lower than the reduced state given in Figure 6a. This energy difference is reasonable considering the thermodynamics cycle shown in eq 3 and Figure 7a, which leads to a ΔG ∼ 1.2 eV, consistent with the DFTB prediction, modulo errors. We used the DFT (with GGA/PBE functional) computed solvation energy of 5.5 eV for gas phase Li+ ion in 5 EC,42 in the estimation, while this value is debated to be 4.4 eV for Li+(EC)4 from Gaussian with Hartree−Fock method,44 to 5.2 eV from AIMD method with GGA/PBE functional at 450 K,23 and 0

4.2. The Charge Transfer Reaction Energy at a Li-Metal|Li2CO3| EC-Electrolyte Interface

The charge transfer reaction in eq 3.1 needs to be rewritten to include the insulating effect of the SEI layer. 2368

+

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370

Article

Accounts of Chemical Research possibly even lower with entropy cost.42 To put the ionized electron back into the Li-metal slab, the reverse of Wf = 2.9 eV is used in the thermodynamics cycle calculation. When the potential decreases, the negative charge density increases on Limetal, Wf will then decrease, and thus the energy of the oxidized state and ΔG will increase accordingly. Some of the Wf decrease may occur in the SEI layer (subsection 3.5). More detailed analysis is needed. We then computed the energies for two intermediate states (Figure 7b): Li+ ion desolvation at the Li2CO3|EC-electrolyte interface and Li+ ion diffusing in the Li2CO3, in order to probe the energy landscape of the charge transfer reaction. In these intermediate states, the electron is also located in Li slab. During the desolvation process, we found Li+ is already absorbed on the Li2CO3 surface after two EC molecules were stripped away from the primary solvation sheath. The relatively high energy barrier observed in Figure 7b might be due to the perfect crystalline Li2CO3 used in the current model. Since Li+ ion transport in Li2CO3 is facilitated by Li+ interstitials, the preexisting interstitial concentration at room temperature may reduce this barrier. The electron potential, liquid state fluctuations, and effects of finite temperature will also alter the energy barrier. We will apply DFTB-based MD simulations and liquid-state free energy methods to re-examine these issues, and to study how much of the EDL is located at the solid−solid interface (subsection 3.5). Nevertheless, this energy landscape provides an important input for thermodynamics based phase-field model of Li plating and dendrite morphology evolution,45,46 which are closely related to the role of SEI,47 which can alter the interface reaction kinetics dramatically.

Biographies Yunsong Li is a postdoctoral researcher at the Michigan State University in the Department of Chemistry Engineering and Material Science. He received his Ph.D. in Materials Physics and Chemistry from Xiamen University in 2012. Currently he focuses on multiscale simulations of the electron and ion transport in interphases. Kevin Leung is a principal research scientist who has worked at Sandia National Laboratories for 18 years. He has specialized in modeling batteries, liquid−solid interfaces, geochemistry, and chemical/electrochemical reactions. Yue Qi is an associate professor at the Michigan State University in the Department of Chemistry Engineering and Material Science. She received her B.S. from Tsinghua University and Ph.D. at Caltech. She worked in General Motors R&D for 12 years before returning to academia in 2013 and starting the “Material Simulation for Clean Energy” lab at MSU.

■

ACKNOWLEDGMENTS We acknowledge the support for degradation mechanism modeling as part of Nanostructures for Electrical Energy Storage (NEES), an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under Award Number DESC0001160. Y.S.L. and Y.Q. also acknowledge the support from NSF GOALI under CMMI-1235092. Sandia National Laboratories is a multiprogram laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC0494AL85000.

■

5. MODEL LIMITATIONS In this Account, we have tried to illustrate how predictive modeling was used to address the fundamental questions related to the electronically insulating but ionic conductive solid electrolyte interface (SEI) in Li-ion batteries. We mainly focused our discussions on Li-metal electrode, while many conclusions are general to other anode materials. Using a crystalline model to represent the dense inorganic SEI layer, we have gained many new insights on Li ion transport, electron tunneling, potential drop inside the SEI, and desolvation on SEI. We just begin to understand the charge transfer reaction process at the Li|SEI|electrolyte interface complex. This model may give us new insights for artificial SEI coating design, but it is oversimplified. In the future, we should consider the multicomponent and multifunctional nature of the SEI. For example, how electrons and ions transport through grain boundaries in the mosaic SEI inorganic phase. We have modeled SEI statically. However, it may evolve due to growth, dissolution, redeposition, and densification of porous phase. SEI formation on Li-metal surface and on artificial SEI with few atomic layers has been explicitly modeled. However, what makes SEI continuously grow beyond its tunneling limit thickness requires immediate attention, since it closely related to the capacity fading of a Li-ion batteries.

■

REFERENCES

(1) Goodenough, J. B.; Kim, Y. Challenges for Rechargeable Li Batteries. Chem. Mater. 2010, 22, 587−603. (2) Xu, K. Nonaqueous Liquid Electrolytes for Lithium-Based Rechargeable Batteries. Chem. Rev. 2004, 104, 4303−4418. (3) Xu, K. Electrolytes and Interphases in Li-ion Batteries and Beyond. Chem. Rev. 2014, 114, 11503−11618. (4) Peled, E. The Electrochemical-Behavior of Alkali and AlkalineEarth Metals in Non-Aqueous Battery Systems - the Solid Electrolyte Interphase Model. J. Electrochem. Soc. 1979, 126, 2047−2051. (5) Aurbach, D.; Eineli, Y.; Chusid, O.; Carmeli, Y.; Babai, M.; Yamin, H. The Correlation between the Surface-Chemistry and the Performance of Li-Carbon Intercalation Anodes for Rechargeable Rocking-Chair Type Batteries. J. Electrochem. Soc. 1994, 141, 603−611. (6) Balbuena, P. B.; Wang, Y. Lithium-Ion Batteries: Solid-Electrolyte Interphase; Imperial College Press: London, 2004. (7) Aurbach, D.; Zaban, A.; Ein-Eli, Y.; Weissman, I.; Chusid, O.; Markovsky, B.; Levi, M.; Levi, E.; Schechter, A.; Granot, E. Recent Studies on the Correlation between Surface Chemistry, Morphology, Three-Dimensional Structures and Performance of Li and Li-C Intercalation Anodes in Several Important Electrolyte Systems. J. Power Sources 1997, 68, 91−98. (8) Aurbach, D.; Zinigrad, E.; Cohen, Y.; Teller, H. A Short Review of Failure Mechanisms of Lithium Metal and Lithiated Graphite Anodes in Liquid Electrolyte Solutions. Solid State Ionics 2002, 148, 405−416. (9) Winter, M. The Solid Electrolyte Interphase - the Most Important and the Least Understood Solid Electrolyte in Rechargeable Li Batteries. Z. Phys. Chem. 2009, 223, 1395−1406. (10) Tarascon, J. M.; Armand, M. Issues and Challenges Facing Rechargeable Lithium Batteries. Nature 2001, 414, 359−367. (11) Whittingham, M. S. Lithium Batteries and Cathode Materials. Chem. Rev. 2004, 104, 4271−4302.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest. 2369

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370

Article

Accounts of Chemical Research (12) Jung, Y. S.; Cavanagh, A. S.; Riley, L. A.; Kang, S. H.; Dillon, A. C.; Groner, M. D.; George, S. M.; Lee, S. H. Ultrathin Direct Atomic Layer Deposition on Composite Electrodes for Highly Durable and Safe Li-ion Batteries. Adv. Mater. 2010, 22, 2172−2176. (13) Xiao, X.; Lu, P.; Ahn, D. Ultrathin Multifunctional Oxide Coatings for Lithium Ion Batteries. Adv. Mater. 2011, 23, 3911−3915. (14) Kozen, A. C.; Lin, C. F.; Pearse, A. J.; Schroeder, M. A.; Han, X. G.; Hu, L. B.; Lee, S. B.; Rubloff, G. W.; Noked, M. Next-Generation Lithium Metal Anode Engineering Via Atomic Layer Deposition. ACS Nano 2015, 9, 5884−5892. (15) Bruce, P. G.; Freunberger, S. A.; Hardwick, L. J.; Tarascon, J.-M. Li-O2 and Li-S Batteries with High Energy Storage. Nat. Mater. 2012, 11, 19−29. (16) Meibuhr, S. G. Electrode Studies in Nonaqueous Solvents. II. Anion Effect on Kinetics of Li/Li+ in Propylene Carbonate. J. Electrochem. Soc. 1971, 118, 1320−1322. (17) Shi, S.; Lu, P.; Liu, Z.; Qi, Y.; Hector, L. G.; Li, H.; Harris, S. J. Direct Calculation of Li-ion Transport in the Solid Electrolyte Interphase. J. Am. Chem. Soc. 2012, 134, 15476−15487. (18) Shi, S.; Qi, Y.; Li, H.; Hector, L. G. Defect Thermodynamics and Diffusion Mechanisms in Li2CO3 and Implications for the Solid Electrolyte Interphase in Li-ion Batteries. J. Phys. Chem. C 2013, 117, 8579−8593. (19) Pan, J.; Cheng, Y. T.; Qi, Y. General Method to Predict VoltageDependent Ionic Conduction in a Solid Electrolyte Coating on Electrodes. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 134116. (20) Lin, Y.-X.; Liu, Z.; Leung, K.; Chen, L.-Q.; Lu, P.; Qi, Y. Connecting the Irreversible Capacity Loss in Li-Ion Batteries with the Electronic Insulating Properties of Solid Electrolyte Interphase (SEI) Components. J. Power Sources 2016, 309, 221−230. (21) Pan, J.; Zhang, Q.; Xiao, X.; Cheng, Y.-T.; Qi, Y. Design of Nanostructured Heterogeneous Solid Ionic Coatings through a Multiscale Defect Model. ACS Appl. Mater. Interfaces 2016, 8, 5687−5693. (22) Zhang, Q.; Pan, J.; Lu, P.; Liu, Z.; Verbrugge, M. W.; Sheldon, B. W.; Cheng, Y.-T.; Qi, Y.; Xiao, X. Synergetic Effects of Inorganic Components in Solid Electrolyte Interphase on High Cycle Efficiency of Lithium Ion Batteries. Nano Lett. 2016, 16, 2011−2016. (23) Leung, K.; Tenney, C. M. Toward First Principles Prediction of Voltage Dependences of Electrolyte/Electrolyte Interfacial Processes in Lithium Ion Batteries. J. Phys. Chem. C 2013, 117, 24224−24235. (24) Elstner, M.; Porezag, D.; Jungnickel, G.; Elsner, J.; Haugk, M.; Frauenheim, T.; Suhai, S.; Seifert, G. Self-Consistent-Charge DensityFunctional Tight-Binding Method for Simulations of Complex Materials Properties. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 58, 7260−7268. (25) Frauenheim, T.; Seifert, G.; Elstner, M.; Niehaus, T.; Kohler, C.; Amkreutz, M.; Sternberg, M.; Hajnal, Z.; Di Carlo, A.; Suhai, S. Atomistic Simulations of Complex Materials: Ground-State and Excited-State Properties. J. Phys.: Condens. Matter 2002, 14, 3015− 3047. (26) Yu, J. M.; Balbuena, P. B.; Budzien, J.; Leung, K. Hybrid DFT Functional-Based Static and Molecular Dynamics Studies of Excess Electron in Liquid Ethylene Carbonate. J. Electrochem. Soc. 2011, 158, A400−A410. (27) Leung, K. Two-Electron Reduction of Ethylene Carbonate: A Quantum Chemistry Re-Examination of Mechanisms. Chem. Phys. Lett. 2013, 568, 1−8. (28) Lu, P.; Harris, S. J. Lithium Transport within the Solid Electrolyte Interphase. Electrochem. Commun. 2011, 13, 1035−1037. (29) Borodin, O.; Zhuang, G. R. V.; Ross, P. N.; Xu, K. Molecular Dynamics Simulations and Experimental Study of Lithium Ion Transport in Dilithium Ethylene Dicarbonate. J. Phys. Chem. C 2013, 117, 7433−7444. (30) Xu, K.; von Cresce, A.; Lee, U. Differentiating Contributions to ″Ion Transfer″ Barrier from Interphasial Resistance and Li+ Desolvation at Electrolyte/Graphite Interface. Langmuir 2010, 26, 11538− 11543.

(31) Li, D. J.; Danilov, D.; Zhang, Z. R.; Chen, H. X.; Yang, Y.; Notten, P. H. L. Modeling the SEI-Formation on Graphite Electrodes in LiFePO4 Batteries. J. Electrochem. Soc. 2015, 162, A858−A869. (32) Ploehn, H. J.; Ramadass, P.; White, R. E. Solvent Diffusion Model for Aging of Lithium-Ion Battery Cells. J. Electrochem. Soc. 2004, 151, A456−A462. (33) Pinson, M. B.; Bazant, M. Z. Theory of SEI Formation in Rechargeable Batteries: Capacity Fade, Accelerated Aging and Lifetime Prediction. J. Electrochem. Soc. 2013, 160, A243−A250. (34) Joho, F.; Rykart, B.; Blome, A.; Novak, P.; Wilhelm, H.; Spahr, M. E. Relation between Surface Properties, Pore Structure and FirstCycle Charge Loss of Graphite as Negative Electrode in Lithium-Ion Batteries. J. Power Sources 2001, 97−8, 78−82. (35) Xu, K.; Zhuang, G. V.; Allen, J. L.; Lee, U.; Zhang, S. S.; Ross, P. N.; Jow, T. R. Syntheses and Characterization of Lithium Alkyl Monoand Dicarbonates as Components of Surface Films in Li-Ion Batteries. J. Phys. Chem. B 2006, 110, 7708−7719. (36) Nie, M.; Chalasani, D.; Abraham, D. P.; Chen, Y.; Bose, A.; Lucht, B. L. Lithium Ion Battery Graphite Solid Electrolyte Interphase Revealed by Microscopy and Spectroscopy. J. Phys. Chem. C 2013, 117, 1257−1267. (37) Soto, F. A.; Ma, Y.; Martinez de la Hoz, J. M.; Seminario, J. M.; Balbuena, P. B. Formation and Growth Mechanisms of SolidElectrolyte Interphase Layers in Rechargeable Batteries. Chem. Mater. 2015, 27, 7990−8000. (38) Fu, L. J.; Chen, C. C.; Samuelis, D.; Maier, J. Thermodynamics of Lithium Storage at Abrupt Junctions: Modeling and Experimental Evidence. Phys. Rev. Lett. 2014, 112, 208301. (39) 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. (40) Lepley, N. D.; Holzwarth, N. A. W. Modeling Interfaces between Solids: Application to Li Battery Materials. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 92, 214201. (41) Pinto, L. M. C.; Quaino, P.; Santos, E.; Schmickler, W. On the Electrochemical Deposition and Dissolution of Divalent Metal Ions. ChemPhysChem 2014, 15, 132−138. (42) Skarmoutsos, I.; Ponnuchamy, V.; Vetere, V.; Mossa, S. Li+ Solvation in Pure, Binary, and Ternary Mixtures of Organic Carbonate Electrolytes. J. Phys. Chem. C 2015, 119, 4502−4515. (43) Zhechkov, L.; Heine, T.; Patchkovskii, S.; Seifert, G.; Duarte, H. A. An Efficient a Posteriori Treatment for Dispersion Interaction in Density-Functional-Based Tight Binding. J. Chem. Theory Comput. 2005, 1, 841−847. (44) Yanase, S.; Oi, T. Solvation of Lithium Ion in Organic Electrolyte Solutions and Its Isotopie Reduced Partition Function Ratios Studied by Ab Initio Molecular Orbital Method. J. Nucl. Sci. Technol. 2002, 39, 1060−1064. (45) Bazant, M. Z. Theory of Chemical Kinetics and Charge Transfer Based on Nonequilibrium Thermodynamics. Acc. Chem. Res. 2013, 46, 1144−1160. (46) Liang, L.; Qi, Y.; Xue, F.; Bhattacharya, S.; Harris, S. J.; Chen, L.-Q. Nonlinear Phase-Field Model for Electrode-Electrolyte Interface Evolution. Phys. Rev. E 2012, 86, 051609. (47) Cohen, Y. S.; Cohen, Y.; Aurbach, D. Micromorphological Studies of Lithium Electrodes in Alkyl Carbonate Solutions Using in Situ Atomic Force Microscopy. J. Phys. Chem. B 2000, 104, 12282− 12291.

2370

DOI: 10.1021/acs.accounts.6b00363 Acc. Chem. Res. 2016, 49, 2363−2370