Article pubs.acs.org/JPCC
Chemical Stability of Lithium 2‑Trifluoromethyl-4,5dicyanoimidazolide, an Electrolyte Salt for Li-Ion Cells Ilya A. Shkrob,*,† Krzysztof Z. Pupek,‡ James A. Gilbert,† Stephen E. Trask,† and Daniel P. Abraham*,† †
Chemical Sciences and Engineering Division, and ‡Energy Systems Division, Materials Engineering Research Facility, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, Illinois 60439, United States S Supporting Information *
ABSTRACT: Lithium hexafluorophosphate (LiPF6) is ubiquitous in commercial lithium-ion batteries, but it is hydrolytically unstable and corrosive on electrode surfaces. Using a more stable salt would confer multiple benefits for high-voltage operation, but many such electrolyte systems facilitate anodic dissolution and pitting corrosion of aluminum current collectors that negate their advantages. Lithium 2-trifluoromethyl-4,5-dicyanoimidazolide (LiTDI) is a new salt that was designed specifically for high-voltage cells. In this study we demonstrate that in carbonate electrolytes, LiTDI prevents anodic dissolution of Al current collectors, which places it into a select group of corrosion inhibitors. However, we also demonstrate that LiTDI becomes reduced on lithiated graphite, undergoing sequential defluorination and yielding a thick and resistive solid-electrolyte interphase (SEI), which increases impedance and lowers electrode capacity. The mechanistic causes for this behavior are examined using computational chemistry methods in light of recent spectroscopic studies. We demonstrate that LiTDI reduction can be prevented by certain electrolyte additives, which include fluoroethylene carbonate, vinylene carbonate, and lithium bis(oxalato)borate. This beneficial action is due to preferential reduction of these additives over LiTDI at a higher potential vs Li/Li+, so the resulting SEI can prevent the direct reduction of LiTDI at lower potentials on the graphite electrode.
1. INTRODUCTION Commercial lithium ion batteries (LIBs) typically have liquid carbonate electrolytes containing molar concentration of lithium salts, of which the most common is LiPF6.1,2 Due to reversible dissociation to PF5 and F−, the hexafluorophosphate anion is suboptimal,3 as the protonation of the fluoride and reactions of PF5 yield hydrofluoric acid that corrodes positive electrodes4,5 and can react with components of negatives electrode (e.g., the SiO2 layer on nanosilicon particles in LixSi electrodes).6−9 Despite these well-known disadvantages, LiPF6 remains the material of preference, as for the great majority of other lithium salts (that are more hydrolytically and solvolytically stable than LiPF6) there is rapid anodic dissolution and pitting corrosion of aluminum current collectors above a certain voltage (∼4.0 V vs Li/Li+).3 This corrosion problem is especially vexing for high-voltage LIBs.10−12 The very chemical and electrochemical stability of hard-to-oxidize anions (such as bis(trifluoromethylsulfonyl)limide, TFSI − , and bis(fluorosulfonyl)imide, FSI−, shown in Scheme 1) becomes their downfall,13−15 as at the exposed aluminum surface, these anions combine with Al3+ ions to form soluble complexes, which sustain pitting corrosion.16−18 In contrast, the poor chemical stability of PF6− and BF4− anions becomes beneficial, as the released fluoride combines with the aluminum oxide to yield a firm protective barrier.1,2,18−20 Because (from a practical © XXXX American Chemical Society
Scheme 1. Examples of Hydrolytically Stable Anions
standpoint) aluminum is the only cheap, light, conductive, and malleable metal for a cathode current collector,18,21 much effort has been invested into finding new lithium salts that (i) inhibit this Al anodic dissolution, while (ii) being sufficiently stable chemically at high voltages on energized electrodes. Lithium 4,5-dicyano-2-(trifluoromethyl)imidazolide (LiTDI, see Scheme 1) was developed by Niedzicki and co-workers for lithium-ion cells as an alternative to LiPF6.22−26 Imidazole is an N-heterocyclic base; electron-withdrawing cyanide and trifluoromethyl groups are included to turn it into a strong acid, so the anion cannot be easily protonated.24,27,28 LiTDI can be Received: September 28, 2016 Revised: November 30, 2016 Published: December 1, 2016 A
DOI: 10.1021/acs.jpcc.6b09837 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
radiolysis, spectroscopy, and computational chemistry to account for these observations. Our general intent is to demonstrate that while LiTDI is less stable on lithiated graphite than LiPF6, this problem can be alleviated by the inclusion of appropriate electrolyte additives. Given that LiTDI appears to be stable on energized positive electrodes and Al collectors,22,24 there is a strong incentive to further pursue this system even though it requires more complex electrolytes. The supporting schemes, tables, and figures have been placed in the Supporting Information. When referenced in the text, these materials have the designator “S”, as in Figure S1.
produced conveniently and in high yield in a one-pot synthesis, and it is thermally stable to 250 °C.22,29 The potential of 4.6 V vs Li/Li+ for the onset of Al anodic currents was estimated for LiTDI in carbonate electrolytes using cyclic voltammetry,22 and this compound began to attract increasing attention, especially for high-energy/high-voltage batteries containing silicon (Li−Si) and graphite (Li−Gr) based negative electrodes. However, before it can be more widely used, its behavior on the negative electrode needs to be understood. Preliminary experiments suggest that LiTDI-based electrolytes are at least as stable as LiPF6-based electrolytes on the positive electrode, whereas on Li-Gr electrodes the electrolyte yields a thick and unstable solid-electrolyte interphase (SEI).30 The latter is a protective coating, composed of electrolyte breakdown products, that isolates the reducing Li−Gr surfaces from the electrolyte thereby preventing its runaway decomposition.2,31−33 To operate properly, this barrier needs to conduct Li+ ions while blocking the diffusion of solvent molecules toward the reactive surface where they become reduced; the thickness of this barrier should not increase during cycling, as that impedes Li+ ion migration and ties them up (as more solvent becomes reduced), depleting the stock of available lithium. For LiTDI -containing electrolytes, these requirements are not met, and there is significant capacity loss.30 However, it was also observed that the addition of 2 wt % fluoroethylene carbonate (FEC, Scheme 2) improved cell performance.30
2. EXPERIMENTAL SECTION Unless specified otherwise, all chemicals were obtained from Sigma-Aldrich. Battery grade fluoroethylene carbonate was obtained from Solvay. LiPF6 was obtained from Strem Chemicals, and LiTDI was synthesized at Argonne’s Materials Engineering Research Facility (MERF) following the method given in ref 22. The electrolyte solvent was 3:7 wt/wt liquid mixture of ethylene carbonate (EC, Scheme 2) and ethyl methyl carbonate (EMC, Scheme 2). LiTDI has good solubility in this solvent (>0.5 M), but it has poor solubility in EMC alone (ca. 87 mM at saturation at 25 °C, as determined by 19F NMR). All electrochemical tests were conducted using a MACCOR Series 4000 test unit. Stainless steel based 2032-type coin cells were used, unless stated otherwise. The Gr/Li (“half-cell”) cells A−E (see Table 1) containing either 0.5 M LiTDI (cells B−E)
Scheme 2. Chemical Structures and C−H Bond Energies (in eV) for Various Carbonate Molecules Compared to a Hypothetical Product TDI(H)− of H Atom Abstraction by F Loss Radical of TDI− Anion
Table 1. Chemical Compositions for Electrolytes cell
salt
M
additive
wt %
A B C D E
LiPF6
1.2
LiTDI
0.5
− − FEC LiBOB VC
− − 10 2 5
or 1.2 M LiPF6 (cell A) were in an Ar-filled glovebox and cycled at 30 °C. Cells C−E contained electrolyte additives: 10 wt % FEC (see Scheme 2), 5 wt % VC (vinylene carbonate, see Scheme S1 in the Supporting Information), or 2 wt % LiBOB (lithium bis(oxalate)borate, see Scheme S1), respectively. The Gr electrode was prepared at Argonne’s Cell Analysis, Modeling, and Prototyping (CAMP) facility and contained 91 wt % Hitachi MAGE graphite, 2 wt % SUPER C45 conductive carbon black (from TIMCAL C-NERGY), 0.17 wt % oxalic acid for improved adhesion to copper, and 6 wt % KF9300 Kureha PVdF polymer as a binder; these materials formed a ∼40 μm thick composite coating on a 10 μm thick copper current collector. The electrode area was 1.77 cm2, the coating loading density was 6.2 mg/cm2 (5.9 mg/cm2 active loading), and the electrode porosity was 30%. Celgard 2325 was used as a microporous separator (25 μm thick, 28 nm pore, 40% porosity). The electrochemical test conditions for the Gr/Li cells are given in Table 2, and Table S1 summarizes the results. All potentials reported in this study are given vs Li/Li+. The typical test involved three galvanostatic cycles with the potential changing between 1.5 V and 1 mV. The lithiation cycling was carried out at a nominal C/20 rate; the actual time for the charge−discharge cycles is shown in Table S1. For delithiation, the C/20 rate for cycles 1−3 was followed by four additional
In this study, we follow up on the chemical stability of LiTDI based electrolytes and suggest a rationale for the observed behavior. We first characterize anodic Al dissolution for LiTDI using the chronoamperometry methodology introduced in ref 3. We then compare the cycling behavior of Gr/Li cells containing LiPF6 and LiTDI at different charging rates in solutions with and without electrolyte additives. Our results indicate that cells with LiTDI alone display higher impedance, with concomitant capacity loss. Apparently, the products of LiTDI-containing electrolyte decomposition yield an inferior SEI, which decreases the rate of Li+ migration, as compared to LiPF6-based electrolytes; addition of certain compounds mitigates this behavior. Finally, we combine the results of B
DOI: 10.1021/acs.jpcc.6b09837 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
semiempirical PM7 method37,38 from MOPAC2016 program suite39 was used.
Table 2. Charge/Discharge Rates for the Different Cell Cycles charge/discharge nominal rate, C/. cycle no.
delithiation
lithiation
charge current, mA
1−3 4 5 6 7
20 10 5 2 1
20 20 20 20 20
0.17 0.33 0.66 1.65 3.30
3. RESULTS AND DISCUSSION 3.1. Inhibition of Anodic Dissolution of Al Current Collector by LiTDI. Potential and current response profiles during a typical electrochemical test for 0.5 M LiTDI electrolyte are shown in Figure 1a. Negligible currents were observed until ∼22 h, when a small increase was observed (indicated with the arrow in the plot). Figure 1b demonstrates that, for 1.2 M LiTFSI, the rapid rise in current is observed at ∼4.0 V, whereas no current is observed for 1.2 M LiPF6 to 5.5 V. For LiTFSI, the current increased rapidly above 4.2 V reaching 1 mA, which indicates rapid dissolution (pitting corrosion) of the Al current collector.3 Smaller currents (
