Abstract
This work aims to unveil the mystery about the suitability of using lithium bis(fluorosulfonyl)imide (LiFSI) as a single salt in low-to-moderate concentration electrolytes of Li metal and Li-ion batteries by studying the corrosion behavior of Al and stainless steel (SS) in 3:7 (wt) ethylene carbonate/ethyl methyl carbonate electrolytes. It is found that regardless of LiFSI concentrations (0.8–1.8 m), there is a ∼4.3 V vs Li/Li+ of "threshold potential," above which Al suffers uncontrollable corrosion. This finding is confirmed in graphite/LiNi0.80Co0.15Al0.05O2 pouch cells, however, none of Li/LiNi0.80Co0.10Mn0.10O2 coin cells can be charged to 4.2 V in the first charge. Post-mortem analysis of the failed coin cells reveals that the failure is due to anodic corrosion of the SS spacer. Further study on two types of spacers, made by SS304 and SS316L, indicates that the SS316L spacer is more corrosion resistant, however, it still cannot support batteries for long-life cycling. The suitability for using LiFSI as a single electrolyte solute in Li-ion batteries can be guaranteed as long as, at the cathode, the charge voltage is strictly limited to the "threshold potential" and the SS components are avoided. This conclusion is further verified by using four different sources of LiFSI.
Export citation and abstract BibTeX RIS
Lithium bis(fluorosulfonyl)imide (LiFSI) has long been considered a magic and controversial salt in the community of Li metal and Li-ion batteries. The magic lies in its super ability in providing high ionic conductivity for electrolytes and forming robust and highly conductive interphases with both the anode and cathode materials, 1–4 whereas the controversy is centralized on the anodic corrosion of Al current collector at high potentials. 2,5–7 On Al corrosion, Han et al. 2 reported that the Cl‒ anions, an impurity in LiFSI salt, are responsible for promoting Al3+ ions to dissolve into the solution, and suggested that using a LiFSI with high purity can mitigate Al corrosion. However, other studies indicated that chlorine-containing solvent 8 and Li salt 9 do not trigger Al corrosion except for the possibility that the chlorine is oxidized at high potentials. Research also showed that different sources of LiFSI do not lead to visible differences in the cycling performance of Li/LiFePO4 cells. 10 Owing to the concern with Al corrosion, LiFSI is at present mainly employed as an additive 11–13 or a co-salt 14–17 to improve the performances of Li and Li-ion batteries. In efforts to use LiFSI as a single electrolyte solute in the Li metal and Li-ion batteries, several strategies have shown success in suppressing Al corrosion, for example, high concentration electrolytes (HCEs), 18–23 fluorinated solvents 24,25 and particularly using fluorinated co-solvents to form localized high concentration electrolytes (LHCEs), 26–28 and ionic liquids (ILs). 29–31 From the viewpoint of practical applications, however, all these strategies are not acceptable because on one hand the high concentration and fluorinated solvents increase the cost and on the other hand the fluorinated solvents introduce detrimental impacts on the environment. Furthermore, the HCEs and ILs do not wet low-polar polyolefin separators, special separator must be developed so as to use these electrolytes.
At present, Al corrosion remains a great challenge for using low-to-moderate concentration (<2.0 M) LiFSI electrolytes in high-voltage Li metal and Li-ion batteries. In the previous literature, inconsistent or even opposite conclusions were reported about the corrosion of Al in low-to-moderate concentration LiFSI electrolytes, 32–38 and the corrosion of stainless steel (SS) spacer increased additional complexity of the coin cells. 30,34 In our laboratory, entirely different results were also observed from pouch cells and coin cells. That is, Li-ion pouch cells operated well in the normal voltage range (2.8–4.2 V) even when LiFSI was used as a single salt in a 1.0 m (molality) electrolyte, while Li-ion coin cells failed to be charged to 4.0 V even if only 10 mol% LiFSI was used as a co-salt. To unveil the mystery of low-to-moderate LiFSI concentration electrolytes in Li metal and Li-ion batteries, in this work, the effect of LiFSI concentration on the anodic corrosion of Al current collector and stainless steel (SS) spacer is studied by using a 3:7 (wt) solvent system of ethylene carbonate and ethyl methyl carbonate. The work leads to two important findings: (1) regardless of LiFSI concentration, there is a ∼4.3 V vs Li/Li+ of "threshold potential," above which Al suffers uncontrollable corrosion, and (2) the failure of coin cells is simply due to the corrosion of SS spacer. The above findings suggest it is suitable for using LiFSI as a single salt in low-to-moderate concentration electrolytes of the Li metal and Li-ion batteries as long as, at the cathode, the charge voltage is strictly limited to the "threshold potential" and the SS components are avoided. This suitability is further verified by using four sources of LiFSI.
Experimental
LiFSI salts were acquired from four manufacturers, coded as A, B, C and D, and used as received. Electrolytes were prepared in an Ar-filled glove box by using a 3:7 (wt) mixture of ethylene carbonate and ethyl methyl carbonate, and referred to as EL08, EL10, EL12, EL15, and EL18 for 0.80 m, 1.0 m, 1.2 m, 1.5 m, and 1.8 m LiFSI, respectively. A dip-type three-electrode cell consisting of an Al wire (1.0 mm in diameter) as the working electrode and Li as the counter and reference electrodes was used to study the corrosion behavior of Al. Machine-made graphite/LiNi0.8Co0.15Al0.05O2 (Gr/NCA, 200 mAh) pouch cells without electrolyte were supplied by Li-Fun Technology (Zhuzhou City, China). Before filling the electrolyte, the cells were dried at 80 oC under vacuum for 16 h by cutting the side of degassing. In an Ar-filled glove box, the cell was filled with 0.90 mL (∼0.95 g) electrolyte, followed by vacuum sealing. Two types of coin cell hardware with a spacer made of SS304 and SS316L, respectively, were purchased from MTI Corporation (Richmond, CA). Machine-coated graphite (Gr) anode (6.07 mg cm−2) and LiNi0.8Co0.1Mn0.1O2 (NCM811) cathode (9.08 mg cm−2) were received from Argonne National Laboratory. The electrodes were punched into the small disks with a 1/2 inch diameter for the Gr anode and a 7/16 inch diameter for the NCM811 cathode, and dried at 110 oC under vacuum overnight. Using a Celgard 2350 membrane as the separator, Li/NCM811 half cell and Gr/NCM811 Li-ion cell were assembled and filled with 40 μL electrolyte.
Cyclic voltammetry of the dip-type cell was performed at a slow rate of 0.1 mV s−1 on a Solartron SI 1287 Electrochemical Interface. Cycling testing of the pouch cells and coin cells was performed on a Maccor Series 4000 cycler. For pouch cells, an additional process was applied to allow electrolyte wetting by charging the cell at 0.1 C to 1.5 V and holding it at 1.5 V for 24 h. Both Gr/NCA pouch cells and Gr/NCM811 coin cells were cycled between 2.8 and 4.2 V at 0.1 C for 2 cycles, followed by cycling at 0.5 C for the rest of the testing. For Li/NCM811 coin cells, the experiment was set to charge the cell at 0.1 C (1 C = 1.68 mA cm−2) to 4.2 V and then hold it at 4.2 V for a total of 100 h. The microscope image of SS spacers was acquired by a Nikon Digital Camera DXM1200F.
Results and Discussion
Corrosion behavior of Al
Using a dip-type three-electrode cell, the corrosion behavior of Al in various LiFSI concentration electrolytes is studied by cyclic voltammetry at a slow rate of 0.1 mV s−1. The parts of interest for the initial three cycles are shown in Fig. 1, from which several important findings are observed: First, regardless of LiFSI concentration, the native aluminum oxide passivation layer formed on the Al surface is able to protect Al from anodic corrosion up to 4.3 V vs Li/Li+, above which Al suffers uncontrollable corrosion, as indicated by the anodic current of the first cycle in Figs. 1a–1e. It is observed that in the first cycle, the anodic current continues to increase even when the potential is swept back from the maximum (5.5 V). This behavior can occur only when the corrosion is out of control because in this case, the current is proportional to the surface area of unprotected Al, and the latter increases with time of the breakdown of the native aluminum oxide passivation layer in the course of uncontrollable corrosion. Similar corrosion patterns were also reported elsewhere, 2,5,7,39 and such corrosions are found to be induced by the oxidation of solvents, 40 which is affected not only by the nature of the solvents but also by the salt concentration and solvent composition in the solutions. In other words, the onset potential of such corrosions is affected by the oxidation of electrolyte solvents, instead of the nature of Al. To distinguish it from others, such an onset potential is referred to as "threshold potential." Thus, many previous controversies about the corrosion of Al in LiFSI electrolytes can be reasonably explained by the influence of electrolyte solvents on the "threshold potential." This also reasonably explains the success of the HCE strategy 18–23 since high salt concentration reduces the molar fraction of free solvents in the solution, which as a result enhances the stability of solvents against oxidation and reduction at extreme potentials. Second, the onset potential of the 2nd and 3rd cycles in all LiFSI concentrations is consistently moved down by 0.2–0.3 V, as compared with that of the 1st cycle. This is because the Li reference electrode reacts with the Al3+ ions, dissolved from the working electrode, to form Li-Al alloy, which consequently raises the standard electrode potential of the Li reference electrode. Third, at the end of the experiment (6 cycles), the color of the surface of Al working electrode and Li reference electrode is changed to black and gray, respectively, and most significantly, numerous grey flocs accumulate around the Li counter electrode (see the picture in each figure, where the Li counter electrode is marked by a symbol "c"). In the above, the black Al surface means that Al corrosion is accompanied by solvent oxidation, and the grey Li reference electrode indicates the formation of Li-Al alloy on the Li surface, whereas the grey flocs, whose amount dramatically increases with a decrease in the LiFSI concentration, are a mixture of Li metal, Li-Al alloy, and solvent decomposition products. On the other hand, Fig. 1f indicates that minor corrosion occasionally occurs near 3.5 V vs Li/Li+ for those with low concentrations from 0.8 m to 1.2 m LiFSI. However, these minor corrosions are self-healed, as indicated by the small peak of anodic current (see the marked dot rectangle in Fig. 1f). From the above results, it can be concluded that LiFSI is suitable for use as a single salt in low-to-moderate concentration electrolytes of Li metal and Li-ion batteries as long as the charge voltage is strictly limited to the "threshold potential" at the cathode.
Figure 1. Part of the interest of the initial three cyclic voltammograms of Al-Li-Li three-electrode cells, in which the inset displays the cell's photo after six consecutive cycles at 0.1 mV s−1 between 2.0 V and 5.5 V vs Li/Li+. (a) EL08, (b) EL10, (c) EL12, (d) EL15, (e) EL18, and (f) comparison of the anodic currents in the first potential sweep.
Download figure:
Standard image High-resolution imageSuccess in Li-ion pouch cells
Following the above findings, LiSFI electrolytes are examined in 200 mAh Gr/NCA pouch cells. Figure 2 compares the voltage profile of the first cycle and discharge capacity of the pouch cells using different LiFSI concentration electrolytes. In particular, the capacity and coluombic efficiency of the initial two cycles are summarized in Table I. It is observed in Fig. 2 and Table I that differences in the cycling performance caused by LiFSI concentration are in an error range. In the 2nd cycle, all cells show a small increase in capacity (see Fig. 2b and Table I). Particularly, the cell with 0.8 m LiFSI electrolyte suffers a small loss in capacity in the 22nd cycle. This is attributed to the minor Al corrosion and it is self-healed in the subsequent cycle, as suggested in Fig. 1f. Compromising the cost and cell performance, a 1.2–1.5 m concentration range appears to be optimal for the LiFSI electrolytes used in Li metal and Li-ion batteries. Additionally, small and regular fluctuations in capacity, as observed in Fig. 2b, are found due to the day-night temperature change in the testing room.
Figure 2. Cycling performance of Gr/NCA pouch cells. (a) Voltage profile of the first cycle at 0.1 C, and (b) discharge capacity of the cells cycling at 0.5 C except for the initial two cycles at 0.1 C.
Download figure:
Standard image High-resolution imageTable I. Effect of LiFSI concentration on specific capacity and coulombic efficiency of Gr/NCA pouch cells.
| Electrolyte | LiFSI, m | 1st cycle | 2nd cycle | ||
|---|---|---|---|---|---|
| Cap., mAh g−1 | CE, % | Cap., mAh g−1 | CE, % | ||
| EL08 | 0.8 | 196.8 | 84.4 | 207.1 | 100.6 |
| EL10 | 1.0 | 194.2 | 84.0 | 203.9 | 99.9 |
| EL12 | 1.2 | 198.5 | 84.8 | 206.6 | 99.8 |
| EL15 | 1.5 | 196.0 | 84.1 | 203.5 | 99.9 |
| EL18 | 1.8 | 193.2 | 84.2 | 203.4 | 99.9 |
Failure in coin cells
The same electrolytes are tested in Li/NCM811 coin cells, unfortunately, none of the cells can be charged to the cutoff voltage (4.2 V). Figure 3 displays the voltage curves of the first charge of Li/NCM811 coin cells, which were recorded by setting to charge the cell at 0.1 C (1 C = 1.68 mA cm−2) to 4.2 V and then hold it at 4.2 V for a total of 100 h. It is shown that before reaching 4.2 V, the voltage of all cells started to decline via a two-plateau pattern. Post-mortem analyses on the tested cells reveal that all failures are due to the corrosion of SS spacer (made by SS304), and that the pattern of the cell's voltage decline corresponds to such two processes as: (1) redox shuttle of the transition ions, such as Fe2+/3+, Ni2+/3+, and Cr2+/3+, dissolved by the anodic corrosion of SS spacer, for the first higher voltage plateau, and (2) localized electric shorting, caused by the penetration of the deposited metal dendrites, for the second lower voltage plateau. Figures 4a and 4b show the pictures of cell components and the corrosion images of SS spacers, respectively. It is very consistently observed that the outer edges of the Li anode are entirely covered by many black deposits, and the SS spacer is severely corroded. In particular, the Al current collector becomes very frangible and immediately generates bubbles in contact with water, which provides excellent support for the localized electric shorting. It is the localized electric shorting that promotes Al current collector to react with the Li anode via a galvanic cell without needing physical contact between the Al current collector and the Li anode, which forms Li-Al alloy and makes the Al foil very frangible.
Figure 3. Voltage curve of the first charge of Li/NCM811 coin cells using an SS304 spacer, recorded by setting to charge the cell at 0.1 C to 4.2 V and then hold it at 4.2 V for a total of 100 h.
Download figure:
Standard image High-resolution image
Figure 4. (a) Photos of coin cell components, as harvested after the Li/NCM811 cells were charged at 0.1 C for 100 h, and (b) a representative spacer showing a ring-shaped corrosion pattern with the same size and shape of the cathode and microscope images of the corrosion randomly selected from the different concentration electrolytes.
Download figure:
Standard image High-resolution imageInterestingly, a ring-shaped corrosion pattern is observed from all SS spacers in Fig. 4a. Representatively, a spacer was cleaned using water and displayed in Fig. 4b, from which the size and shape of the corrosion ring are found to exactly match those of the cathode. This is because on one hand the ionic transport path is blocked by the solid Al current collector, which protects SS spacer beneath the cathode from anodic corrosion, and on the other hand the excess liquid electrolyte is preferentially accumulated along the outer jointing edge of the smaller cathode, which makes these regions less polarized and consequently easier to corrode than the others.
SS304 vs SS316L spacer
There are two types of coin cell spacers, made by SS304 and SS316L, available from the suppliers. The chemical composition and corrosion resistance of SS304 and SS316L are compared in Table II, 41 and the performances of Gr/NCM811 coin cells with a 1.2 m LiFSI electrolyte are presented in Fig. 5. As indicated in Fig. 5a, in the first charge, the cell with SS304 spacer fails to reach 4.2 V, while the other using SS316L spacer completes the first charge and discharge between 2.8 V and 4.2 V, revealing that SS316L spacer is superior in corrosion resistance to SS304 spacer. Therefore, the SS316 cells are selected for further cycling test by using a cutoff voltage of 4.1 V and 4.2 V, respectively, and the results, with two cells for each, are exhibited in Fig. 5b. It is shown that both the specific capacity and coulombic efficiency of the cells with the 4.2 V cutoff voltage are significantly lower than those cut off at 4.1 V. Even if cutting off at 4.1 V, the capacity of the cells still fades rapidly with the cycle number. The above results indicate that LiFSI electrolytes are electrochemically incompatible with the stainless steel materials, which is probably part of the reasons for the inconsistent results, reported previously by different researchers who use coin cells for testing.
Table II. Chemical composition and corrosion resistance of SS304 and SS316L. 41
| Composition, wt% | SS304 | SS316L |
|---|---|---|
| C | ≤0.08 | ≤0.08 |
| Si | ≤1.00 | ≤1.00 |
| Mn | ≤2.00 | ≤2.00 |
| P | ≤0.045 | ≤0.045 |
| S | ≤0.030 | ≤0.030 |
| Cr | 18.0 ∼ 20.0 | 16.0 ∼ 18.0 |
| Ni | 8.0 ∼ 11.0 | 10.0 ∼ 14.0 |
| Mo | — | 2.0 ∼ 3.0 |
| Corrosion resistance | good | better than SS304 |
Figure 5. Performance difference of Gr/NCM811 coin cells caused by the SS material of spacer. (a) Voltage curve of the first cycle at 0.1 C, and (b) discharge capacity (hollow symbols) and coulombic efficiency (solid symbols) with two cells for each cutoff voltage.
Download figure:
Standard image High-resolution imageFigure 6 shows the picture and corrosion image of a tested Gr/NCM811 cell using an SS304 spacer. Again, a ring-shaped corrosion pattern is observed from the spacer, and the outer edges of both the separator at the side facing Gr anode and Gr anode are entirely covered by the black deposits that have been reported to be the decomposition products of the electrolyte solvents. 34 A large amount of black decomposition products suggests that the decomposition of the solvents must have been catalyzed by either the solvation with the transition metal ions or the reduction of the transition metal ions (Fe, Ni, and Cr), dissolved from the SS spacer.
Figure 6. Corrosion behavior of SS304 spacer in Gr/NCM811 coin cell after the cell was charged at 0.1 C for 100 h. (a) SS spacer, (b) separator facing the side of Gr anode, and (c) Gr anode.
Download figure:
Standard image High-resolution imageDifferent sources of LiFSI
Based on the above analyses, it is believed that using LiFSI as a single salt in a low-to-moderate concentration electrolyte of Li metal and Li-ion batteries is suitable as long as, at the cathode, the charge voltage is strictly limited to the "threshold potential" (4.2–4.3 V vs Li/Li+ in the present work) and the SS components are avoided. To verify this hypothesis, four sources of LiFSI are used to prepare a 1.2 m electrolyte, coded as EL12 [A], EL12 [B], EL12 [C], EL12 [D], respectively, and examined in Gr/NCA pouch cells. The cycling performances of these cells are compared in Fig. 7 and the data of the initial two cycles at 0.1 C are summarized in Table III. Similar to those observed in Fig. 2 and Table I, the differences related to the sources of LiFSI are in an error range. Interestingly, the small increase in capacity from the 1st to 2nd cycle is again observed. Since such a small capacity gain is consistently observed from all cells, it can be considered to be a common phenomenon of the pouch cells and explained by the elimination of mechanical stress between the electrodes and separator by the first charge and discharge cycling, which consequently homogenizes electrolyte distribution and reduces the internal resistance of the cells.
Figure 7. Cycling performance of Gr/NCA pouch cells with a 1.2 m LiFSI electrolyte made by different sources of LiFSI. (a) Voltage curve of the first cycle at 0.1 C, and (b) discharge capacity.
Download figure:
Standard image High-resolution image
Table III. Specific capacity and coulombic efficiency of Gr/NCA pouch cells by different sources of LiFSI.
| 1st cycle | 2nd cycle | ||||
|---|---|---|---|---|---|
| Code | LiFSI, m | Cap., mAh g−1 | CE, % | Cap., mAh g−1 | CE, % |
| EL12 [A] | 1.2 | 199.7 | 85.2 | 204.2 | 100.7 |
| EL12 [B] | 1.2 | 198.6 | 85.1 | 207.3 | 100.2 |
| EL12 [C] | 1.2 | 197.3 | 84.6 | 206.7 | 100.6 |
| EL12 [D] | 1.2 | 193.4 | 84.8 | 203.8 | 100.3 |
Conclusions
Anodic corrosion of the Al current collector has long been a concern with the use of LiFSI as a single salt in low-to-moderate (<2.0 M) concentration electrolytes of the high-voltage Li metal and Li-ion batteries. Inconsistent or even opposite results on it are frequently reported by different researchers. To unveil the mystery about LiFSI, in this work, the corrosion behavior of Al is studied in 0.8–1.8 m LiFSI electrolytes by using a 3:7 (wt) solvent system of ethylene carbonate and ethyl methyl carbonate. The work leads to several important findings: (1) regardless of LiFSI concentration, there is a solvent-related "threshold potential" (4.3 V vs Li/Li+ in the present system), above which Al suffers uncontrollable corrosion, (2) minor corrosion that increases with a decrease in the LiFSI concentration occasionally occurs at ∼3.5 V vs Li/Li+, however, it can be self-healed, (3) LiFSI is rather corrosive to SS materials, as such, the coin cells made of SS materials without appropriate protection are not suitable for using LiFSI electrolytes, (4) LiFSI can be used as a single salt in low-to-moderate concentration electrolytes of the Li metal and Li-ion batteries as long as, at the cathode, the charge voltage is strictly limited to the "threshold potential" and the SS materials are avoided, and (5) different sources of LiFSI (and hence the purity) do not lead to a significant difference in the cycling performances of the Li-ion batteries. Compromising the cost and battery performance, a 1.2–1.5 m concentration range would be optimal for the LiFSI electrolytes used in the Li metal and Li-ion batteries. The previous controversy about Al corrosion in LiFSI electrolytes can ascribe to the influence of electrolyte solvents on the "threshold potential," and the corrosion of SS spacer increases additional complexity if the coin cells are used in testing.
Acknowledgments
The author acknowledges the support of DEVCOM Army Research Laboratory, and thanks Drs. Arthur von Cresce, Lin Ma, Marshall Schroeder, and Kang Xu for providing materials and discussion.