Dramatic Effects of Low Salt Concentrations on Li-Ion Cells Containing EC-Free Electrolytes

Different concentrations of LiPF 6 (0.3 M–2 M) in ethyl methyl carbonate (EMC) electrolyte and ethylene carbonate (EC)-based electrolyte were studied in LiNi 0.4 Mn 0.4 Co 0.2 O 2 (NMC442)/graphite pouch cells. Fresh cells containing 0.3 M LiPF 6 in EMC electrolyteshowedextremelylargechargetransferresistancewhilethosewith0.3MLiPF 6 inEC/EMCelectrolytedidnot.Impedance spectra taken on symmetric cells and ionic conductivity measurements suggest this difference is due to difﬁculty in dissociating and desolvating Li + ions from the EMC-based electrolyte to intercalate into both the electrodes. After elevated temperature storage experiments at 4.5 V, cells with 0.3 M LiPF 6 in EC/EMC showed a large increase in positive electrode charge transfer impedance, presumably caused by electrolyte oxidation. With salt concentrations greater than 1 M, charge transfer resistance was much smaller in EMC-based electrolytes and was stable during storage for both electrolyte types. Conductivity and cycle testing measurements suggest that 1.5 M LiPF 6 should be used in EC-free EMC-based electrolytes to optimize cell performance. ©

Using high potential positive electrodes can increase the energy density of Li-ion cells. [1][2][3][4] Layered LiNi x Mn y Co z O 2 (x + y + z = 1) (NMC) electrodes have been extensively studied because they are cheaper materials than LiCoO 2 (LCO) and can operate at high working potentials. [5][6][7][8][9][10][11][12][13][14][15] LiNi 0.4 Mn 0.4 Co 0.2 O 2 (NMC442) is a possible choice for high voltage application due to its high working potential, high specific capacity, moderate cost and excellent safety. [16][17][18][19] NMC442 has been shown to be structurally stable up to 4.7 V 16,17 and its specific capacity increases almost linearly as the upper cutoff voltage increases from 4.1 V to 4.7 V. At 4.7 V, NMC442 can deliver a reversible specific capacity of ∼207 mAh/g. NMC442 shows attractive safety features compared to other NMC grade materials as well. 19 It is difficult to create NMC442/graphite cells that show long lifetime when operated continually to an upper cutoff potential above 4.4 V. Electrolyte additives such as pyridine boron fluoride (PBF), 20,21 triallyl phosphate (TAP) 22 and additive blends such as prop-1-ene-1,3-sultone (PES) + 1,3,2-dioxathiolane-2,2-dioxide (DTD) + tris-(trimethyl-silyl) phosphite (TTSPi) [23][24][25][26][27] in EC-based electrolyte have been used to improve charge-discharge cycle life and calendar life of cells operated to 4.4 V. Sulfone-based and fluorinated electrolyte systems have also shown to be of value in getting NMC442/graphite cells to operate beyond 4.4 V. 28,29 In spite of these improvements, further improvement is required. Recently, it was found that EC-free electrolyte using only ethyl methyl carbonate as the solvent can enhance high voltage operation and lifetime of NMC442/graphite pouch cells compared to EC-based electrolyte. [30][31][32] However, only a salt concentration of 1 M LiPF 6 in this novel electrolyte was investigated.
The concentration of LiPF 6 in EC-based electrolyte can greatly affect cell performance. [33][34][35][36] Wang et al. found that different concentrations of LiPF 6 in EC-based electrolyte can affect LiCoO 2 /graphite cell cycle life and calendar life. 33 Petibon et al. observed that high concentrations of LiPF 6 (more than 2 M) in EC-based electrolyte hinder impedance growth at the positive electrode when cells were tested to 4.5 V. 34 In this report, a study of different concentrations of LiPF 6 in EMC and EC-based electrolytes (0.3-2 M) was carried out. The electrolytes were used in NMC442/graphite pouch cells that were subjected to elevated temperature storage, charge-discharge and impedance spectroscopy studies. Pouch cells were taken apart so that symmetric cells could be constructed to learn at which electrode impedance increases originated. In order to achieve a fair comparison between these two different electrolyte systems at different salt concentrations, the same * Electrochemical Society Fellow. z E-mail: jeff.dahn@dal.ca electrolyte additive blend with the same weight percentage was added to all electrolytes used. Ionic conductivity measurements versus salt concentration in EMC electrolyte (0.1-2.5 M) and EC-based electrolyte (0.002-0.2 M) were taken at various temperatures to gain insights on what salt concentrations should be used in EMC electrolytes to optimize cell performance and why the low salt concentrations in EMC electrolytes cause large cell impedance.

Experimental
Electrolytes used in this work.-Components used to make the electrolytes described in this paper are listed in Table I. Electrolytes used in this work are listed in Table II. Table II Table III summarizes detailed information about the electrode materials in these cells. The separator used in these cells was microporous polyethylene with an Al 2 O 3 ceramic coating on the side in contact with the positive electrode.
Before shipping to Canada, these cells were vacuum sealed in a dry room in China without electrolyte. Prior to electrolyte filling, these cells were cut below the seal and vacuum dried at 80 • C for 14 h to remove the trace amount of water. After the pouch cells were filled with 0.76 mL of electrolyte in an argon-filled glove box, they were sealed under vacuum in the same glove box. After electrolyte filling, cells were placed in a 40. ± 0.1 • C temperature controlled box, and held at 1.5 V for 24 h. They were charged to 3.5 V at C/20 and held for 1 h, and then transferred to a glove box for degassing (the first degassing step) (cut open below the seal and re-sealed under vacuum). After degassing, they were charged to 4.5 V and held for 1 h and then degassed again (the second degassing step). These degassing voltages were selected based on the in-situ gas measurements, which indicated that most of the gas was produced during the first cycle at voltages below 3.5 V and 4.3 V. 37 After the two degassing steps, cells were discharged to 3.8 V, held for 5 h and cell impedance was then measured.
Storage experiments.-After cell impedance measurements, the cells with X_E_VC_TTSPi and X_EE_VC_TTSPi were discharged to 2.8 V and charged back to 4.5 V at C/20 and held at 4.5 V for 30 hours. They were then moved to a storage system and stored at 40 or  Long term cycling experiments.-The NMC442/graphite pouch cells with 0.5_E_VC_TTSPi, 1_E_VC_TTSPi, 1.5_E_VC_TTSPi, 0.5_EE_VC_TTSPi, 1_EE_VC_TTSPi and 1.5_EE_VC_TTSPi after 500 h storage at 40 • C were selected for long term cycling at 40 • C between 2.8 and 4.4 V using a constant current of C/2.2. A constant voltage step was added at the top of charge and applied until the current dropped below C/20. A low rate C/10 cycle was also included every 50 cycles to estimate what fraction of the capacity loss was due to impedance growth during the high rate cycling.

Electrochemical impedance spectroscopy measurements of pouch cells, coin full cells and symmetric cells.-NMC442/graphite
pouch cells with 0.3_E_VC_TTSPi, 2_E_VC_TTSPi, 0.3_EE_VC_TTSPi and 2_EE_VC_TTSPi after 500 h storage at 40 • C were first equilibrated at 3.8 V. They were then moved to an argon filled glove box and disassembled there. The harvested positive and negative electrodes from the pouch cells were used to construct NMC442/graphite coin full cells, NMC442/NMC442 and graphite/graphite symmetric cells with 0.3_E, 2_E, 0.3_EE or 2_EE electrolytes. Eight nominally identical NMC442/graphite pouch cells with 1_E_VC_PPF electrolyte underwent the two degassing steps and then were given ten charge-discharge cycles between 2.8 and 4.5 V at 40 • C at a rate of C/10. Then the cells were equilibrated at 3.8 V and opened in the argon-filled glove box to make coin full cells and symmetric cells with 0.2_E, 1_E, 0.2_D, 1_D, 0.2_EE, 1_EE, 0.2_FT and 1_FT electrolytes. The harvested electrodes for symmetric cells and rebuilt coin full cells mentioned above were washed using DMC two or three times to remove the residual electrolyte from the electrodes.
Two "brother" NMC442/graphite pouch cells with 1_E_VC_PPF underwent the two degassing steps and were then equilibrated at 3.8 V. The cells were moved to the glove box and disassembled there to make coin full cells with 0.2_E, 0.2_EE, 0.02_EE and 0.002_EE. The harvested electrodes used for coin full cells with 0.02_EE and 0.002_EE were washed six times using DMC to completely remove any remaining LiPF 6 in the electrodes.
In most cases, more than two symmetric and coin full cells of each type were made to ensure repeatability. The importance of using symmetric cells to determine charge transfer impedance, R ct , (representing the motion of Li + ions and electrons through the solid electrolyte interface) of electrodes was described by Chen et al. 38 by Petibon et al. 39 The symmetric cells and coin full cells were made using one polypropylene blown microfiber separator (available from 3M Co., 0.275 mm thickness). Most EIS measurements were conducted using a Biologic VMP-3 potentiostat, in a two wire configuration with ten points per decade from 100 kHz to 100 mHz or 10 mHz using a 10 mV input signal amplitude. The two wire configuration was used because this VMP3 is connected to a switching system that allows 16 cells to be measured automatically in sequence. The lead resisatnce (about 1 ) is very small compared to the impedances of the coin-type cells measured here. However, to probe the impact of low electrolyte conductivity, some EIS measurements were made using a Biologic SP150 potentiostat in 4-wire mode with ten points per decade from 100 kHz to 5.1 mHz using a 10 mV input signal amplitude. The EIS spectra for all the coin full cells were measured at 3.8 V. All EIS spectra were collected at 10 • C. This temperature was chosen to magnify the difference between the impedances of the cells.

Determination of gas evolution in pouch cells.-Ex-situ gas mea-
surements were used to measure gas evolution during formation and during storage. The measurements were made using Archimedes' principle with cells suspended from a balance while submerged in deionized water. The change in the weight of the cells suspended in fluid, before and after testing is directly related to the volume change by the change in the buoyant force. The change in weight of a cell, ( W), which can be read from the balance, suspended in a fluid of density (ρ) is related to the change in cell volume ( V) by V = -W/(ρ.g). Ex-situ measurements were made by suspending pouch cells from a fine wire "hook" attached under a Shimadzu balance (AUW200D). The pouch cells were immersed in a beaker of de-ionized water that was at 20. ± 1 • C for measurement.
Conductivity measurements.-The conductivity of EMC electrolyte with different concentrations of LiPF 6 was measured using a conductivity meter (METTLER TOLEDO, FE30). Three different standards (84 μS/cm, 1413 μs/cm and 12.88 mS/cm) were used for calibration. A temperature controlled bath (VWR Scientific, Model 1157) was used to adjust the temperature. After the desired temperature was reached, conductivity readings were recorded at least twice to ensure that the readings were stable. Figure 1 shows the dQ/dV vs. V curves of NMC442/graphite cells with X_E_VC_TTSPi or X_EE_VC_TTSPi electrolytes during the formation cycle. Sinha et al. found that there were clear interactions between VC and TTSPi. 40 Therefore, it was difficult to assign reduction peaks for VC and TTSPi, respectively. Figure 1 shows that the reduction peak of the additives shifts slightly to higher voltage as the concentration of LiPF 6 increases in both electrolytes, which corresponds to a lower potential vs Li + /Li.

Results and Discussion
It is of importance to minimize gas generation in pouch cells. Figure 2 shows the volume of gas produced in the cells of Figure  1 during the two steps of formation (a, b) and during the 500 h storage periods at 4.5V at both 40 and 60 • C (c, d). Figures 1a and  1b show that the concentration of LiPF 6 does not strongly affect gas generation during formation. Figures 1c and 1d show that the volume of gas generated during 500 h storage increased as the concentration of LiPF 6 increased at both temperatures, for cells containing both X_E_VC_TTSPi and X_EE_VC_TTSPi electrolytes. Much more gas was observed after 500 h of storage at 60 • C compared to 40 • C. This is due to more severe electrolyte oxidation at higher temperature. 41 Figure S1 shows the open circuit voltage versus time of NMC442/graphite cells with X_E_VC_TTSPi and X_EE_VC_TTSPi electrolytes during the 500 h storage periods at 40 • C and 60 • C. Figure  3 summarizes the voltage drop during storage. The voltage drop during open circuit storage mainly originates from electrolyte oxidation at the positive electrode which causes lithium ions to intercalate into the positive electrode to combine with electrons from the electrolyte oxidation. 41 Figure 3 shows that higher concentrations of LiPF 6   a slightly greater voltage drop suggesting that higher concentrations of LiPF 6 cause slightly larger parasitic reaction rates at the positive electrode. Figure 4 shows the area-specific Nyquist plots of NMC442/ graphite pouch cells with X_E_VC_TTSPi and X_EE_VC_TTSPi electrolytes collected after formation and after the 500 h storage experiments described by Figures S1 and 3. The impedance spectra were measured at 3.8 V at 10 • C. The width of the overlapping semicircles represents the sum of R ct of the positive and negative electrodes. The first semicircle at higher frequency originates from the negative electrode while the second semicircle at lower frequency is due to the positive electrode. 39,42 Figure 4 shows that R ct of NMC442/graphite pouch cells with X_E_VC_TTSPi or X_EE_VC_TTSPi electrolytes decreases as the concentration of LiPF 6 increases. This is consistent with Wang et al.'s findings for LCO/graphite pouch cells. 33 Figures  4d and 4f show that R ct of NMC442/graphite pouch cells after storage with 0.3_EE_VC_TTSPi is much larger compared to other concentrations of LiPF 6 in X_EE_VC_TTSPi. The impedance spectra in Figures 4d and 4f for cells with 0.3_EE_VC_TTSPi are well separated into two "semicircles" where the second semicircle at the lower frequency, due to the positive electrode, is much larger than the first semicircle, due to the negative electrode, at higher frequency. However, R ct of NMC442/graphite pouch cells with 0.3_E_VC_TTSPi does not separate into two semicircles. This suggests that R ct originating from the positive electrode does not dominate the impedance spectra of cells with 0.3_E_VC_TTSPi electrolyte. The reasons for the differences in the impedance spectra as a function of salt content in these cells will be discussed in more detail in the text pertaining to Figures 7,8,9 and 10. Figure 5 shows discharge capacity vs cycle number for NMC442/graphite pouch cells tested at 40 • C and 55 • C. The cells in Figure 5 contained 0.5_E_VC_TTSPi, 1_E_VC_TTSPi, 1.5_E_VC_TTSPi, 0.5_EE_VC_TTSPI, 1_EE_VC_TTSPi and 1.5_EE_VC_TTSPi electrolytes. These cells were cycled between 2.8 and 4.4 V at C/2.2 using the CCCV cycling protocol described in the Experimental section. At 55 • C, the cells with EMC electrolyte had better capacity retention than those with the same salt concentration in EC/EMC electrolyte. The cells containing 0.5_E_VC_TTSPi or 0.5_EE_VC_TTSPi electrolyte experienced rapid capacity fade at both testing temperatures. Nelson et al. 25 Ma et al. 24 and others [43][44][45] have shown that the major reason for rapid capacity loss in NMC/graphite cells tested at 55 • C and to 4.4 V and above is impedance growth, not Li-inventory loss. Day et al. 46 showed that depletion of LiPF 6 salt occurred during aggressive charge-discharge cycling to 4.5 V in NMC442/graphite cells at 55 • C. If this occurs, it is certainly important to know how both the electrolyte conductivity and the electrode charge transfer impedances depend on salt concentration. Increasing the salt concentration to 1.5 M improved capacity retention in both cases. This is consistent with Petibon et al.'s findings that higher concentrations of LiPF 6 help prevent impedance growth at the positive electrode when cells are cycled to 4.5 V. 34 Ding et al. 47 conductivity increasing linearly with salt content at low concentration and then reaching a maximum and declining as ion pairing and viscosity increase. By contrast, a search of the literature did not reveal any measurements for the conductivity of LiPF 6 /EMC solutions versus molarity and temperature. Therefore, measurements were made over a wide range of LiPF 6 (0.1-2.5 M) concentrations over a wide range of temperature. Figure 6 shows the conductivity of X_E electrolytes versus molarity and temperature. Figure 6 shows that the conductivity of X_E electrolyte drops dramatically when the concentration of LiPF 6 decreases below 0.4 M, completely unlike the results for LiPF 6 :PC:DEC electrolytes described by Ding et al. 48 The conductivity versus molarity results for X_E electrolytes are very similar to the results described by Doucey et al. 49 and Delsignore et al. 50   electrolyte at low salt concentration. Figure 6 also shows that 1.5_E electrolyte has optimum conductivity over a wide temperature range, suggesting 1.5_E should be used in cells with EC-free electrolytes based on EMC. Figure 7 shows the area-specific Nyquist plots of the reassembled negative/negative and positive/positive symmetric cells and coin full cells with 0.3_E (a, e, i) and 2_E (b, f, j) electrolytes where the electrodes were taken from the two "brother" pouch cells with 0.3_E_VC_TTSPi electrolytes. These two "brother" cells were stored at 40 • C for 500 h and their impedance spectra, shown in Figure 4c, have been plotted in Figure 7a for comparison The symmetric cells and coin full cells with 0.3_E electrolyte had larger impedance than the corresponding ones with 2_E. Figure 7 shows the impedance spectra of the reassembled negative/negative and positive/positive symmetric cells and coin full cells with 0.3_E (c, g, k) and 2_E (d, h, l) electrolytes where the electrodes were taken from brother pouch cells with 2_EE_VC_TTSPi electrolyte. These two "brother" cells were stored at 40 • C for 500 h and their impedance spectra, shown in Figure 4c greatly affect R ct in symmetric cells even when the electrodes used for the symmetric cells have very different histories. Figure 8 shows the area-specific Nyquist plot of reassembled negative/negative symmetric cells, positive/positive symmetric cells and coin full cells with 0.3_EE and 2_EE electrolytes where the electrodes were taken from the "brother" pouch cells with 0.3_EE_VC_TTSPi electrolyte. These two "brother" pouch cells were stored at 40 • C for 500 h and their impedance spectra, shown in Figure 4d, have been plotted in Figure 8a for comparison. Figure 8 shows that the positive electrode contributes most to the full cell impedance. Corresponding symmetric cells and coin full cells with 0.3_EE had roughly the same impedance as those with 2_EE ( Compare Figures 8a and 8b; compare Figures 8e and 8f; compare Figures 8i and 8j). Figure 8 also shows the area-specific Nyquist plot of negative/negative symmetric cells, positive/positive symmetric cells and coin full cells with 0.3_EE and 2_EE electrolytes where the electrodes were taken from brother pouch cells with 2_EE_VC_TTSPi electrolyte. These two "brother" pouch cells were stored at 40 • C for 500 h and their impedance spectra, shown in Figure 4d, have been plotted in Figure 8d for comparison. The symmetric cells and coin full cells with 0.3_EE had roughly the same impedance as the corresponding ones with 2_EE ( Compare  Figures 8c and 8d; compare Figures 8g and 8h; compare Figures 8k  and 8l). Figures 8a, 8b, 8i and 8j show that the large R ct in the full cells with 0.3_EE_VC_TTSPi after storage is caused by the large positive electrode impedance and that this is not affected by the salt concentration used in the symmetric cells during the impedance measurement. Similarly, Figures 8c, 8d, 8k and 8l show that the modest R ct in the full cells with 2_EE_VC_TTSPi after storage is mirrored by modest positive electrode impedance and that this is not affected by the salt concentration used in the symmetric cells during the impedance measurement. Figure 7 is important because it shows that charge transfer resistance at either electrode is always large when measured in pouch cells, coin full cells or symmetric cells when the cells contain 0.3_E or 0.3_E_VC_TTSPi electrolytes, no matter the prior history of the pouch cells prior to symmetric cell construction. Figure 7 also shows that that charge transfer resistance at either electrode is always small when measured in pouch cells, coin full cells or symmetric cells when the cells contain 2_E or 2_E_VC_TTSPi electrolytes, no matter the prior history of the pouch cells prior to symmetric cell construction. Figure 8 shows that the charge transfer resistance is only high for the positive electrodes taken after storage from pouch cells containing 0.3_EE_VC_TTSPi while it is not high if the positive electrodes were extracted after storage from a pouch cells containing 2_EE_VC_TTSPi. Therefore, in EC-free, EMC based electrodes, the large charge transfer resistance appears to be caused by the presence of low salt concentration during measurement, which is a surprising result while in EC:EMC electrolyte low salt concentration during  300  400  30  60  90  120 150  30  60  90  120 150   0  100  200  300  400  100  200  300  400  30  60  90 120 150 30  60  90 120 Figure 8. Area-specific Nyquist plots of reassembled NMC442/graphite full cells, NMC442/NMC442 and graphite/graphite symmetric cells where the electrodes were taken from pouch cells with 0.3_EE_VC_TTSPi or 2_EE_VC_TTSPi electrolytes equilibrated at 3.8 V. The impedance of the NMC442/NMC442 and graphite/graphite symmetric cells has been divided by two. The pouch cells were stored at 4.5 V for 500 h at 40 • C prior to disassembly. The reassembled cells either contained 0.3_EE or 2_EE electrolyte. elevated temperature storage at 4.5 V leads to high positive electrode charge transfer resistance, no matter how it is measured afterwards, provided EC:EMC electrolytes are used in the symmetric cells. To further confirm these observations, the experiments below were performed. Figure 9 shows the area-specific Nyquist plots of reassembled full cells, positive/positive symmetric cells and negative/negative symmetric cells filled with 0.2_E, 0.2_D, 0.2_EE and 0.2_TF electrolytes. Figure 10 shows the area-specific Nyquist plots of reassembled full cells, positive/positive symmetric cells and negative/negative symmetric cells filled with 1_E, 1_D, 1_EE and 1_TF electrolytes. The electrode used to make the various coin cells in Figures 9 and 10 were harvested from "brother" pouch cells that contained 1_E_VC_PPF electrolyte after ten charge discharge cycles. Figure 9 shows that the charge transfer impedance of both the positive and negative electrodes are much larger when measured in 0.2_E or 0.2_D electrolytes than in 0.2_EE or 0.2_TF electrolyte. The reader should note the large differences in horizontal scale between those of Figures 9a, 9b, 9e, 9f, 9i and 9j compared to those of Figures 9c, 9d, 9g, 9h, 9k and 9l. By contrast, Figure 10 shows that all the charge transfer impedances of symmetric and coin full cells with 1_E, 1_D, 1_EE and 1_TF electrolytes are roughly the same. The electrodes for all the reassembled symmetric cells and coin full cells in Figures 9 and 10 were taken from "brother" cells and experienced the same rinsing procedure, so the SEI at these electrodes were roughly the same before coin cell construction. This further verifies that low LiPF 6 concentration in EC-free linear alkyl carbonate electrolytes, based on EMC or DEC, causes large charge transfer impedance.
In order to convince readers of the robustness of these results, it is important to use Bode plots to demonstrate that the impedances of the reassembled coin full cells are equal to the sum of the impedances of the negative/negative symmetric cells divided by two and the impedance of positive/positive symmetric cells divided by two. 39 Figures S2, S3, S4, and S5 are the corresponding Bode plots to the Nyquist plots in Figures 7, 8, 9 and 10. Figures S3, S4 and S5 show good agreement between the impedances of the reassembled full cells and the sum of the impedances of the symmetric cells while Figure  S2 does not. We do not know why this is the case, so the reader is cautioned that the results in Figure 7 are less robust than those in the other Figures.
Why is the charge transfer resistance for electrodes so large in low molarity solutions of LiPF 6 in EMC and DEC? As mentioned before, LiPF 6 :EMC electrolyte shows an unconventional conductivity vs concentration relation below 0.4 M as shown in Figure 6. This is caused by the presence of far fewer ions that can contribute to conductivity (free ions and/or triple ions) than in a high dielectric solvent blend like EC:EMC at the same concentration. In order to test if the lack of charged ions causes the increase in R ct , coin full cells with 0.2_E, 0.2_EE, 0.02_EE and 0.002_EE electrolytes were made. These electrodes were taken from the same "brother" pouch cells at 3.8 V. Figure 11a and Figure 11b shows the area-specific Nyquist plots of the reassembled coin full cells obtained using four wire EIS measurements at 10 • C, while Figure 11c shows the conductivity of the same electrolytes also measured at 10 • C. Since four-wire measurements were used, the high frequency intercept is meaningful and represents the ionic resistance due to the electrolyte. Figure 11 shows that the diameter of the impedance "semicircle", R ct , increases as the LiPF 6 concentration in EC/EMC electrolyte decreases from 0.2 M to 0.002 M. R ct for 0.2_E is similar to that of R ct for 0.002_EE. The conductivity of 0.2_E is also similar to that 0.002_EE which suggests a similar concentration of charged ions that can contribute to conductivity. At these extremely low concentrations of charged ions, it is clear from Figure 11 that R ct , is dramatically impacted in both E (0.2 M) and EE (0.002 M) electrolytes. This suggests that the lack of charged ions (free ions and/or triple ions) in the LiPF 6 :EMC electrolyte at low salt concentration causes the large R ct in Li-ion cells where such electrolytes are used. Figure 12 shows the electrolyte resistivity (the reciprocal of the conductivity plotted in Figure 11c) plotted versus the high frequency intercept of the impedance spectra of Figures 11a  and 11b. Figure 12 shows a good correlation, as expected if the high frequency intercept is primarily due to the electrolyte resistance in the cell. 52 Xu et al. reported that the extraction of the lithium ion from its solvation sheath is the rate determining step for lithium intercalation into graphite. 53 As the salt concentration in the LiPF 6 :EMC electrolyte decreases, the solvation number increases and the desolvation energy also increases. But the question is: "do they increase enough to make a meaningful impact on R ct , as compared to R ct at high salt concentration?". Figure 4 shows that when the salt concentration decreases from 2M to 0.5M in the LiPF 6 :EMC electrolyte, R ct of the cells does not change significantly. This suggests that changes in desolvation energy due to changes in salt concentration in the 0.5 to 2.0 M range in LiPF 6 :EMC electrolyte does not contribute significantly to the change in R ct of the cells. However, In dilute LiPF 6 solutions in EMC or DMC, by analogy to LiAsF 6 in DMC, as reported by Doucey et al. 49 the salt is not well-dissociated at low concentration, which is why the conductivity is so small. Doucey   constant solvents. Therefore, the large charge transfer impedance at the electrodes of NMC442/graphite cells which use 0.2_E (Figure 9), 0.2_D (Figure 9) or 0.3_E (Figure 7) electrolyte probably stems from the lack of charged ions (free ions and/or triple ions) and also from the difficulty of extracting a Li + cation from the ion pairs that predominate at low concentration. When the salt concentration is large enough, i.e. 1 M as in Figure 10, then sufficient dissociated Li + cations exist and the charge transfer impedance in EMC-based electrolytes becomes comparable to that in EC/EMC-based electrolytes.

Conclusions
The effects of LiPF 6 concentration in EMC and EMC/EC electrolytes were studied in NMC442/graphite pouch cells. Low concentrations of LiPF 6 in EMC electrolyte lead to large charge transfer resistance at both electrodes of NMC442/graphite cells. This is because the LiPF 6 salt is poorly dissociated at low concentration, which also leads to extremely low electrolyte conductivity. At higher concentrations, e.g. 1 M, the charge transfer resistance at electrodes in EMCbased electrolytes becomes similar to that of electrodes in EC/EMC electrodes.
There are significant consequences of these findings for the possible application of EC-free electrolytes based on EMC in Li-ion cells designed for high voltage operation. Any time a Li-ion cell is operated at high rate, concentration polarization in the electrolyte develops. Figure 6 shows that the electrolyte conductivity drops precipitously near 0.4 M and this would be a concern if large concentration polarization developed. In addition, at the same time, the charge transfer impedance at the electrode surfaces also rises dramatically as the salt concentration decreases which would also limit the rate capability of cells. Based on these findings, it is important to select an initial electrolyte concentration that minimizes the possibility that low concentration would develop at some portions of the electrodes when high currents are demanded from the cell. Based on Figure 6, 1.5 M rather than 1 M LiPF 6 in EMC should be used. However, based on the voltage drop during storage ( Figure 3) and gas evolution during storage (Figure 2), there may be applications where it is desirable to limit LiPF 6 concentration to 1 M. Tradeoffs need to be considered.