Studies of the SEI layers in Li(Ni_0.5Mn_0.3Co_0.2)O₂/Artificial Graphite Cells after Formation and after Cycling

The chemicals used were as follows: ethylene carbonate (EC), dimethyl carbonate (DMC), lithium hexafluorophosphate (LiPF₆), vinylene carbonate (VC), and lithium difluorophosphate (LiPO₂F₂—called LFO here). All electrolytes were prepared in an argon-filled glove box using chemicals from BASF and Shenzhen Capchem (EC:DMC 3:7 (w:w) from BASF (purity > 99.9%, water < 20 ppm), LiPF₆ from Capchem (water < 20 ppm, HF < 5 ppm), VC from Capchem (purity > 99.9% and water < 20 ppm), and LiPO₂F₂ from Capchem). For all cells in these experiments, the control electrolyte mixture used was 1.2 M LiPF₆ in EC:DMC 3:7 (w:w). Other electrolytes used were the control electrolyte mixture with two percent by weight VC (called control + 2% VC) and the control electrolyte mixture with one percent by weight LiPO₂F₂ (called control + 1% LFO).

Single crystal Li(Ni_0.5Mn_0.3Co_0.2)O₂ (NMC 532)/artificial graphite (AG) 402035-size wound pouch cells with a capacity of 210 mAh at 4.2 V were used in these experiments. Dry (no electrolyte), vacuum sealed cells were received from LiFun Technology (Xinma Industry Zone, Golden Dragon Road, Tianyuan District, Zhuzhou City, Hunan Province, PRC, 412000). The positive electrode loading was 21.1 mg cm⁻², the electrode density was 3.5 g cm⁻³, and the active weight fraction was 94%. The negative electrode loading was 11.5 mg cm⁻², the electrode density was 1.55 g cm⁻³, and the active weight fraction was 95.4%. The pouch cells were dried under vacuum in a 100 °C oven for 14 h to remove any moisture before filling with electrolyte. Approximately 0.85 ml (∼1 g) of electrolyte was used to fill each cell. Pouch cells underwent a formation process on a Maccor 4000 series charger, which involved wetting the cells at 1.5 V for 24 h, followed by a C/20 charge to either 4.2 V or 4.4 V, and a hold at top of charge for one hour. Cells were then discharged to ∼50% state of charge (SOC), which was ∼3.8 V for these cells. Pouch cells were either disassembled after formation, to make symmetric cells and coin cells, or were cycled for long term cycling tests. After formation, pouch cells that were to be cycled were degassed and re-sealed in an argon-filled glovebox. These cells were externally clamped to provide uniform pressure to the electrode stack while cycling. Pouch cells were cycled on a Neware battery tester (Shenzhen, China) at 40 °C. Cells were cycled between 3.0 V and either 4.2 V or 4.4 V at a charge/discharge rate of C/3 in constant current, constant voltage (CCCV) mode, with a cut-off current of C/20. One constant current C/20 cycle was done every 50 cycles.

For EIS testing, the pouch cells were disassembled in an argon-filled glovebox and negative electrode symmetric cells, positive electrode symmetric cells, and full cell coin cells were made from the charged electrodes. Two or three cells of the same type were fabricated to ensure repeatability. The same electrolyte that was used in the parent pouch cells was also used in the assembly of the symmetric cells and full coin cells. The electrodes punched for these cells had a surface area of 0.95 cm² and were separated by a polypropylene blown microfiber separator (BMF—from 3 M Co., 0.275 mm thickness, 3.2 mg cm⁻²). The symmetric cell assembly process can be seen in detail in Petibon et al.⁴⁴ The electrodes used in the symmetric cells were two-side coated, not single side coated, as shown in Keefe et al.³⁶ Keefe et al.³⁶ also showed that the additional layer of electrode material (the coating on the back side of the electrode) affects only the high frequency, contact resistance part of the Nyquist plots for these cells, while the R_ct semicircle is unaffected.

A BioLogic VMP3 was used to measure electrochemical impedance spectra of the symmetric cells and full coin cells. The spectra were collected over a frequency range of 100 kHz to 10 mHz with ten points per decade with a perturbation amplitude of 10 mV. Spectra were collected at −10 °C, 0 °C, 10 °C, 20 °C, 30 °C, and 40 °C. Measurements were taken at 10 °C at the beginning, middle, and end of the experiments to ensure repeatability. All Nyquist plots in this work show area specific impedances calculated based on the active electrode area of the coin cells. The negative and positive symmetric cell impedances were divided by two in order to only account for one of the two electrodes. The addition of the positive and negative symmetric cell data by frequency, as in a Bode plot, should give the full coin cell spectrum for cells from the same parent pouch cell.⁴⁴ This is demonstrated in Fig. S1 (available online at stacks.iop.org/JES/167/120507/mmedia).

Figure S2 shows the equivalent circuit models used to fit impedance data in this paper. The high frequency semicircle is interpreted as contact impedance at the electrode active particle/current collector interfaces, while the mid-low frequency semicircle represents charge transfer resistance. More details on the EIS interpretation can be found in Keefe et al.³⁶ Fits were completed using RelaxIS 3—Impedance Data Analysis (rhd instruments GmbH & Co. KG, Germany).

Square samples (1 cm²) were cut from double sided electrodes in an argon-filled glovebox and rinsed three times in DMC solvent to remove residual EC and LiPF₆ salt. Once dry, samples were mounted on sample holders using double-sided copper tape (3 M). To ensure that samples never came in contact with air, samples were transferred to the XPS apparatus using a custom-made transfer suitcase built at Dalhousie University.⁹ The analysis chamber was under ultra-high vacuum (UHV) (< 2 × 10⁻⁹ mbar) for data acquisition. XPS was done using a SPECS spectrometer with a PHOIBOS 150 hemispherical energy analyzer. Samples were irradiated with a Mg Kα source (hυ = 1253.5 eV). Data analysis was performed using CasaXPS software (version 2.3.12). For all spectra, calibration of BE was done using the 284.8 eV C-C peak. A Shirley-type background was fit around the regions of interest and the background was subtracted from the data. Peaks were fit in CasaXPS and assigned to different SEI and electrode material components based on BE. See Ref. 9 for more details about the experimental procedure.

Table III shows the matrix of cells in this comparison. All cells were formed or cycled to 4.2 V and disassembled at 3.8 V. The number of cycles and the capacity retention are indicated in the table. Cycled cells are labeled by approximate cycle number (∼400c or ∼700c) in future figures. Nyquist plots for negative and positive symmetric cells and full coin cells at −10 °C, 10 °C, 30 °C and 40 °C are shown in Figs. S8—S22.

Figure 3 shows 10 °C Nyquist plots for (−/−) and (+/+) cells for all cells in this comparison. Figures 3a–3c show good agreement in R_contact across all (−/−) cells. Figures 3d–3f show minor increases in R_contact for (+/+) cells after cycling. R_contact will be compared more closely at all temperatures in Fig. 5.

Zoom In Zoom Out Reset image size

Figure 3a shows that R_ct decreases after cycling for (−/−) cells with control electrolyte, while Fig. 3d shows that (+/+) cell R_ct increases significantly for control cells after cycling. The addition of the impedance from the positive electrode and the negative electrode in a control cell gives an overall increase in impedance after cycling. This increase in impedance after cycling is consistent with the ΔV results for the control cell as seen in Fig. 2c.

Figure 3b shows that R_ct for the negative electrode changes only slightly over cycle life for cells with 2% VC additive, and Fig. 3c shows virtually no change in R_ct for the negative electrode in cells with 1% LiPO₂F₂. Negative electrode R_ct for 1% LiPO₂F₂ cells is significantly smaller than R_ct for the 2% VC and control cells in agreement with Ma et al.¹² and Liu et al.⁴⁷ Figs. 3e and 3f show a reduction in positive electrode R_ct for both 2% VC and 1% LiPO₂F₂ cells with cycle number. This suggests good SEI formation on the positive electrode in cells containing these additives. This differs from the trend for cells containing control electrolyte, in which positive electrode R_ct increases significantly with cycling. Figure 10 summarizes R_ct changes with cycling.

Figure 4 shows the logarithm of R_ct vs 1/T for all cells in this comparison. The relationship is linear for all cells, indicating that R_ct follows the Arrhenius relationship after formation and also after cycling. E_a values and errors are shown for each cell. Linear regression was used to fit the data and errors were calculated based on the standard deviation of the slope found by the fit. E_a values are stated with standard deviation in brackets. Data points at 40 °C (0.0032 K⁻¹) were not included in the linear regression because of the large errors in fitting the charge transfer semicircle at this temperature. As shown in Figs. S8-S22, at 40 °C the charge transfer semicircle is often similar in size or smaller than the high frequency, contact impedance semicircle, making it difficult to resolve features and achieve reliable fitting results at high temperatures.

Zoom In Zoom Out Reset image size

For the negative electrode in control cells, E_a decreases from 0.59 (0.02) eV to 0.48 (0.01) eV after ∼700 charge-discharge cycles. This decrease in activation energy suggests a different SEI composition with a lower energy barrier to Li⁺ transport after cycling. Figure 3a also showed that R_ct decreased by a factor of two after cycling in these control cells. Perhaps these decreases in R_ct and E_a are caused by increasing negative electrode SEI porosity after cycling in control cells. This is discussed further below in relation to capacitance data.

Figure 4b shows that E_a for negative electrode cells with 2% VC remains constant after ∼400 cycles (0.62 (0.02) eV after formation and 0.63 (0.01) eV after ∼400 cycles), and then starts to decrease after ∼700 cycles (0.54 (0.02) eV). Figure 3b showed no major changes in R_ct values for these cells. Figure 4c shows that E_a remains constant after cycling for LiPO₂F₂ cells. The mean and standard deviation across the formed cell, cell after ∼400 cycles, and cell after ∼700 cycles was 0.50 (0.01) eV. This suggests that cells containing LiPO₂F₂ form the most stable negative electrode SEI.

Figure 4d shows a significant increase in R_ct magnitude for the positive electrode after cycling in the cells with control electrolyte. Although the magnitude of R_ct changes dramatically, the slope of the line, and therefore the magnitude of E_a, remains the same which may surprise the reader. The increase in R_ct suggests that the SEI becomes more resistive after cycling. The similarity in E_a before and after cycling suggests similar SEI compositions on these electrodes, however, the XPS results in Fig. 7 below indicate that the SEI surface composition changes after long term cycling.

Figures 4e and 4f show changes in positive electrode E_a after cycling for cells with 2% VC and 1% LFO respectively. Figure 4e shows that E_a decreases from 0.71 ± 0.01 eV (after formation) to 0.63 (0.04) eV after cycling (mean and standard deviation of the two cycled cells). Figure 4f shows that E_a decreases from 0.64 ± 0.01 eV (after formation) to 0.55 (0.01) eV after cycling (mean and standard deviation of the two cycled cells with 1% LiPO₂F₂). These E_a decreases are accompanied by R_ct decreases for the positive electrode. Figures 4e and 4f showed that R_ct decrease by about a factor of two after cycling for cells containing 2% VC and 1% LFO. These decreases indicate better Li⁺ transport through the positive electrode SEI after cycling, which may be due to changes in SEI composition or morphology.

The behavior of the SEI on negative and positive electrodes does appear to differ upon cycling. In control cells, E_a for the negative electrode decreased after cycling, while E_a for the positive electrode remained the same. In cells containing 2% VC, E_a for the negative electrode remained the same after ∼400 cycles and then decreased after ∼700 cycles, while E_a for the positive electrode decreased after ∼400 cycles and then stayed the same after ∼700 cycles. In cells containing 1% LiPO₂F₂, E_a for the negative electrode saw no change as a function of cycling, while E_a for the positive electrode noticeably decreased between formation and cycling. While the mechanisms for these differences may not be clear, the differences between the behavior of the two electrodes is consistent with reports of differing SEI composition between positive and negative electrodes. XPS results for electrode samples from these cells are examined below.

Figure 5 shows R_contact vs temperature for the cells formed and cycled to 4.2 V in this comparison. No major differences are seen across (−/−) cells. R_contact for the (+/+) cells with control electrolyte appears to increase with cycling. However, for (+/+) cells with 2% VC and 1% LiPO₂F₂ electrolyte additives, R_contact appears to increase after ∼400 cycles and then decrease after ∼700 cycles. Differences are more prominent in cells cycled to higher voltages, which will be discussed below.

Zoom In Zoom Out Reset image size

Figure 6 shows capacitance vs temperature for these cells. Capacitance values were calculated using Eq. 2. The characteristic frequencies associated with the charge transfer impedance used to calculate these capacitance values can be seen in Fig. S3 in the supplementary material. Using the parallel plate capacitor equation—Eq. 3—to model SEI capacitance suggests that a strong temperature dependence is not expected since neither the area of the SEI (A), the thickness of the SEI (d), or the dielectric constant of the SEI (ε) should vary with temperature for a stable SEI film over the temperature range tested. Only small changes in capacitance are observed here over the temperature range under study, excluding high temperature data due to large fitting errors—fitting becomes less accurate at high temperatures with this data as contributions from contact impedance and charge transfer impedance overlap.

Zoom In Zoom Out Reset image size

Equation 3 suggests that capacitance should decrease with increased SEI thickness. Capacitance values are about an order or magnitude different between positive and negative electrodes, suggesting a thinner positive electrode SEI. Figure 6a shows that the negative electrode capacitance decreases by > 60% after cycling in control cells. In contrast, Figs. 6b and 6c show no changes in negative electrode capacitance in cells with VC and LiPO₂F₂ additives. This suggests that SEI thickening is happening on the negative electrode in the control cell, which is consistent with the literature.^10,27 Since SEI thickening consumes Li⁺, this is consistent with reports of Li⁺ inventory loss leading to capacity fade.^48,49 Figure 3a showed a decrease in R_ct for the control cell after cycling, and Fig. 4a showed a decrease in E_a after cycling. Perhaps this suggests a highly porous, thick SEI after cycling. Compared to a dense SEI, a porous SEI could have better Li⁺ transport, leading to a smaller measured R_ct.

Figure 6d shows that the capacitance of the positive electrode SEI increases slightly after cycling in control cells. This increase in capacitance would indicate a decrease in SEI thickness, which will be discussed in the next section where increases in positive electrode capacitance are larger.

Figure 7 shows the F 1 s, O 1 s, C 1 s, P 2p, Ti 2p, and "TM + Li" (transition metals and lithium) XPS spectra for NMC532 electrodes. Spectra from a fresh, uncharged NMC532 electrode are shown in the top row. The following rows contain spectra from electrodes extracted from cells after formation and after ∼700 cycles. These are electrodes from the same cells that underwent temperature dependent EIS studies shown above in this section. All formed and cycled cells were disassembled at 3.8 V. Spectra for electrodes from control cells are shown in the second and third rows; 2% VC cells are shown in the fourth and fifth rows; 1% LiPO₂F₂ cells are shown in the sixth and seventh rows. Note that y-axes for similar spectra are the same (i.e. all F 1 s spectra, all O 1 s spectra etc.), allowing for the comparison of similar peaks between spectra.

Zoom In Zoom Out Reset image size

The O 1 s spectra will be highlighted first. In the fresh electrode, two peaks are observed at 532 eV and 529.5 eV. The peak at 532 eV is typically assigned to surface oxygen,^10,29 and the peak at 529.5 eV is due to lattice oxygen anions in the NMC active material (shown in yellow here).^10,50 In formed and cycled cells, peaks at ∼533 eV and ∼531.5 eV are attributed to C–O and C=O bonds respectively. These peaks may be assigned to carbonaceous species from oxidation of electrolyte solvents at the NMC electrode.¹⁰ These peaks increase by different amounts in all cell types after cycling.

Compared to the fresh electrode, the lattice oxygen peak reduces slightly after formation, and a significant reduction is seen after cycling for all electrolytes. The lattice oxygen peak can be used as an indicator of SEI thickness. Upon SEI thickening, the active material gets covered in SEI, thus this peak signal weakens. Additionally, the Ti 2p spectra for all electrodes show two peaks which are associated with TiO₂ from the NMC electrode coating. The "TM + Li" spectra shows Ni, Co, and Mn peaks from the NMC active material. The TiO₂ and TM peaks are significantly smaller after cycling, in agreement with the decrease in the lattice oxygen peak. The magnitudes of both the lattice oxygen and TiO₂ peaks do not differ substantially between electrolytes after cycling. This suggests that SEI thickness on the NMC532 electrode may be similar for all electrolyte systems after cycling. Major differences in cycling data for control cells compared to cells with 2% VC and 1% LiPO₂F₂ additives were presented above for these cells. Additionally, R_ct magnitudes differed for (+/+) cells containing the three electrolytes after cycling. After cycling, R_ct of 2% VC cells was about two times larger than R_ct of 1% LiPO₂F₂ cells, and R_ct of control cells was nearly three times larger than R_ct of 2% VC cells. This suggests that SEI thickness alone may not be a major contributor to impedance and to capacity fade.

An additional peak at ∼534.3 eV is observed in the electrodes from 2% VC cells. This peak is associated with an oligomer of VC (oligo-VC) formed at the surface, and is highlighted in blue here.^10,29,30 This peak increases slightly after cycling, suggesting continued oligo-VC production during cycling.

Blue peaks are also shown in the C 1 s spectra as they may also be associated with the production of oligo-VC. Ouatani et al.,²⁹ Ota et al.,³⁰ and Madec et al.^9,10 associated peaks around these binding energies with CO₃ and CO₂ in oligo-VC. However, it is difficult to deconvolute these peaks from other peaks in the C 1 s spectrum. In the C 1 s spectrum of the fresh electrode, four peaks are shown with their assignments. The C-C peak appears to reduce after cycling for all electrolytes. The C-F peak at ∼290.8 eV is associated with the PDVF binder in the NMC electrodes and is highlighted in orange.^51,52

In the F 1 s spectrum, C-F appears around 688 eV and is also highlighted in orange.^51,52 C-F, associated with PVDF binder is the only visible peak in the fresh electrode, however, after formation and cycling, two more peaks appear in the F 1 s spectra. A peak at ∼687 eV is associated with fluorophosphates such as LiPF₆ and Li_xPO_yF_z (purple).^9,50 A fluorophosphate peak is also seen in the P 2p spectra at ∼137 eV.^9,50 The fluorophosphate peaks increase significantly after cycling for all electrolytes. The P 2p spectra show a second peak around 134 eV associated with phosphates (P_xO_y).^9,50 When comparing the electrodes with different additives after formation, it is clear that this P_xO_y is more prominent in the cell containing 1% LFO. After cycling, the cell with 1% LFO has the largest increase in P_xO_y.

Finally, LiF production is seen in the F 1 s spectra around 685 eV (green).^9,32 LiF results from the degradation of LiPF₆ salt in the electrolyte^2,20 and production of LiF has been observed to increase with electrolytes containing LFO additive due to the additional fluorine.²⁷ Fig. 7 shows LiF in all cells after formation, with the control cell having the most. After cycling, the control cell gains some LiF, the 2% VC cell has no visible LiF, and the 1% LiPO₂F₂ cell has the largest LiF peak. LiF peak intensities correlate well with the Li peak intensities in the "TM + Li" spectra. While the 2% VC cell lacks a clear LiF peak, the large P-F peak for this cell indicates that there has been LiPF₆ degradation. The lack of a visible LiF peak in the cycled 2% VC cell suggests either that 2% VC suppresses LiF production in comparison with the other electrolytes tested, or that LiF is no longer present in the surface layer of the SEI.

The biggest differences in SEI composition between electrolyte additives appears to be in inorganic compounds such as fluorophosphates and LiF. The P_xO_y peak also shows differences between cycled cells with different electrolytes, and oligo-VC may be a factor in the SEI of cells containing VC additive. Figures 3d and 3f show significantly different positive electrode charge transfer impedance after cycling for control cells and 1% LiPO₂F₂ cells respectively. Figures 4d and 4f showed different activation energies for cycled positive electrode control cells (0.63 (0.02) eV) and 1% LiPO₂F₂ cells (0.55 (0.01) eV for both cycled cells). In contrast, XPS results in Fig. 7 showed small differences in the SEI composition on the NMC532 electrodes of these cells. Further research is required to determine the cause of the SEI differences. It should also be noted that XPS is a surface technique, and therefore the composition of the entire depth of the SEI is not captured.

Figure 8 shows the F 1 s, O 1 s, C 1 s, and P 2p XPS spectra for the graphite (AG) electrodes. Spectra from a fresh AG electrode are shown in the top row. The following rows contain spectra from AG electrodes extracted from the same formed and cycled cells shown in Fig. 7.

Zoom In Zoom Out Reset image size

The C 1 s spectrum for the fresh electrode shows a peak at 284.2 eV associated with graphite.^9,10 After formation, the graphite is lithiated, shifting the peak to ∼282.5 eV.^9,10 The lithiated graphite peak is shown in dark gray. In a similar manner to the lattice oxygen peak in NMC532 electrodes, the lithiated graphite peak is an indicator of SEI thickness. After formation, the 2% VC cell has the smallest lithiated graphite peak, suggesting that it also has the thickest SEI. This is in agreement with the work of Madec et al.¹⁰ and Qian et al.²⁷ After cycling, the lithiated graphite peak is no longer observed for all electrolytes, indicating that the SEI is too thick for the photoelectrons to penetrate from below the SEI on the electrode. A minor increase in the C–O/C–H peak and a more significant decrease in the C–C peak is observed after cycling for all electrolytes. Again, peaks that may be associated with oligo-VC are shown in blue. In the C 1 s spectra, it is difficult to distinguish these peaks from C-F and C=O peaks. However, the additional peak in the O 1 s spectra at ∼534.5 eV is apparent for cells with 2% VC, suggesting the presence of oligo-VC in the SEI. The O 1 s spectra contain two peaks associated with carbonaceous species from the decomposition of solvents on the graphite electrode.^10,29 In both control and 1% LiPO₂F₂ cells, the C-O peak increases significantly after cycling.

Fluorophosphates appear in all F 1 s and P 2p spectra after formation (purple). The cell with 2% VC initially has more P-F than control and 1% LFO cells, but the P-F peaks increase to roughly the same magnitude in all electrolyte systems after cycling. P-O peaks also begin with different intensities after formation with the 1% LiPO₂F₂ cell showing a significantly larger peak. P-O peaks also increase to roughly the same magnitude after cycling. Lastly, the LiF peak in the F 1 s spectra is of interest. After formation, LiF appears with different intensities following: 2% VC > control > 1% LiPO₂F₂. After ∼700 charge-discharge cycles, the peak intensities of LiF follow: 1% LiPO₂F₂ > control > 2% VC.

The differences in SEI composition on the graphite electrodes with different electrolyte additives after cycling include differences in: LiF content, C-O content, C-C content, and oligo-VC for the cell containing VC. Figure 3 showed that the 2% VC cell had the largest negative electrode R_ct after cycling, while the control and 1% LiPO₂F₂ cells had similar, smaller negative electrode R_ct. Qian et al. have suggested that inorganic compounds, such as LiF, create SEIs with lower impedance by lowering the energy barrier and favoring Li⁺ diffusion.²⁷ Figure 8 shows higher LiF content in control and 1% LiPO₂F₂ negative electrode SEIs after cycling. Figure 4 also showed low E_a for control (0.48 (0.01) eV) and 1% LiPO₂F₂ cells (0.50 (0.01) eV) after cycling. Along with the small R_ct measurements, this may support Qian et al.'s hypothesis.

Table IV shows the matrix of cells in this comparison. All cells were formed or cycled to 4.4 V and disassembled at 3.8 V. The number of cycles and capacity retention are indicated in the table. Cycled cells are labeled by approximate cycle number (∼300c, ∼400c, or ∼700c) in future figures. Nyquist plots for negative and positive symmetric cells and full coin cells at −10 °C, 10 °C, 30 °C and 40 °C are shown in Figs. S8–S22.

Figure 9 shows the logarithm of R_ct vs 1/T for all cells in this 4.4 V comparison. E_a values and errors are shown in each panel of Fig. 9. In contrast to the cells cycled to 4.2 V shown in Fig. 4, no significant changes to positive electrode E_a are seen for these 4.4 V cells with all electrolytes after cycling. However, changes are seen in negative electrode E_a values after cycling for all electrolyte types. For negative electrodes, E_a reduces from 0.62 (0.01) eV (after formation) to 0.50 (0.01) eV (after ∼300 cycles) in control cells; 0.65 (0.02) eV (after formation) to 0.52 (0.01) eV (after ∼400 cycles) in cells with 2% VC; and 0.55 (0.02) eV (after formation) to 0.44 (0.01) eV (averaged over both cycled cells) in cells with 1% LiPO₂F₂. These reductions in E_a after cycling are consistent with reductions in R_ct after cycling in these cells, and the activation energy is lowest in cells with 1% LiPO₂F₂, supporting the improved transport qualities of SEIs containing LiPO₂F₂.

Zoom In Zoom Out Reset image size

Figures 9d and 9e show a significant increase in R_ct magnitude for the positive electrode after cycling in the cell with control and control + 2% VC electrolytes. However, E_a values do not differ significantly between formation and cycling. Averaged over all formed and cycled cells shown, the means and standard deviations for (+/+) cells are: 0.61 (0.01) eV for control cells, 0.65 (0.01) for control + 2% VC cells, and 0.63 (0.01) eV for control + 1% LiPO₂F₂ cells. There are only minor differences between values for all three electrolytes, suggesting similarities in Li⁺ transport through the positive electrode SEI for all three additives, despite very different R_ct behavior.

Nyquist plots at 10 °C, R_contact vs temperature, capacitance vs temperature, and characteristic frequency vs temperature plots can be seen in Figs. S4–S7 of the supplementary material. Figure S4 shows that negative electrode R_ct decreased after cycling in all electrolyte systems. Positive electrode R_ct increases significantly for control cells after cycling. R_ct also increased for the positive electrode with control + 2% VC electrolyte. This is different than the 2% VC results for the 4.2 V cells, where R_ct decreased after cycling. Ma et al. have observed higher impedance in full NMC111/graphite cells containing 2% VC additive cycled to 4.4 V, compared with cells cycled to 4.2 V.¹³ For cells with 1% LiPO₂F₂ additive, positive electrode R_ct initially decreases after ∼400 cycles, and then increases after ∼700 cycles to almost the same magnitude as after formation.

Figure S5 shows that R_contact for the negative electrode again did not show significant changes over cycling. However, R_contact for the positive electrode did increase after cycling which will be discussed with Fig. 11 below.

Figure S6 shows consistent capacitance values across the evaluated temperature range (excluding high temperature data due to large errors in fitting). Again, positive electrode capacitance is an order of magnitude larger than negative electrode capacitance, suggesting a thinner positive electrode SEI. Panels (d) and (e) show slightly increased capacitance after cycling for positive electrode symmetric cells with control and control + 2% VC additives. This suggests somewhat thinner SEI layers after cycling, which contradicts the literature,^9,10,28 and thus capacitance measurements may be an inaccurate indicator of SEI thickness in cases where the dielectric constant is unknown and could be changing with cycle number.

Two figures are presented in this section to summarize R_ct and R_contact results for all cells in this study. Figures 10 and 11 show R_ct vs cycle number and R_contact vs cycle number respectively. Only 10 °C measurements are presented in these figures for simplicity. Duplicate and triplicate cells data is shown and repeat measurements at 10 °C are also included.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figures 10a–10c show generally stable or decreasing negative electrode R_ct values over cycling for 4.2 V and 4.4 V cells with all additives. Figure 10d shows a significant increase in (+/+) R_ct after cycling in both the 4.2 V and 4.4 V cells with control electrolyte. A larger increase in impedance is observed for the cell cycled to 4.4 V, and dashed lines are shown in this plot as a guide to the eye. Figure 10e shows that (+/+) R_ct decreases with cycling for the 4.2 V cells with 2% VC but increases with cycling for the 4.4 V cells. This agrees with the higher impedance growth observed by Ma et al.¹³ in cells containing 2% VC electrolyte additive tested above 4.3 V. Figure 10f shows stable or decreased (+/+) R_ct throughout cycling for the cells with 1% LiPO₂F₂. This data suggests that cells with 2% VC cycled to 4.2 V and cells with 1% LiPO₂F₂ cycled up to 4.4 V suppress impedance growth over cycle life. This is consistent with the good cyclability of these cells, shown in Fig. 2 above.

Figures 11a–11c show no significant changes in negative electrode R_contact over cycle life for all electrolyte mixtures. Figure 11d shows increases in positive electrode R_contact after cycling for both the 4.2 V and 4.4 V cells containing control electrolyte. The increase is more sizeable for the 4.4 V cell after cycling. Landesfeind et al.⁵³ have shown increasing contact impedance with cycling in LiNi_0.5Mn_1.5O₄ cells (cycled to 4.9 V at 40 °C). The authors suggest that loss of contact between active cathode material and the current collector occurs under aggressive cycling conditions in these cells.⁵³ Atebamba et al. also suggest that volume change during repeated Li⁺ intercalation may affect the distribution of active particles and conductive additives.⁵⁴ Perhaps this may lead to the increased contact impedance after aggressive cycling observed here.

Figures 11e and f also show an increase in positive electrode R_contact over cycle life for the 4.4 V 2% VC and 1% LiPO₂F₂ cells. The trend is not the same for the 4.2 V 2% VC and 1% LiPO₂F₂ cells which show an initial increase followed by a decrease in R_contact over cycle life. The reason for this behavior is unclear.