Corrosion/Passivation Behavior of Concentrated Ionic Liquid Electrolytes and Its Impact on the Li-Ion Battery Performance

NMC532 cathode (LiNi_0.5Mn_0.3Co_0.2O₂; active material area loading 10.26 mg/cm²) laminates were supplied by the Cell Analysis, Modeling, and Prototyping (CAMP) Facility at Argonne National Laboratory. LFP cathode (LiFePO₄) was prepared by mixing LiFePO₄ (90%), C45 carbon black (5%), and PVDF binder (5%) in N-methyl-2-pyrrolidinone (NMP) and the resulting slurry was then cast onto Al foil (active material area loading: 10.79 mg/cm²). The electrode laminates were punched into 14 mm discs in diameter and dried at 120°C under vacuum for overnight. Li chips were purchased from MTI Corporation (15.6 mm in diameter and 0.2 mm in thickness).

PMpipFSI was synthesized by the ion exchange reaction of 1-methyl-1-propylpiperidinium bromide (IoLiTec Inc.) with LiFSI (Nippon ShokuBai Co., Ltd.). The ionic liquids were extracted with ethyl acetate and washed with de-ionized water until no residual halides were detected by saturated AgNO₃ solution. ILs were dried on lyophilizer for at least 2 days, and then stored over 4 Å molecular sieves before use. The water content was measured to be < 30 ppm by Mettler Toledo coulometric Karl-Fischer titrator C30.

The viscosity was measured by ViscoLab 4000 at room temperature. Ionic conductivity (σ) was determined by electrochemical impedance spectroscopy (EIS) using Solartron Analytical 1400 Cell test working station. The distance between the two stainless steel electrodes (l) was fixed to 0.172 cm by using a Teflon ring spacer with the inner area (A) of 0.502 cm². EIS was measured to obtain the ohmic resistance (R_Ω) of the electrolyte sample from the Nyquist plot at the intersection with Z' axis at the high frequency. The ionic conductivity was then calculated by Equation 1.

Cyclic voltammetry was performed using a Gamry Instrument potentiostat reference 600+. The measurements were carried out in a 2032-coin cells using Al (14 mm in diameter) or stainless-steel (SS) disc (15.6 mm in diameter) as working electrode, lithium as both counter and reference electrode (15.6 mm in diameter), and glass-fiber as the separator (16 mm in diameter). The scan rate was 10 mV/s.

Standard and Al-coated 2032-coin cell configurations are illustrated in Figure 1.³⁶ In the standard cell configuration (Figure 1a), two SS spacers are included: one is between the cathode/working electrode and cathode cap, and the other one is between the anode/Li chip and spring/anode cap. In the Al-coated 2032-coin cell configuration (Figure 1b), both SS spacers are placed on the anode side (between anode/Li chip and spring) with cathode cap coated with aluminum.

Zoom In Zoom Out Reset image size

Galvanostatic charge-discharge cycling of LFP/Li and NMC532/Li cells were conducted with standard and Al-coated 2032-coin cells on Maccor (MIMSclient) cycler. The glass fiber separator GF/F (16 mm in diameter) was used and the total electrolyte amount is 100 μL for both Gen 2 and PMpipFSI electrolytes. The effective electrode area was 1.54 cm² for LFP and NMC532 cathode. These cells were first formed with C/20 rate for 3 cycles followed by 100 cycles with C/10 rate. The cycling testing was conducted at 25°C and 55°C.

PMpipFSI based electrolyte with various LiFSI concentrations were prepared in an argon-filled glove box, and their physicochemical properties are summarized in Table I.

Figure 2a shows the cycling performance of 1 M LiFSI-PMpipFSI electrolyte in a LiFePO₄/Li cell with a cutoff voltage of 3.8-3.0 V, which is the widely used cathode for IL-based electrolyte evaluation.^37,38 The cell was tested in a standard 2032-coin cell setup (Figure 1a). Surprisingly, when a higher voltage NMC532/Li cell was used (cutoff voltage 4.3-3.0 V), a low initial capacity of 140.6 mAh/g was obtained and the capacity fades rapidly within first 30 cycles with only 80% Coulombic efficiency as shown in Figure 2b. Apparently, the poor cell performance was not caused by the electrode or the testing condition since the conventional Gen 2 electrolyte (1.2 M LiPF₆ EC/EMC 3/7) cell cycles well as evidenced by the cycling data shown in Figure 2b. The inherent property of high voltage stability of IL-electrolytes excludes the possibility of oxidative decomposition at 4.3 V,¹⁷ indicating other side reactions occur at this voltage. Interestingly, when the LiFSI concentration increases, the cycling performance dramatically improved. Figure 2c showed the capacity retention and Coulombic efficiency for 2–5 M LiFSI-PMpipFSI cells. 4 M LiFSI is a threshold value since no obvious change in cell performance when salt concentration is further increased to 5 M, the maximum salt concentration. It is noticeable that although the increased salt concentration improves the cell performance, the cell capacity still degrades cycle by cycle and the Coulombic efficiency is still low (only 90–95%), indicating that a different underpinned mechanism dictates the observed electrochemical behavior, which leads us to investigate the corrosion/passivation behavior on the metals in the standard 2032-coin cell setup.

Zoom In Zoom Out Reset image size

Cyclic voltammetry was conducted in the standard 2032-coin cell setup using Al foil as working electrode and Li chip as counter and reference electrode. Figure 3 shows the CV profiles of 1–5 M LiFSI-PMpipFSI and Gen 2 electrolytes. Gen 2 electrolyte shows the typical passivation behavior after the 1^st anodic scan to 6.0 V and no corrosion was observed afterwards. However, all the IL-electrolytes exhibits different degree of pitting corrosion^39–41 during the 1^st anodic scan (Figure 3a), and this behavior continues in the following scans (Figure 3b) indicating these electrolytes could not passivate the metal. Another important observation is that the corrosion current density decreases with increased LiFSI salt concentration, suggesting the highly concentrated PMpipFSI electrolyte could kinetically suppress the corrosion to a certain degree that could support the normal redox reactions on the cathode at 4.3 V. The CV data correlate well with the improved cycling of 5 M LiFSI PMpipFSI electrolyte in NMC532/Li cell reported in Figure 2c. Cho et al. reported the same cyclic voltammetry of 1 M LiTFSI in 1-methyl-1-propylpyrrolidinium bis(fluorosulfonyl)imide (PMpyrFSI) ascribing to the Al corrosion issue.²⁸

Zoom In Zoom Out Reset image size

However, there are two possible corrosion sources in this standard 2032-coin cell setup: Al and SS.^42–45 To isolate these two corrosion sources, an Al-coated 2032-coin cell setup was employed for cyclic voltammetry study. As shown in Figure 1b, the internal side of the cathode cap (made of SS) is coated with Al and the SS spacer placed between the cathode cap and the cathode/working electrode was moved to the anode side in order to evaluate the Al corrosion/passivation behavior. Figure 4 summarize the cyclic voltammograms of 1 M, 5 M LiFSI-PMpipFSI and Gen 2 electrolyte, respectively. For 1 M and 5 M PMpipFSI electrolytes, the Al oxidation starts at 3.2 V and 3.5 V in the first anodic scan and the corrosion current density is 50 times smaller than that obtained in the standard 2032-coin cell setup shown in Figure 3a (Al working electrode with an adjacent SS spacer). Different from the results tested in standard 2032-coin cell setup, no corrosion reaction was observed at the 2^nd and 3^rd scan, indicating both 1 M and 5 M electrolyte could effectively passivate the Al and thus inhibit its corrosion. As expected, Gen 2 electrolyte showed the passivation behavior on Al as evidenced from the CV data in Figure 4c. At this point, it is evident that both low and high concentration PMpipFSI electrolytes could effectively passivate the Al, and the large corrosion current observed in the standard 2032-coin cell setup were originated from the corrosion of SS spacer used in the cell.

Zoom In Zoom Out Reset image size

To further confirm it, cyclic voltammetry analysis using SS as working electrode in standard 2032-coin cell setup was carried out. Cyclic voltammograms for the 1^st and 3^rd cycles are provided in Figure 5. As expected, Gen 2 electrolyte is capable of passivating SS and inhibits corrosion in the subsequent cycles, while all the PMpipFSI electrolytes showed strong corrosion currents. The corrosion current density is 3 times larger than that shown in Figures 3a–3b when Al working electrode was used in the standard coin cell setup. Similar results as shown in Figure 2c, when the LiFSI concentration increases from 1 M to 4 M and 5 M, the SS corrosion is mitigated to a degree that the electrolyte enables the regular cycling of NMC532 cathode at 4.3 V. In conclusion, these cyclic voltammetry studies indicate: (1) the LiFSI-PMpipFSI IL electrolytes could well passivate Al current collector, but corrodes the SS, and (2) the highly concentrated PMpipFSI electrolytes could kinetically suppress the stainless-steel corrosion reaction to a low degree that it could support the normal redox reaction of NMC cathode.

Zoom In Zoom Out Reset image size

The cyclic voltammetry study elucidates the corrosion behavior of the PMpipFSI based electrolyte on Al and SS. We re-evaluated the electrochemical performance of PMpipFSI electrolytes in NMC532/Li cells using a revised 2032-coin cell setup (illustrated in Figure 1b where an Al-coated cathode cap without a SS spacer was used on the cathode side). Figure 6a showed the specific discharge capacity and Coulombic efficiency of 1 M and 5 M LiFSI-PMpipFSI electrolytes. Both electrolytes showed excellent cycling stability. Specifically, 1 M electrolyte cell delivers an initial capacity of 152 mAh/g and 149 mAh/g at 100 cycles. For 5 M cell, it delivers a lower capacity of 139 mAh/g due to its high viscosity (Table I). The capacity retention is superior for this electrolyte (a higher capacity of 147 mAh/g at 103 cycle is due to slow electrode wetting). Compared with only 90% obtained in the standard 2032-coin cells, the average Coulombic efficiency for the Al-coated cells is 98.9% and 99.8% for 1 M and 5 M LiFSI-PMpipFSI, respectively.

Zoom In Zoom Out Reset image size

To evaluate the stability of the Al passivation layer at elevated temperature, NMC532/Li cell were extensively cycled at 55°C. Due to the decrease in viscosity, these cells could be cycled with a high C-rate after the formation step. The capacity retention and Coulombic efficiency are shown in Figure 6b. At increased temperature, the initial capacity increases from 152 mAh/g to 176 mAh/g for 1 M LiFSI-PMpipFSI cell and from 139 mAh/g to 181 mAh/g for 5 M LiFSI-PMpipFSI cell. When the C-rate increased to C/3 for extended cycling, the cell capacity slightly decreased to 159 mAh/g for 1 M LiFSI-PMpipFSI and 169 mAh/g for 5 M LiFSI-PMpipFSI, both are still higher than those of room temperature with C/10 rate. The capacity retention is 99.2% in 106 cycles for 1 M LiFSI PMpipFSI with an averaged Coulombic efficiency only 97.4%, which is presumably due to the dendrite formation during the cycling. For 5 M LiFSI-PMpipFSI electrolyte cell, 99.3% capacity retention and a high average Coulombic efficiency of 99.9% was observed. These cell data indicate that the Al passivation layer formed in the PMpipFSI based electrolytes is stable even at elevated temperatures.