Comparative Study on Charge–Discharge Behavior of Graphite Positive Electrode in FSA- and FTA-Based Ionic Liquid Electrolytes with Different Alkali Metal Cations

To investigate the graphite intercalation/deintercalation behavior of the FSA anions, charge–discharge tests were conducted for 5 cycles using M/graphite coin cells (M = Li, Na, and K) with M[FSA]–[C₃C₁pyrr][FSA] (x = 0.20) IL electrolytes. In addition, the structural changes in graphite were analyzed using ex situ XRD measurements of the graphite electrodes charged to the upper cut-off voltages, as shown in Table I. To assign the obtained XRD patterns to stage-n GICs (n: stage index), calculations are performed as follows. First, the d-spacing (d) is obtained based on Bragg's equation:

$$\begin{eqnarray}&&\lambda =2dsin\theta \end{eqnarray} \tag{ 1 }$$

where λ is the wavelength of the X-ray radiation (1.5418 Å) and θ is the diffraction angle. The stage index (n) is then determined using the following equation:

$$\begin{eqnarray}&&{I}_{c}=\left(n-1\right){d}_{0}+{d}_{i}\end{eqnarray} \tag{ 2 }$$

where I_c is the repeat unit distance including the graphene layer and the layer forming GIC, d₀ is the distance between pristine graphene layers (3.35 Å), and d_i is the interlayer distance between the layers in which intercalants are present.

Figure 1 shows the charge–discharge curves of the graphite positive electrodes in M/graphite half-cells with FSA-based IL electrolytes and the XRD patterns of the graphite electrodes charged to the upper cut-off voltages. In the Li[FSA]–[C₃C₁pyrr][FSA] electrolyte (Fig. 1a), the graphite positive electrode exhibits an initial charge capacity of 60 mAh g⁻¹, which is accompanied by two plateau regions at approximately 4.6 and 4.8 V. In contrast, the initial discharge curve mainly consists of one plateau at 4.1 V, exhibiting an initial discharge capacity of 29 mAh g⁻¹. Thus, the initial coulombic efficiency as low as 49% is obtained, possibly due to the irreversible decomposition of the electrolyte such as the formation of solid electrolyte interphase (SEI) on the electrode surface. Similar behavior is observed after the 2nd cycle, except for the shrinkage of the charging plateau at 4.6 V, and the 5th discharge capacity is found to be 23 mAh g⁻¹. As illustrated in Fig. 1d, the graphite electrode charged to 4.9 V in the Li[FSA]–[C₃C₁pyrr][FSA] electrolyte exhibits two prominent peaks at approximately 23.2° and 31.7°, which are different from the 002 diffraction peak of graphite (26.6°) ²⁹ observed for the pristine electrode. These diffraction peaks can be attributed to stage-2 GICs with d_i = 8.2 Å using Eqs. 1 and 2, which is almost consistent with previous studies on FSA-GIC formation. ²¹ Furthermore, as shown in Figs. S1a and S2a, when the upper cut-off voltage was set to 5.2 V, graphite finally changed into stage-1 FSA-GIC at the full-charged state; however, the graphite electrode exhibits more unstable charge–discharge behavior and loses its capacity after 5 cycles, which is probably due to the severe electrode degradation at higher voltages.

Zoom In Zoom Out Reset image size

In the case of the Na[FSA]–[C₃C₁pyrr][FSA] electrolyte (Fig. 1b), the charge–discharge behavior was completely different from that of its Li-based counterpart. The charge–discharge curves are composed of a multi-step plateau region that overlaps each other for 5 cycles, indicating the stable and reversible reactions of the graphite electrode. In the 1st cycle, the discharge capacity is 80 mAh g⁻¹ with a coulombic efficiency of 78%, which is larger than the capacity of the Li-based electrolyte, and approximately the same values are retained after 5 cycles. Notably, while the initial charge capacity appears around 4.4 V, the onset potential is slightly lower (4.3 V) after the 2nd cycle, which can be attributed to an activation process arising from the kinetic hindrance of the graphite electrode in the 1st cycle. As shown in Fig. 1d, the XRD pattern of the graphite electrode charged to 4.9 V mainly comprises three sharper peaks at 20.6°, 27.6°, and 34.7° and two broader peaks at approximately 23.7° and 31.9°. Although the broader peaks are assigned to stage-2 FSA-GIC with d_i = 7.9 Å as in the case of the Li-based electrolyte, the d_i values of the sharper peaks are calculated to be greater than 8 Å, which has not been previously reported. Assuming that these unknown peaks can be attributed to the formation of stage-1 or stage-2 GICs, the d_i values obtained are 12.9 or 9.6 Å, respectively. As described in Figs. S1b and S2b, graphite completely transforms into an unknown phase at 5.0 V. Furthermore, the graphite electrode shows relatively stable cycling behavior even at an upper cut-off voltage of 5.0 V. One of the possible reasons for this behavior is the different liquid structure of Na[FSA]–[C₃C₁pyrr][FSA] such as the complex ion state (e.g., [Na(FSA)_n ]¹⁻ⁿ ) in the vicinity of the graphite electrode surface compared to that of other FSA-based ILs. The details of this interesting behavior will be investigated in the near future.

In the K[FSA]–[C₃C₁pyrr][FSA] electrolyte, as shown in Fig. 1c, the graphite electrode exhibits charge–discharge behavior similar to that of its Li-based counterpart. The initial charge and discharge capacities are 63 and 26 mAh g⁻¹, respectively, resulting in a coulombic efficiency as low as 42%. Then, the discharge capacity decreases to 21 mAh g⁻¹ in the 2nd cycle, and remains the same till the 5th cycle. The onset potential in the charging process of the 1st cycle is approximately 4.8 V, and the potential plateaus appear at approximately 4.8 and 5.1 V, which corresponds well with the reaction potential considering the potential difference between lithium and potassium. ¹⁴ The XRD pattern of the graphite electrode charged to 5.15 V consists of two peaks around 23.9° and 31.4°, which are similar positions as those of the Li-system, thus indicating the formation of stage-2 FSA-GIC with d_i = 7.8 Å. Although stage-1 FSA-GIC is formed in the charged state when the graphite electrode is charged to 5.4 V, the reversible capacities are approximately 50 mAh g⁻¹ with a large voltage polarization (See Figs. S1c and S2c).

To ascertain the charge–discharge mechanism of graphite positive electrodes in FTA-based IL electrolytes, charge–discharge tests were conducted using M[FTA]–[C₄C₁pyrr][FTA] (x(M[FTA]) = 0.20) ILs, followed by XRD measurements at the full-charged states. Figures 2a–2c show the charge–discharge curves of the graphite electrodes in the Li-, Na-, and K-based ILs, respectively, for the initial 5 cycles. Overall, the graphite electrodes exhibit similar charge–discharge behavior in all the electrolytes, where the charge–discharge curves consist of a multi-step plateau region, suggesting the reversible formation of FTA-GICs. The initial charge capacities are 141, 140, and 133 mAh g⁻¹ for the Li, Na, and K systems, respectively. The XRD results (Fig. 2d) indicate that the graphite electrodes in their full-charged states are mainly composed of stage-1 FTA-GIC with d_i = 8.0 Å for all the systems, which is consistent with the results of previous studies. ²⁷ Small amount of stage-2 FTA-GIC remains in the case of the Na system, corresponding to a smaller initial discharging capacity as described below. In addition, the d_i values of FTA-GIC are almost equal for all FTA-based electrolytes, which is different from the case of FSA-based electrolytes (See also Fig. 1d), suggesting that the charge–discharge mechanism is not identical for different amide anions.

Zoom In Zoom Out Reset image size

The initial discharge capacity and coulombic efficiency are 103 mAh g⁻¹ and 73% for the Li system, 89 mAh g⁻¹ and 64% for the Na system, and 95 mAh g⁻¹ and 71% for the K system. In all cases, larger capacities are obtained compared to those obtained using FSA-based electrolytes, which is attributable to the long voltage plateau appearing in the highest voltage region. Concerning the cycling stability, a slight capacity decrease is observed in the Li system; the discharge capacity is 97 mAh g⁻¹ in the 5th cycle, with a capacity retention of 94%. In contrast, for the Na and K systems, the charge–discharge curves almost overlap each other for 5 cycles, with discharge capacities of 89–90 mAh g⁻¹ and 95–98 mAh g⁻¹, respectively. The performance of the graphite positive electrodes is summarized in Table I. From the above, it can be affirmed that the K[FTA]–[C₄C₁pyrrr][FTA] electrolyte is the most suitable for the graphite positive electrodes among all electrolytes, including FSA-based ILs.

To discuss the common features and differences in the graphite positive electrode performance in the ILs, charge–discharge measurements were performed using a three-electrode cell with the corresponding alkali metal used as the reference electrode. Figure 3a displays the charge–discharge curves of the graphite positive electrode at the 2nd cycle in the FSA-based electrolytes operated at 20 mA g⁻¹ with respect to the Li⁺/Li potential, which is calibrated by the redox potential of the alkali metals in the FSA-based ILs. ¹⁴ The onset potential of charging begins at approximately 4.5 V in all the systems, which suggests that FSA⁻ anions in the electrolyte are intercalated between the graphene layers at a similar potential to form anion-GICs. However, in the Li and K systems, the potential polarization during charging is higher than that in the Na system, leading to lower charge–discharge capacities. In addition, during the discharge process, the potential plateau appears at approximately 4.8 V in the Na system, whereas it is absent in the Li and K systems, indicating that the charging plateau at 4.95 V observed in the Na-based electrolyte is reversible.

Zoom In Zoom Out Reset image size

Similar tests were conducted for the M[FTA]–[C₄C₁pyrr][FTA] electrolytes, as shown in Fig. 3b. The potentials are calibrated based on the redox potentials of the alkali metals in the FTA-based ILs. ¹⁵ Evaluation using three-electrode cells reveals that the charge–discharge curves showed a similar trend for all systems after potential calibration. The onset potentials are around 4.5 V vs Li⁺/Li, which is close to that of FSA-based electrolytes. Furthermore, the plateau potential appears at approximately 5.0 V vs Li⁺/Li for all the electrolytes. The discharge curves consist of three gentle plateaus, whereas only two plateaus are observed for the Na[FSA]–[C₃C₁pyrr][FSA] electrolyte. In conclusion, the intercalation/deintercalation potentials of amide anions are found to be common and independent of the anionic species and metal cations present in the IL electrolytes.

Figure 3c shows the comparison of the charge–discharge curves of the graphite electrodes in the 2nd cycle in the Na[FSA]–[C₃C₁pyrr][FSA] and Na[FTA]–[C₄C₁pyrr][FTA] IL electrolytes. The potentials are calibrated to the Fc⁺/Fc redox couple, as described in Supplementary Information. Long plateaus during charging and discharging processes are seen at 1.7–1.8 V vs Fc⁺/Fc for both systems, which is characteristic of the formation of amide anion GICs, suggesting that FSA-GIC and FTA-GIC are likely to form at similar potentials. Several potential plateaus corresponding to higher-stage GICs are also observed below the long plateaus. However, the staging mechanism does not seem necessarily identical because a distinct plateau is observed at 0.8 V vs Fc⁺/Fc for the FTA-based IL whereas it is absent for the FSA-based IL.

To further explain the effect of the electrolytes on the performance of the graphite positive electrodes, chronoamperometric measurements were conducted on the Al current collector in the FSA- and FTA-based IL electrolytes, as shown in Fig. 4. Stepwise constant voltages were applied to the M/Al two-electrode coin cells to simulate the charge–discharge profiles of the graphite positive electrode, as described in the experimental section. In the Li[FSA]–[C₃C₁pyrr][FSA] electrolyte (Fig. 4a), the anodic current density increases during the initial 6 h when higher voltages are applied to the cell. At a constant voltage of 5.0 V, the current density is initially approximately 7 μA cm⁻² and is stabilized at a value as high as 6 μA cm⁻² after 72 h. In contrast, for the Li[FTA]–[C₄C₁pyrr][FTA] IL, no substantial change in the current density is observed, except for a short period immediately after the potential steps. The anodic current densities are suppressed to a low value of approximately 0.5 μA cm⁻² during the entire period.

Zoom In Zoom Out Reset image size

For the Na system, the current densities in the Na[FSA]–[C₃C₁pyrr][FSA] electrolyte are lower than those of its Li-based counterpart, reaching only 1.1 μA cm⁻² even after 78 h (Fig. 4b). The Al current collector in Na[FTA]–[C₄C₁pyrr][FTA] exhibits slightly larger anodic currents over 1 μA cm⁻² in the initial 9 h; however, the value has gradually decreased to 0.7 μA cm^–2 after 78 h. Therefore, the FTA-based IL is more resistant to Al corrosion in the long term. Similar trends are observed in the K-based ILs as shown in Fig. 4c. For K[FSA]–[C₃C₁pyrr][FSA], the anodic current density of an Al foil is maintained at approximately 2.0 μA cm⁻² after polarization at 5.2 V. In contrast, extremely low currents around 0.2 μA cm⁻² flow into the Al current collector at all times in the case of the K[FTA]–[C₄C₁pyrr][FTA] electrolyte, which indicates that it exhibits the highest tolerance against Al corrosion.

From the above results, it can be concluded that the corrosion of the Al current collector in FTA-based ILs is suppressed compared to that in FSA-based ILs, regardless of the alkali metal cation species in the electrolytes, which is consistent with the superior charge–discharge performance of the graphite positive electrode at high voltages. In addition, according to our previous studies, ^14,15 the FSA-based ILs show wider electrochemical windows on glassy carbon electrodes compared to the FTA-based ILs as shown in Table S1, which also suggests that electrolyte decomposition except for Al corrosion does not significantly affect the graphite positive electrode behavior under the voltage conditions used in the present study. While the corrosion behavior of the FTA-based ILs is almost independent of the alkali metal cations, the FSA-based IL exhibits a strong dependence on the alkali metal cations. Furthermore, the Na[FSA]–[C₃C₁pyrr][FSA] electrolyte suppresses Al corrosion to a level as low as that of the Na[FTA]–[C₄C₁pyrr][FTA] electrolyte, which may correspond to the differences in charge–discharge behaviors of other FSA-based ILs, as shown in Figs. 1 and 2.

The previous studies reported that Al corrosion currents are suppressed in TFSA-based ILs containing Li⁺ cation compared to FSA-based ILs, ^28,29 which may be caused by the lower solubility of Al-TFSA compounds (e.g., Al[TFSA]₃). ^28–30 Since the magnitude of Al corrosion currents in FTA-based ILs lies between FSA- and TFSA-based ILs, ²⁸ the solubility of Al-FTA compounds is likely to be lower than that of Al-FSA compounds and higher than that of Al-TFSA compounds. Thus, the FTA-based ILs suppressed the Al corrosion compared to the FSA-based ILs in the present study. Among the FSA-based ILs, only the IL containing Na⁺ (Na[FSA]–[C₃C₁pyrr][FSA]) exhibited significantly smaller currents comparable to FTA-based ILs. One possible explanation of this unique behavior is that the liquid structure, i.e., the complex ion state of the cations and anions, is different only in Na[FSA]–[C₃C₁pyrr][FSA], as mentioned before, which suppresses the dissolution of Al-FSA compounds and allows high corrosion resistance.

As discussed above, K[FTA]–[C₄C₁pyrr][FTA] (x(K[FTA) = 0.20) was found to confer good charge–discharge performance to the graphite positive electrode. To further substantiate the performance for constructing DCBs, the detailed electrochemical behavior of graphite in this IL electrolyte was investigated using ex situ XRD measurements and long-term cycling tests.

Figure 5a shows the initial charge–discharge curve of the graphite electrode in the K[FTA]–[C₄C₁pyrr][FTA] IL electrolyte at 20 mA g⁻¹ and 298 K, which is identical to the initial charge–discharge curve in Fig. 2c. The charge and discharge capacities were 133 and 95 mAh g⁻¹, respectively. Figure 5b summarizes the ex situ XRD patterns of the graphite electrode during the initial cycle, which depict the phase transition of the FTA-GICs; each number corresponds to the state of charge provided in Fig. 5a. The samples were prepared by charging (and discharging) at constant current densities of 20 mA g⁻¹ for points 1–6 and 5 mA g⁻¹ for point 7. In the pattern of the pristine graphite electrode, a sharp peak appeared at 26.6° assignable to 002 diffraction of graphite. At point 1 (30 mAh g⁻¹), the intensity of the graphite diffraction peak decreases, and two peaks appear at approximately 24.2° and 30.3°; the latter can be assigned to stage-3 FTA-GIC with d_i = 8.0 Å. Upon further charging to point 2 (53 mAh g⁻¹), the diffraction peaks have shifted to 23.6° and 31.7°, corresponding to the formation of stage-2 FTA-GICs with d_i = 8.0 Å. When the charging capacity reaches point 3 (91 mAh g⁻¹), four diffraction peaks are obtained in the XRD pattern; those detected at 23.4° and 32.0° are assigned to stage-2 FTA-GIC and those at 22.2° and 33.5° to stage-1 FTA-GIC. The d_i value of the stage-2 GIC seems slightly larger than that observed at point 2, implying the existence of a single-phase region with different interlayer concentrations for this compound. When the graphite electrode is further charged to an upper cut-off voltage of 5.4 V (point 4, 133 mAh g⁻¹), two dominant peaks are observed at 22.2° and 33.6°, which are assigned to stage-1 FTA-GIC. Thus, the long plateau at approximately 5.2 V corresponds to the coexisting state of stage-1 and stage-2 FTA-GICs.

Zoom In Zoom Out Reset image size

The peak intensities in the XRD pattern during discharging are smaller than those during charging, indicative of a decrease in graphite crystallinity owing to large volume changes during the charging and discharging processes. Upon discharging to point 5 (102 mAh g⁻¹), four broad diffraction peaks are confirmed: the two peaks appearing at 23.4° and 31.9° are assigned to stage-2 GIC and the two at 22.3° and 33.5° to the remaining stage-1 GIC. This indicates that the discharge plateau at approximately 5.0 V corresponds to the charge plateau at 5.2 V. After further discharging to point 5, commensurate with a cumulative capacity of 62 mAh g⁻¹, four diffraction peaks are observed, with two peaks attributable to stage-3 GIC appearing at 24.4° and 29.9° and two peaks attributable to stage-2 GIC at 23.6° and 31.1°. Finally, the XRD pattern of the graphite electrode discharged to 3.6 V at 5 mA g⁻¹ is shown at point 7. A significant decrease in peak intensity is confirmed, and the diffraction peak is detected at almost the same position as that of pristine graphite, indicating that the discharge is almost complete. However, the existence of another peak at 25.5° corresponding to dilute stage or high stage GICs suggests that a small amount of FTA⁻ remains in the graphite even after a complete discharge. If incomplete deintercalation is the only cause for coulombic efficiencies lower than 100%, anions would accumulate between graphite layers during long-term cycling, leading to a decrease in reversible capacity. In contrast, the results show the stable cycling characteristics for 50 cycles, as shown in Fig. 6. Therefore, Al corrosion or other side reactions such as SEI formation cannot be completely ruled out, and such side reactions that may occur on graphite electrodes could be one of the causes.

Zoom In Zoom Out Reset image size

Figure 6 shows the results of the cycling performance tests of the graphite positive electrode at a current density of 20 mA g⁻¹ at 298 K with the K[FTA]–[C₄C₁pyrr][FTA] electrolyte. The initial charge and discharge capacities are 115 and 88 mAh g⁻¹, respectively. The resulting coulombic efficiency is 76%, which is attributed mainly to irreversible reactions such as SEI layer formation on the graphite electrode or the Al current collector, as discussed in the previous sections. The discharge capacities are maintained at approximately 90 mAh g⁻¹ for 50 cycles, and the coulombic efficiency gradually increases from the 2nd cycle, reaching 91–95% after the 5th cycle. As mentioned above, the irreversible capacities observed in every cycle are possibly due to a small amount of continuous side reactions (e.g., Al corrosion), although their origin is still unclear. As shown in Fig. 6a, the charge–discharge curves almost overlap for 50 cycles except for the 1st cycle with a negligible increase in polarization, indicating that graphite is a promising positive electrode material for DCBs in the K[FTA]–[C₄C₁pyrr][FTA] IL electrolyte.