3.1. DME, DOL, and Their Mixture as a Solvent
The suitability of DME, DOL, and their mixtures for Li-ion battery electrolytes was first examined using Li/Gr and Li/NCM811 coin cells.
Figure 1 shows the voltage profile of the first cycle of Li/Gr cells at 0.1 C. As predicted, the Li/Gr cell with DME (EL02) produces a large amount of irreversible capacity, especially in the voltage regions above 0.5 V due to the co-intercalation of DME with the Li
+ ions and the resultant reduction, which, meanwhile, exfoliates the structure of graphite. As such, in the following charge, no reversible capacity can be observed except for some capacitance behaviors. However, the Li/Gr cell with DOL (EL04) presents a typical characteristic of Li
+ ion intercalation in graphite, as indicated by three long voltage plateaus below 0.3 V, which generates 2.67 mAh cm
−2 of reversible capacity and 85.3% of Coulombic efficiency (CE), confirming previous reports [
7,
8]. The outperformance of the LiFSI DOL electrolyte can be attributed to the weak coordination between DOL and Li
+ ions, which facilitates the solvation and desolvation of the Li
+ ions at the electrolyte–electrode interface and consequently mitigates co-intercalation of DOL in the Gr electrode. Additionally, it is found that the 1:1 DME/DOL mixture does not lead to a complementary effect in the electrochemical performances of the Li/Gr cell, as indicated by the cell with EL06.
The effect of LiFSI concentration on the performance of LiFSI DOL electrolytes in Li/Gr cells is further evaluated by comparing the voltage profile of the first discharge, as indicated in
Figure 2. It is observed that there is an optimum concentration at 1.2 m (EL04) for CE. Regardless of the LiFSI concentration, however, the voltage of all Li/Gr cells shows a sharp drop at the very beginning of the first discharge and quickly recovers (see the inset of
Figure 2). Such a phenomenon was also observed in previous publications [
7,
8]; therefore, it can be considered a typical characteristic of the LiFSI DOL electrolytes in the Li/Gr cells. This voltage drop reveals that the initial formation of a SEI experiences a high polarization, suggesting that the initial SEI formed with the LiFSI DOL electrolyte is rather resistive.
In addition, it is noticed that as of the time of writing this paper (after ca. three months), EL03, EL04, and EL05 have become a gel-like solution as a result of the acid-initialized ring-opening polymerization of DOL. This can be attributed to the structural rearrangement of FSI anions, as suggested by Equation (1), where Compound I is an SO
3-like strong Lewis acid.
The suitability of the DME, DOL, and 1:1 DME/DOL electrolytes for Li-ion batteries was further evaluated using high-voltage Li/NCM811 cells.
Figure 3a,b indicates the voltage profile of the first cycle and cycling performance of the cells. As observed in
Figure 3a, all three cells can be successfully charged to 4.3 V, generating a CE higher than 80%. However, the capacity of two cells with DME (EL02) and DOL (EL04) fades quickly, not lasting for up to 10 cycles, as indicated in
Figure 3b. Interestingly, the 1:1 DME/DOL electrolyte (EL06) shows a synergistic effect. In the first cycle (
Figure 3a), it obtains the highest CE (86.1% vs. 84.8% and 82.8% of the DME and DOL electrolytes), and in the following cycles (
Figure 3b), it retains the most stable values of capacity retention and CE.
Among the three electrolytes, the DOL electrolyte (EL04) shows the worst performance, although similar electrolytes have been reported to work well in 4.0 V Li/LiFePO
4 cells [
3,
7,
8]. Besides the high charging cutoff voltage (4.3 V) of Li/NCM811 cells, oxygen release and the resultant transitional metal ion (M
2+) dissolution of the delithiated layer-structured cathode materials must play an essential role in affecting the performance of DOL electrolytes. For example, under the catalysis of a trace number of Co
2+ ions, the oxidation of DOL by oxygen at 60 °C for 24 h can reach over 70% of yield according to the route of Equation (2) [
9], and ethylene glycol monoformate (namely Compound II) can be further oxidized to produce detrimental organic acids, H
2O, and CO
2.
The LiFSI–DOL electrolytes are also evaluated in a 200 mAh Gr/NCA pouch cell by using 1.0 m LiFSI (EL03) and 1.5 m LiFSI (EL05) electrolytes and comparing them with the 1.2 m LiFSI 3:7 EC/EMC baseline electrolyte (EL01). The voltage profiles and cycling performances of these three pouch cells are compared in
Figure 4. In the first charge, the EL03 and EL05 cells show a small voltage peak in the beginning, as marked by an arrow in the inset of
Figure 4a, which corresponds to the initial voltage drop in the Li/Gr cell of
Figure 2. It can be observed from
Figure 4a that although high concentration (EL05) is beneficial to CE (75.3% of EL05 vs. 70.0% of EL03), the CEs of the EL03 and EL05 cells are much lower than that of the EL01 cell. Furthermore, in the continuous cycling test (
Figure 4b), the capacity of the EL03 and EL05 cells fades rapidly, accompanied by low CEs and gas generation, whereas that of the EL01 cell remains very stable. After the cycling test, the EL03 and EL05 cells were found to be severely swollen, and the post-mortem analysis found that the electrolyte in the EL03 and EL05 cells was completely depleted and that the electrodes and separator were tightly stuck together by a dried polymer. The above results reveal that the poor performances of the EL03 and EL05 cells are attributed to not only the M
2+-catalyzed oxidation of DOL by oxygen but also the electrochemical polymerization of DOL at high potentials, as suggested by La Monaca et al. [
10].
3.2. Effect of FEC on DME Electrolyte
FEC has been established as a multi-functional additive that enables ethers, such as tetramethylene glycol dimethyl ether [
11], to form a robust SEI with Gr and Si anodes. Our previous work [
6] demonstrated that using 5 wt% FEC as the additive can enable the use of up to 20 wt% DME in the electrolyte without adverse effects on the cell’s performances while significantly enhancing the fast-charging capability of Li-ion batteries. Therefore, in the present work, FEC is used to attempt to facilitate the formation of a SEI in DME electrolytes.
Figure 5 displays the voltage profile of the first cycle and the performance of the following cycles for three Li/Gr cells using EL07, EL08, and EL09. It can be observed from
Figure 5a that when 20% FEC is used as the co-solvent (EL07), a decent SEI can be formed, which leads to a reversible capacity of 2.43 mAh cm
−2 and a CE of 81.4%. Surprisingly, further increasing the content of FEC does not improve the cells’ performances; instead, it results in a decrease in the reversible capacity and CE, as indicated by EL08 and EL09. This is probably due to the chemical equilibrium of FEC, as described by Equation (3).
When the amounts of unsaturated vinylene carbonate (VC) and acidic HF, formed by Equation (2), become excessive, the excess VC and HF hurt the performances of Li/Gr cells. After the formation of the SEI, all three cells remain very stable in capacity retention (
Figure 5b). However, their capacities are largely different and far lower than the designed capacity (2.6 mAh cm
−2). In other words, the capacity loss resulting from the formation of the SEI is permanent, which suggests that some graphite structure must have been exfoliated even if the DME electrolytes contain as much as 20–60% FEC as the co-solvent.
Up to this point, it may be concluded that even when used with LiFSI salt and FEC as the co-solvent, neither DME, DOL, nor their mixtures constitute a suitable electrolyte for high-voltage Li-ion batteries, either due to the inability of DME to form a robust SEI with graphite anodes or due to the oxidative instability of DOL against the oxygen released from delithiated layer-structured cathode materials. It should be mentioned that the electrochemical polymerization of DOL at high potentials, as proposed by La Monaca et al. [
10], would not be a problem for the Li-ion batteries because the resultant polymer can serve as a polymer matrix to form a gel polymer electrolyte, as long as DOL is used together with the other co-solvents.
3.3. Effect of FEC on DME/DOL Electrolyte
Thanks to the synergistic effect of DME and DOL, which stabilizes DME at high potentials, using 30% FEC as the co-solvent can make a 1:1 DME/DOL mixture compatible with high-voltage Gr/NCA Li-ion chemistry.
Figure 6 compares the voltage profile of the first cycle and the cycling performance of the subsequent fast-charging cycles for two Gr/NCA pouch cells using EL01 and EL10, in which EL01 serves as the baseline and EL10 contains a total of 70% 1:1 DME/DOL mixture. In the first cycle (
Figure 6a), the EL10 cell shows 187 mAh of reversible capacity and a CE of 78.4%, both of which are lower than those of the EL01 cell. However, these two cells exhibit similar capacity retention in a 4C12 fast-charging test (
Figure 6b). It is observed that in the several initial cycles of fast charging (as marked by the shaded region in
Figure 6b), the EL01 cell suffers a much greater capacity loss, compared with the EL10 cell. This is because the 1:1 DME/DOL electrolyte is more stable against Li metal that is often plated at the Gr anode in fast charging, as compared with the 3:7 EC/EMC electrolyte (EL01). In the subsequent fast-charging cycles, the capacities of these two cells decayed at a similar rate despite the low capacities of the EL10 cell. The low capacities of the EL10 cell are due to its low CE in the first cycle, which irreversibly consumes some of the cyclable Li
+ ions from the limited amount of the cathode material.
3.5. Role of DME and DOL Solvents in Fast Charging
AC impedance was studied to unveil the role of ether solvents in enhancing the fast-charging capability of Li-ion batteries.
Figure 8a displays overviews of the impedance spectroscopies at selected SOCs for two Gr/NCA pouch cells using EL01 and EL12. Ascribing to their specific geometry, the pouch cells’ impedance spectroscopies consistently show a long inductance loop at high frequencies, followed by two overlapped semi-circuits, and a sloped straight line. The main body of such impedance spectroscopies can be fitted by an equivalent circuit, as shown in the inset of
Figure 8a, where R
b, R
sl, and R
ct are the bulk resistance, surface layer (or SEI) resistance, and charge transfer resistance, respectively. Ascribed to the offset of R
sl in cycling between the cathode and the anode, R
b and R
sl are shown to be independent of the SOC; however, R
ct varies vastly with the SOC. In particular, the impedance spectroscopies at 100% SOC of these two cells are plotted together in
Figure 8b, and the resistances fit by the equivalent circuit are compared in the inset of
Figure 8b. It is observed that these two cells have very similar R
b and R
sl; however, the R
ct associated with EL12 is much smaller than that associated with EL01 (i.e., 0.260 Ω vs. 0.422 Ω). These results reveal that the DOL and DME solvents do not alter the ionic conductivity of the electrolyte and the SEI; instead, their presence greatly reduces the charge transfer resistance. This suggests that by altering the solvation shell structure of Li
+ ions, the presence of ether solvents lowers the solvation and desolvation activation energies of the Li
+ ions, which consequently enhances the electrode reaction kinetics at the Gr–electrolyte and NCA–electrolyte interfaces, respectively.