The IL materials were successfully synthesized with purity level overcoming 99.9 wt.%, i.e., particularly, the bromide, moisture, and lithium contents were found to be below 5 ppm. In total, four ionic liquids were prepared and investigated as electrolyte components: tetra-butyl-phosphonium (trifluoromethylsulfonyl)(nonafluorobutylsulfonyl)imide (P
4444IM
14), tetra-butyl-phosphonium bis(trifluoromethylsulfonyl)imide (P
4444TFSI), tetra-butyl-phosphonium (fluorosulfonyl)(trifluoromethylsulfonyl)imide (P
4444FTFSI), and 1-ethyl-3-methyl-imidazolium bis(trifluoromethylsulfonyl)imide (EMITFSI). The (IM
14)
- and (FTFSI)
- anions were selected because of their asymmetric structure with the aim of lowering the IL melting point [
8,
9]. The P
4444TFSI and P
4444FTFSI samples are solid at room temperature.
Figure 1 shows a picture of the pure P
4444TFSI sample (left panel) and its electrolyte mixture with LiTFSI (right panel, IL:LiTFSI mole ratio = 4:1).
2.1. Thermal Properties
Figure 2 compares the DSC (Differential Scanning Calorimetry) trace of the pristine IL materials and of their electrolyte mixtures with the LiTFSI salt. The pure ionic liquid samples show a well-defined melting feature (labelled with an asterisk in
Figure 2) whereas additional endothermic peaks (not evidenced in the thermal trace of EMITFSI) are ascribable to solid–solid phase transitions prior to melting the IL sample, likely due to internal structural rearrangements [
8,
9], which seem to depend on the anion asymmetry. For instance, the P
4444TFSI sample (i.e., containing the symmetric TFSI anion) shows three peaks (located at −70, 28, and 65 °C, respectively) whereas two features are detected in the DSC trace of P
4444FTFSI (around −28 and 12 °C, respectively) and P
4444IM
14 (around −35.9 and 7.0 °C, respectively), which house an asymmetric anion (in particular, IM
14), prior to melting.
The (P
4444)
+-based ionic liquids displays higher fusion temperature than EMITFSI (i.e., the melting peak onset is located around −7.5 °C), likely attributed to the large steric hindrance of the tetra-butyl-phosphonium cation that enhances the van der Waals interactions with the anion [
8]. For instance, P
4444TFSI, P
4444FTFSI, and P
4444IM
14 exhibit melting feature onsets located at 84.5, 65, and 21.8 °C, respectively, i.e., the melting feature is seen to progressively shift down to lower temperatures when passing from P
4444TFSI to P
4444IM
14, likely due to increasing asymmetry of the anion, which leads to more and more unfavourable ion packing [
8]. Conversely, the highest melting point is displayed from the P
4444TFSI sample attributed to the symmetric structure of both cation and anion, resulting in easier ion packing and, therefore, remarkable ion lattice energy value [
8].
The addition of the LiTFSI salt (IL:LiTFSI mole ratio = 4:1) to the investigated ionic liquids results in different behaviour. For instance, the 0.8P
4444TFSI-0.2LiTFSI and 0.8P
4444FTFSI-0.2LiTFSI electrolytes exhibit a shift in the melting feature, similarly to the solid–solid phase transition profiles, towards lower temperatures (with respect to the neat IL material, as depicted in
Figure 2), likely ascribable to the remarkably reduced steric hindrance of Li
+ with respect to the IL cation, which hinders the crystal lattice formation [
8]. In most of the IL families, this effect overcomes the increase in the cation–anion interactions, due to the higher surface charge density of the smaller lithium cation, resulting in melting temperature decrease [
8]. It is also interesting to note that the endothermal features, i.e., ascribed to both melting and solid–solid transitions, appear less split with respect to those observed in the DSC trace of the neat ILs. Conversely, the 0.8P
4444IM
14-0.2LiTFSI and 0.8EMITFSI-0.2LiTFSI samples show no feature apart from the glass transition profile. For instance, no crystallization (endothermal) peak was evidenced even under slow cooling or repeated low temperature thermal cycling. This unexpected behaviour, however, recorded in other IL materials [
17] is likely ascribable to much slower crystallization kinetics of the ionic liquid in the presence of the LiTFSI salt.
2.2. Ion Transport Properties
The ion transport properties of the neat ionic liquid materials (open squares) and their electrolyte mixtures (full squares) with the LiTFSI salt are reported in
Figure 3, in terms of ionic conductivity vs. temperature dependence, as Arrhenius plots.
Conduction values lower than 10
−8 S cm
−1, i.e., not detectable through the used equipment, were not reported. At lower temperature, the pure (P
4444)
+ IL samples show a marked progressive conductivity increase prior to melting (less evidenced in EMITFSI), suggesting that the ions are able to move even if the samples are still in solid phase (e.g., as a result of substantial gained ion mobility prior to material being fully molten). This behaviour agrees with the DSC results (
Figure 2), which display low temperature solid–solid phase transitions. The solid–liquid transition is revealed by a step conductivity jump up to four orders of magnitude, once more in good agreement with the thermal measurements. For instance,
Figure 4 compares the DSC trace and the conductivity vs. temperature dependence of the neat ionic liquid materials. It is worthy noting the correspondence of the step conduction rises with the endothermal melting feature and the progressive conductivity increases, detected prior to melting, with the solid–solid phase transition peaks.
As clearly evidenced, the endothermal feature attributed to the melting of the IL sample occurs just in correspondence of the conductivity step rise, which can be ascribable to the solid–liquid phase transition. Such a behaviour was also observed in the other ILs investigated, even in the presence of the LiTFSI salt. However, despite a well-evidenced conduction jump, the 0.8EMITFSI-0.2LiTFSI and 0.8P
4444IM
14-0.2LiTFSI electrolytes did not show any feature except for the glass transition profile (
Figure 2). This apparent discrepancy, previously observed in other IL materials [
17], can be addressed to the different protocols followed for conductivity measurements (i.e., preliminary treatment in liquid nitrogen, overnight sample hosting at −40 °C, rough cell electrodes, much larger sample mass) with respect to that adopted for the DSC ones (see the
Section 3), which allowed for full crystallization of both the neat IL and the IL-LiTFSI electrolyte mixtures. Further rise in temperature above the melting point results in a more moderate increase in conductivity, which exhibits a VTF behaviour previously encountered in other IL materials [
8,
17] and typical of amorphous electrolytes [
18].
The incorporation of LiTFSI leads to, even if modest, conductivity decay (i.e., reported in the literature for various IL electrolyte families) in the molten state ascribable to the higher surface charge density of Li
+ with respect to the IL cation, thus increasing the ion–ion interactions and, therefore, the viscosity of the IL material. More interestingly, the presence of the lithium salt is found to remarkably lower the melting temperature of the ionic liquid sample. For the sake of truth, this behaviour was previously observed in several ionic liquid typologies [
8], however no IL class until now investigated has exhibited melting point decay exceeding 40 °C (0.8P
4444TFSI-0.2LiTFSI) or even 60 °C (0.8P
4444FTFSI-0.2LiTFSI). For instance, the P
4444TFSI ionic liquid is solid at room temperature whereas the 0.8P
4444TFSI-0.2LiTFSI electrolyte mixture is in the molten state (
Figure 1). Such behaviour, in good agreement with the DSC results in
Figure 2, has to be attributed (as discussed above) to the much more reduced steric hindrance of the Li
+ cation with respect to the (P
4444)
+ one, which strongly hinders the formation of the IL crystal lattice (i.e., largely overcoming the increase in the ion–ion interactions due to the higher surface charge density of Li
+ with respect to the IL cation) and, therefore, leading to fusion point decay. This remarkably extends the operative temperature range of the investigated ionic liquids.
Table 1 summarizes the conductivity values, recorded at selected temperatures, for both the neat IL materials and the 0.8IL-0.2LiTFSI electrolyte mixtures. As also evidenced from
Figure 3, the (EMI)
+ formulation exhibits the highest ion conduction in the whole temperature range investigated in conjunction with the widest molten state interval. This is mainly due to the reduced steric hindrance of the imidazolium cation, with respect to the (P
4444)
+ one, and its quasi-planar structure, which allows for faster mobility [
8]. For instance, a conductivity exceeding 10
−4 S cm
−1 is recorded already at −20 °C, making the 0.8EMITFSI-0.2LiTFSI formulation potentially appealing also for low temperature applications, whereas 10
−3 S cm
−1 and 10
−2 S cm
−1 are overcome at 10 °C and 80 °C, respectively. The better ion transport properties of the 0.8EMITFSI-0.2LiTFSI electrolyte are also witnessed by its lower glass transition temperature (around −86 °C), as shown in
Figure 2. Below 0 °C, the 0.8P
4444IM
14 formulation displays the highest conductivity values with respect to the other tetra-butyl-phosphonium electrolytes, likely attributed to the high asymmetry of the (IM
14)
- anion, which hinders the ion packing and, consequently, lowers the melting temperature of the IL sample [
8]. Conversely, the 0.8P
4444TFSI-0.2LiTFSI and 0.8P
4444FTFSI-0.2LiTFSI formulations show larger conductivity in the molten state, as the lower steric hindrance of the (TFSI)
- and (FTFSI)
- anions with respect to (IM
14)
- results in enhanced ion mobility. For instance, conductivity values approaching 2 mS cm
−1 and overcoming 4 mS cm
−1 are recorded at 70 °C and 100 °C, i.e., more than two and three times higher, respectively, with respect to the 0.8P
4444TFSI-0.2IM
14 formulation.
2.3. Thermal Stability
Strong thermal robustness is a crucial property in view of application in high-temperature devices.
Figure 5 reports on the variable-temperature TGA (Thermo-Gravimetrical Analysis) traces of the ionic liquid materials investigated, as measured in nitrogen atmosphere. The EMITFSI material exhibits no appreciable weight loss up to 300 °C, i.e., about 20 °C (P
4444TFSI and P
4444FTFSI) and 50 °C (P
4444IM
14) higher with respect to the (P
4444)
+-based samples. At higher temperatures, degradation of the IL sample takes place, resulting in almost full weight loss (i.e., suggesting complete volatilization of the investigated ILs) in a temperature range from 300 to 400–500 °C.
To investigate the thermal robustness under hard conditions, isothermal step TGA tests were also carried out. The results, depicted in
Figure 6, indicate remarkable lower stability with respect to the data obtained from variable-temperature measurements (
Figure 5). This behaviour, previously observed in several IL families [
8,
9,
17], supports the better reliability of the isothermal tests. For instance, the (P
4444)
+-containing materials start to exhibit a minimal weight loss (however, below 0.4%) above 175 °C, whereas no appreciable weight variation is observed for EMITFSI up to 200 °C. The nature of the anion (i.e., at least, the typologies investigated in the present work) does not seem to remarkably affect the thermal stability of the tetra-butyl-phosphonium ionic liquids. To summarize, the reported results clearly highlight an excellent thermal robustness up to 150 °C for the investigated ionic liquids.
2.4. Electrochemical Stability
Electrochemical stability is one of the most important requirements for electrolyte formulations, especially for those addressed to devices operating at high temperatures and high voltages. This topic peculiarity was evaluated through repeated anodic cyclic voltammetry (CV) tests performed on Li/0.8IL-0.2LiTFSI/C cells at 110 °C. The CV profiles, depicted in
Figure 7, reveal relatively low current densities (normalized with respect to the geometrical area of the carbon working electrode) in the first anodic scan (dotted trace). However, a remarkable current decay is recorded in the following cycles (solid traces) in combination with the absence of cathodic features. This indicates that the irreversible oxidation processes (occurring during the first anodic sweep) can be related to contaminants rather than to the ionic liquid materials and/or to the LiTFSI salt.
Table 2 compares the voltage at which the current flowing through the cell achieves 10 (V
10) and 20 (V
20) μA cm
−2 during the third anodic scan, with the value (anodic limit voltage, V
L(An)) obtained by fitting the step raise region of the CV traces (third anodic sweep) vs. the X axis. As clearly evidenced from
Table 2, the 20 μA cm
−2 threshold is achieved above 4.6 V for the 0.8P
4444FTFSI-0.2LiTFSI electrolyte and is overcome above 4.3 V and 4.1 V for the 0.8P
4444IM
14-0.2LiTFSI and 0.8P
4444TFSI-0.2LiTFSI formulations, respectively. Conversely, the 0.8EMITFSI-0.2LiTFSI electrolyte matches a current density of 20 μA cm
−2 already at 3.06 V. The 0.8P
4444FTFSI-0.2LiTFSI sample exhibits a current value even lower than 10 μA cm
−2 (V
10) up to 4.5 V, whereas this limit is reached above 4 V only by the 0.8P
4444IM
14-0.2LiTFSI formulation. Therefore, the anodic electrochemical robustness of the investigated 0.8IL-0.2LiTFSI electrolytes is seen to mainly depend on the nature of the anion [
8,
9].
From a more accurate examination of the data reported in
Figure 7 and
Table 2, it appears evident that the nature of the cation can also play a role in determining the oxidation stability. For instance, the 0.8P
4444TFSI-0.2LiTFSI electrolyte exhibits higher electrochemical robustness than 0.8EMITFSI-0.2LiTFSI, likely highlighting the better stability of the phosphonium cation with respect to the imidazolium one and confirming the influence of the cation on IL oxidation processes. For instance, the electron-donor effect of the
n-butyl chain [
19] could stabilize the positive charge localized onto the P atom of the (P
4444)
+ cation; however, further investigation (out of the purpose of the present work) should be carried out for understanding this behaviour.
Finally, the anodic limit voltage (V
L(An) values in
Table 2) was determined through fitting of the step current riise region of the CV traces (third anodic sweep of
Figure 7) vs. the X axis. It should be noted, as clearly reported in
Table 2, that the V
L(An) parameter always largely exceeds 4.6 V, i.e., from about 0.1 (0.8P
4444FTFSI-0.2LiTFSI electrolyte) to 1.6 V (0.8EMITFSI-0.2LiTFSI) higher with respect to the V
10 and V
20 values (
Table 2). However, the corresponding current flow recorded at the V
L(An) voltage ranges from 60 (0.8P
4444FTFSI-0.2LiTFSI) to 120 (0.8P
4444TFSI-0.2LiTFSI) μA cm
−2, i.e., the oxidation of the electrolyte sample is not fully negligible. Therefore, this approach, even if widely adopted in the literature [
8,
9] for estimating the electrochemical stability of electrolytes, could not provide an optimal evaluation. Overall, at temperatures overcoming 100 °C, the (P
4444)
+-based electrolyte formulations show electrochemical stabilities against oxidation well above 4.5 V, whereas the EMITFSI electrolyte is electrochemically stable only up to 3 V.
2.5. Preliminary Tests on Battery
Preliminary charge–discharge cycling tests were performed at 100 °C on Li/LiFePO
4 cells (operating up to 4 V) to check the feasibility of the developed ionic liquid electrolyte technologies for devices operating at high temperatures. The 0.8P
4444TFSI-0.2LiTFSI formulation was selected as the electrolyte with the aim to validate the possibility to use solid ILs at room temperature. The results are plotted in
Figure 8 as the voltage vs. capacity profile recorded at different current rates. A flat plateau (i.e., maintaining the same voltage during almost the entire charge/discharge step), typical of the Li+ insertion/de-insertion process into the LiFePO
4 active material [
20], is observed (in the 3.0–3.3 V range) at 0.1C (black traces), with very good reproducibility in the following cycles (i.e., the voltage profiles are practically overlapped). The increase in the current rate up to 0.5C results in a progressively higher slope of the voltage profile as well as increasing ohmic drop (as expected) likely due to enhanced diffusive phenomena within the electrolyte. A nominal capacity close to the theoretical value (170 mA h g
−1) [
20] is delivered at 0.1C, highlighting the good behaviour of the 0.8P
4444TFSI-0.2LiTFSI electrolyte operating at 100 °C with LiFePO
4 cathodes. Interestingly, a capacity of 160 mA h g
−1 (>94% of the theoretical value) is still discharged at 0.5C, suggesting good capability up to medium current rates and good compatibility at the electrolyte/electrode interface.
These results, even if preliminary, give clear indication about promising performance of the investigated electrolyte technologies for difficult operating conditions (high operative temperature/voltage, cycling tests run at 100% of deep discharge) without any apparent degradation. Of course, further optimization is required at the level of electrode formulation and cell design. An extended investigation on the cycling behaviour in batteries for these ionic liquid electrolyte technologies will be the subject of a future paper.