Effect of Zwitterionic Additive on Electrode Protection through Electrochemical Performances of Anatase TiO2 Nanotube Array Electrode in Ionic Liquid Electrolyte

In this work, a functionalized zwitterionic (ZI) compound 1-butylsulfonate-3-methylimidazole (C1C4imSO3) was synthesized and tested as an additive to LiTFSI/C2C2imTFSI ionic liquid-based electrolytes for lithium-ion batteries. The structure and purity of C1C4imSO3 were confirmed by NMR and FTIR spectroscopy. The thermal stability of the pure C1C4imSO3 was examined by simultaneous thermogravimetric–mass spectrometric (TG–MS) measurements and differential scanning calorimetry (DSC). The LiTFSI/C2C2imTFSI/C1C4imSO3 system was tested as a potential electrolyte for lithium-ion batteries by using anatase TiO2 nanotube array electrode as the anode material. This electrolyte with 3% C1C4imSO3 showed significant improvement of lithium-ion intercalation/deintercalation properties, such as capacity retention and Coulombic efficiency compared to electrolyte without additive.


Introduction
Lithium-ion batteries (LIBs) are becoming increasingly essential for electrifying the transportation systems for sustainable mobility in the future. However, large-scale application of lithium-ion batteries is limited by serious safety concerns when the LIBs are exposed to thermal, mechanical, or electrical abuse conditions [1,2]. Their high energy density associated with the use of volatile, flammable electrolytes based on organic solvents creates safety risks [3,4]. Since the electrolyte is the component that connects all parts, it notably impacts a lot of technological and chemical aspects of LIBs.
To overcome the safety concerns related to interactions between conventionally used electrolytes and electrode materials, researchers have developed functionalized electrolyte systems in combination with robust electrodes. Electrode materials that enhance the battery's capacity and energy density could be TiO 2 nanotube arrays (NTAs) [16][17][18][19][20][21][22][23][24][25]. TiO 2 nanotubes are specific materials, which can combine with current collector and electrode material. TiO 2 is a robust and resistant material, allowing on the use of higher battery's capacity and energy density could be TiO2 nanotube arrays (NTAs) [16][17][18][19][20][21][22][23][24][25]. TiO2 nanotubes are specific materials, which can combine with current collector and electrode material. TiO2 is a robust and resistant material, allowing on the use of higher temperatures because there is no possibility of removing materials from the electrodes at higher temperatures [16][17][18][19][20]. In addition, such NTA electrodes enable immediate, direct recording of Raman and Fourier Transform Infrared (FTIR) spectra on their surface.
Current demands for LIB electrolytes include developing electrolytes containing components that synchronously achieve different functions, such as electrolyte stabilization towards the electrode materials and electrode functionalization [26][27][28][29][30]. As a result, the knowledge based on ionic liquids induced the synthesis of novel families of compounds established on them-the zwitterionic compounds, because it is assumed that there is a possibility of a stabilization effect towards the electrode [31][32][33][34][35][36][37][38]. One of the imperatives of our research is thermal and safety improvement of LIBs. This research study examined the electrochemical properties of 0.5 mol·dm −3 LiTFSI (Figure 1a) in ionic liquid 1,3-diethylimidazolium bis(trifluoromethylsulfonyl)imide, C2C2imTFSI (Figure 1b) with 3% a functionalized additive (zwitterionic compound 1-butylsulfonate-3-methylimidazole C1C4imSO3; ZI, Figure 1c), along with the performance of anatase Ti/TiO2 nanotube array (NTA) cells. For this purpose, the zwitterionic compound C1C4imSO3 was synthetized in his work, as shown in Figure 1d. The electrochemical testing was conducted to compare the stability of the electrolyte studied in this work with that in a previous study [28], which did not contained the zwitterionic compound C1C4imSO3 as an additive. In comparison with their IL counterparts, the present ZI had a more polar character, suggesting that this compound may become attractive in battery electrolytes. A comprehensive consideration of the ZI compound's electrochemical behavior and mechanism is needed for deeper thermochemical and physicochemical profiling.

Thermogravimetric and DSC Analysis
The thermal properties of pure synthesized zwitterionic salt C1C4imSO3 were determined by thermogravimetric (TG) and differential scanning calorimetry (DSC) measurement. The obtained TG and DSC decomposition curves of C1C4imSO3 are shown in Figure  2. The electrochemical testing was conducted to compare the stability of the electrolyte studied in this work with that in a previous study [28], which did not contained the zwitterionic compound C 1 C 4 imSO 3 as an additive. In comparison with their IL counterparts, the present ZI had a more polar character, suggesting that this compound may become attractive in battery electrolytes. A comprehensive consideration of the ZI compound's electrochemical behavior and mechanism is needed for deeper thermochemical and physicochemical profiling.

Thermogravimetric and DSC Analysis
The thermal properties of pure synthesized zwitterionic salt C 1 C 4 imSO 3 were determined by thermogravimetric (TG) and differential scanning calorimetry (DSC) measurement. The obtained TG and DSC decomposition curves of C 1 C 4 imSO 3 are shown in Figure 2.
From the presented thermogravimetric curve (Figure 2, black line), the synthesized C 1 C 4 imSO 3 was stable up to 330 • C, implying its high thermal stability. The TGA thermograms in Figure 2 did not show any significant weight loss near 100 • C, demonstrating that this zwitterion is not hygroscopic. From TG-DSC measurements, we observed the softening of the C 1 C 4 imSO 3 compound with loss of weight and accompanied with an endothermic peak in the simultaneous DSC curve (Figure 2, blue line), demonstrating the melting and decomposition process [20,21]. The DSC curve ( Figure 2, blue line) showed one sharp peak around T melt = 236 • C. The second DSC peak corresponds to the complete decomposition of C 1 C 4 imSO 3 (around T = 365 • C). The comparison of the T melt to that of similar zwitterionic compounds, regarding the length of the spacer between the cation and anion group, was in accordance with other research [31]. Comparison of the sulfonate anion and imidazole cation part with other compounds presented in the work of Yoshizava et al. [31] and Galin et al. [36], where the only difference in length is in the molecule part that connects them, suggests that longer spacer indicate lower melting points. This is probably because of the flexibility increase with increasing spacer length. Figure 3 illustrates the DSC thermogram of the synthesized C 1 C 4 imSO 3 compound in an extended lower temperature range. From the presented thermogravimetric curve ( Figure 2, black line), the synthesized C1C4imSO3 was stable up to 330 °C, implying its high thermal stability. The TGA thermograms in Figure 2 did not show any significant weight loss near 100 °C, demonstrating that this zwitterion is not hygroscopic. From TG-DSC measurements, we observed the softening of the C1C4imSO3 compound with loss of weight and accompanied with an endothermic peak in the simultaneous DSC curve ( Figure 2, blue line), demonstrating the melting and decomposition process [20,21]. The DSC curve ( Figure 2, blue line) showed one sharp peak around Tmelt = 236 °C. The second DSC peak corresponds to the complete decomposition of C1C4imSO3 (around T = 365 °C). The comparison of the Tmelt to that of similar zwitterionic compounds, regarding the length of the spacer between the cation and anion group, was in accordance with other research [31]. Comparison of the sulfonate anion and imidazole cation part with other compounds presented in the work of Yoshizava et al. [31] and Galin et al. [36], where the only difference in length is in the molecule part that connects them, suggests that longer spacer indicate lower melting points. This is probably because of the flexibility increase with increasing spacer length. Figure 3 illustrates the DSC thermogram of the synthesized C1C4imSO3 compound in an extended lower temperature range.  From the presented thermogravimetric curve (Figure 2, black line), the synthesized C1C4imSO3 was stable up to 330 °C, implying its high thermal stability. The TGA thermograms in Figure 2 did not show any significant weight loss near 100 °C, demonstrating that this zwitterion is not hygroscopic. From TG-DSC measurements, we observed the softening of the C1C4imSO3 compound with loss of weight and accompanied with an endothermic peak in the simultaneous DSC curve ( Figure 2, blue line), demonstrating the melting and decomposition process [20,21]. The DSC curve ( Figure 2, blue line) showed one sharp peak around Tmelt = 236 °C. The second DSC peak corresponds to the complete decomposition of C1C4imSO3 (around T = 365 °C). The comparison of the Tmelt to that of similar zwitterionic compounds, regarding the length of the spacer between the cation and anion group, was in accordance with other research [31]. Comparison of the sulfonate anion and imidazole cation part with other compounds presented in the work of Yoshizava et al. [31] and Galin et al. [36], where the only difference in length is in the molecule part that connects them, suggests that longer spacer indicate lower melting points. This is probably because of the flexibility increase with increasing spacer length. Figure 3 illustrates the DSC thermogram of the synthesized C1C4imSO3 compound in an extended lower temperature range.  The obtained solid compound exhibited a prominent melting point upon heating (T m = 236 • C) as the top of an endothermic peak. In the literature, zwitterions which have sulfonate with a longer spacer between the cation and anion show lower T m values at the range of spacer length studied [31]. The DSC data showed that the zwitterionic compound based on imidazolium cations has a significantly higher melting point than those of common imidazolium-based ILs. In this zwitterionic structure, the cation and anion are tethered, and therefore contribute to restricting the molecule's vibrational and rotational motions [31]. Therefore, the negative charge is delocalized by the sulfonate group electron-withdrawing effect, and the interaction force between the anion and cation is lower than in the case of sulfonate anions [31,34].

Thermogravimetry Coupled with Mass Spectrometric (TG-MS) Analysis
We performed additional thermal characterizations of the synthetized C 1 C 4 imSO 3 using thermogravimetry coupled with the mass spectrometry (TG-MS) because it does not exist in the research literature. During the entire process of degradation, the evolution of the essential fragment products were collected by the TG-MS technique. From Figure 4 we can see the thermochemical decomposition in one weight loss step, the first derivative curve with time (DTG), as well as the TG-MS fragment ion intensities for formed products. The temperature interval from 320 to 390 • C mostly shows the mass loss of C 1 C 4 imSO 3 .
The high intensity of the MS peaks for ions with m/z = 18 (H 2 O) and m/z = 17 (OH) are shown as sharp DTG peaks at 380 • C. The intensity ratio of fragments with m/z = 18 (H 2 O): m/z = 17 (OH) = 5:1, is in accordance with the National Institute of Standards and Technology (NIST) mass spectral data [39][40][41][42][43], that is attributed to these fragments. As Figure 4a implies, a relatively large amount of H 2 O (m/z = 18) and OH (m/z = 17) was formed during the recombination of the atoms and heating, due to the oxygen atoms from the structure of C 1 C 4 imSO 3 .
Characteristic for this thermal decomposition is the departure of products with fragments m/z = 28 (CO) and 32 (O 2 ). They are frequently formed from the removal of the carbon and its recombination with oxygen leaving CO m/z = 28. That process occurs at 375 • C, as seen in Figure 4b. The removal of SO and SO 2 from C 1 C 4 imSO 3 released during heating (T = 370 • C) is indicated with the (m/z = 48), and SO 2 (m/z = 48) is released at T = 375 • C. The m/z = 48 and m/z = 64 indicate that two overlapping steps occurred in the range T = 370-375 • C and were difficult to distinguish (Figure 4c).
The products with fragments m/z = 14 and 16 are ingrained in this peak at 370 • C: m/z = 14 (N) and m/z = 16 had a similar shape (NH 2 ) with an equally low magnitude ( Figure 4d). The m/z = 14 and m/z = 16 fragments started around 375 • C. The C-C and C-H bonds broke to form free radicals, and recombined into small fragments with the increase in temperature. Accordingly, the most significant fraction of m/z = 36 can be attributed to C-C-C (C 3 ) from a side chain attached to the imidazole ring of C 1 C 4 imSO 3 . With the increase in temperature, the C-C and C-H bonds broke to form free radicals, which are recombined into more considerable fragments as m/z = 44 (CH 3 -CH 2 -CH 3 ), that were detected with lower intensities (Figure 4e).
The obtained solid compound exhibited a prominent melting point upon heating (Tm = 236 °C) as the top of an endothermic peak. In the literature, zwitterions which have sulfonate with a longer spacer between the cation and anion show lower Tm values at the range of spacer length studied [31]. The DSC data showed that the zwitterionic compound based on imidazolium cations has a significantly higher melting point than those of common imidazolium-based ILs. In this zwitterionic structure, the cation and anion are tethered, and therefore contribute to restricting the molecule's vibrational and rotational motions [31]. Therefore, the negative charge is delocalized by the sulfonate group electronwithdrawing effect, and the interaction force between the anion and cation is lower than in the case of sulfonate anions [31,34].

Thermogravimetry Coupled with Mass Spectrometric (TG-MS) Analysis
We performed additional thermal characterizations of the synthetized C1C4imSO3 using thermogravimetry coupled with the mass spectrometry (TG-MS) because it does not exist in the research literature. During the entire process of degradation, the evolution of the essential fragment products were collected by the TG-MS technique. From Figure 4 we can see the thermochemical decomposition in one weight loss step, the first derivative curve with time (DTG), as well as the TG-MS fragment ion intensities for formed products. The temperature interval from 320 to 390 °C mostly shows the mass loss of C1C4imSO3.  [39][40][41][42][43], that is attributed to these fragments. As Figure 4a implies, a relatively large amount of H2O (m/z = 18) and OH (m/z = 17) was formed during the recombination of the atoms and heating, due to the oxygen atoms from the structure of C1C4imSO3.
Characteristic for this thermal decomposition is the departure of products with fragments m/z = 28 (CO) and 32 (O2). They are frequently formed from the removal of the carbon and its recombination with oxygen leaving CO m/z = 28. That process occurs at 375 °C, as seen in Figure 4b. The removal of SO and SO2 from C1C4imSO3 released during heating (T = 370 °C) is indicated with the (m/z = 48), and SO2 (m/z = 48) is released at T = 375 °C. The m/z = 48 and m/z = 64 indicate that two overlapping steps occurred in the range T = 370-375 °C and were difficult to distinguish (Figure 4c).
The products with fragments m/z = 14 and 16 are ingrained in this peak at 370 °C: m/z = 14 (N) and m/z = 16 had a similar shape (NH2) with an equally low magnitude ( Figure  4d). The m/z = 14 and m/z = 16 fragments started around 375 °C. The C-C and C-H bonds broke to form free radicals, and recombined into small fragments with the increase in temperature. Accordingly, the most significant fraction of m/z = 36 can be attributed to C-C-C (C3) from a side chain attached to the imidazole ring of C1C4imSO3. With the increase in temperature, the C-C and C-H bonds broke to form free radicals, which are recombined into more considerable fragments as m/z = 44 (CH3-CH2-CH3), that were detected with lower intensities (Figure 4e).
With further heating (T = 407 °C), one of the highest intensities for the sample was released and correlates to the removal of H from C1C4imSO3 (m/z = 1; Table S1). Table S1 summarizes the identification results and normalized ion currents for the thermal decomposition of the pure compound in this work.

Flammability of Electrolyte
Regarding the electrolyte safety aspect, it is essential to estimate the flammability of 0.5 mol·dm −3 LiTFSI/C2C2imTFSI/C1C4imSO3 electrolyte. A flame did not appear during 120 s of heating, which indicates that the LiTFSI/C2C2imTFSI/C1C4imSO3 electrolyte is With further heating (T = 407 • C), one of the highest intensities for the sample was released and correlates to the removal of H from C 1 C 4 imSO 3 (m/z = 1; Table S1). Table  S1 summarizes the identification results and normalized ion currents for the thermal decomposition of the pure compound in this work.

Flammability of Electrolyte
Regarding the electrolyte safety aspect, it is essential to estimate the flammability of 0.5 mol·dm −3 LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 electrolyte. A flame did not appear during 120 s of heating, which indicates that the LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 electrolyte is nonflammable ( Figure S1). The flammability of the electrolyte with same concentration of lithium salt in ionic liquid electrolyte was investigated in a previous publication [28].

Electrochemical Characterization 2.4.1. Galvanostatic Cycling
The discharge/charge capacities during Li + -ion insertion/deinsertion into/from TiO 2 NTAs for electrolyte LiTFSI/C 2 C 2 imTFSI and for the same electrolyte with the zwitterion additive LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 , were obtained at a current density of 50 µA/cm 2 and are shown in Figure 5a-c, respectively. For the LiTFSI/C 2 C 2 imTFSI electrolyte without additive, the initial discharge/charge capacity of the Ti/TiO 2 NTA electrode was 370/303 mAh·g −1 . After seven cycles and increase in the charge capacity, both the discharge and charge capacities decreased duslightring further cycling, (Figure 5a). Although the capacity decreased with the increase in the number of GS cycles, the stabilization of the capacity was not achieved after 100 cycles, when the discharge/charge capacity was 228.0/217.3 mAh·g −1 . The Coulombic efficiency increase during cycling and amounted 93.7% after 50 cycles, 94.6% after 75 cycles, and 95.3% after 100 cycles. The electrolyte at the beginning was colorless, but after 100 cycles it became a pale yellow, which shows the presence of electrolyte decomposition (Figure 5c). The increase of Coulombic efficiency as the number of cycles increased, indicates that the electrolyte decomposition process was more pronounced at the beginning of cycling, but continued during further GS cycling.
93.7% after 50 cycles, 94.6% after 75 cycles, and 95.3% after 100 cycles. The electrolyte at the beginning was colorless, but after 100 cycles it became a pale yellow, which shows the presence of electrolyte decomposition (Figure 5c). The increase of Coulombic efficiency as the number of cycles increased, indicates that the electrolyte decomposition process was more pronounced at the beginning of cycling, but continued during further GS cycling. In the case of the electrolyte with additive, LiTFSI/C2C2imTFSI/C1C4imSO3, the initial discharge capacity of the Ti/TiO2 NTAs electrode was 325.4 mAh·g −1 , and the charge capacity was significantly lower, at 231.8 mAh·g −1 (Figure 5b). During the initial 14 cycles, the charge capacity increased slightly. Then, both discharge and charge capacities decreased until the 43 rd cycle, when the decrease of both capacities stopped and their values became almost identical. Further GS cycling even led to a slight increase in capacity. The Coulombic efficiency was 99.0% after 43 cycles, 99.2% after 50 cycles, 99.5% after 75 cycles, and 99.6%after 100 cycles. After 100 cycles the final discharge/charge capacity was 249.4/250.5 mAh·g −1 and the electrolyte remained colorless. The presence of the zwitterion additive C1C4imSO3 led to the formation of a film on the working electrode, with this process being completed in ~40 cycles. This prevented further decomposition of the electrolyte and allowed the Li + -ion insertion/deinsertion process to occur with high Coulombic efficiency, and without capacity fade during further cycling. The comparison of the capacity of the Ti/TiO2 NTAs electrode in the electrolyte with and without additive can be seen in Figure 5c.
The voltage profiles, for the 100th GS cycle, during lithiation/delithiation of the Ti/TiO2 NTA electrode in the LiTFSI/C2C2imTFSI are presented in Figure 5d. The region in discharge during lithiation, where the voltage profile was relatively flat (~1.74 V in the middle part), designates the existence of two phases: the lithium poor Li0.026TiO2 and lithium rich Li0.52TiO2 phases [44][45][46]. When the transition from the first to the second phase during lithiation was completed, the inclined part of voltage profile started. This part ended at the lower cut-off voltage of 1.0 V, and can be attributed to surface storage mechanisms [45,46]. It can be seen in Figure 5d that in the case of the LiTFSI/C2C2imTFSI electrolyte, the contribution of the inclined part of the voltage profile to the overall capacity was larger from flat part, while for the electrolyte with additive, LiTFSI/C2C2imTFSI/C1C4imSO3, it is vice versa. This shows that the film formed during GS cycling at the surface of the TiO2 NTs, due to the presence of the zwitterion additive C1C4imSO3, decreased the surface storage, but has the effect of increasing the depth of lithium insertion into the TiO2 NTs. This increased of the overall capacity of the active electrode in the electrolyte with additive comparing to the electrolyte without additive, In the case of the electrolyte with additive, LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 , the initial discharge capacity of the Ti/TiO 2 NTAs electrode was 325.4 mAh·g −1 , and the charge capacity was significantly lower, at 231.8 mAh·g −1 (Figure 5b). During the initial 14 cycles, the charge capacity increased slightly. Then, both discharge and charge capacities decreased until the 43 rd cycle, when the decrease of both capacities stopped and their values became almost identical. Further GS cycling even led to a slight increase in capacity. The Coulombic efficiency was 99.0% after 43 cycles, 99.2% after 50 cycles, 99.5% after 75 cycles, and 99.6%after 100 cycles. After 100 cycles the final discharge/charge capacity was 249.4/250.5 mAh·g −1 and the electrolyte remained colorless. The presence of the zwitterion additive C 1 C 4 imSO 3 led to the formation of a film on the working electrode, with this process being completed in~40 cycles. This prevented further decomposition of the electrolyte and allowed the Li + -ion insertion/deinsertion process to occur with high Coulombic efficiency, and without capacity fade during further cycling. The comparison of the capacity of the Ti/TiO 2 NTAs electrode in the electrolyte with and without additive can be seen in Figure 5c.
The voltage profiles, for the 100th GS cycle, during lithiation/delithiation of the Ti/TiO 2 NTA electrode in the LiTFSI/C 2 C 2 imTFSI are presented in Figure 5d. The region in discharge during lithiation, where the voltage profile was relatively flat (~1.74 V in the middle part), designates the existence of two phases: the lithium poor Li 0.026 TiO 2 and lithium rich Li 0.52 TiO 2 phases [44][45][46]. When the transition from the first to the second phase during lithiation was completed, the inclined part of voltage profile started. This part ended at the lower cut-off voltage of 1.0 V, and can be attributed to surface storage mechanisms [45,46]. It can be seen in Figure 5d that in the case of the LiTFSI/C 2 C 2 imTFSI electrolyte, the contribution of the inclined part of the voltage profile to the overall capacity was larger from flat part, while for the electrolyte with additive, LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 , it is vice versa. This shows that the film formed during GS cycling at the surface of the TiO 2 NTs, due to the presence of the zwitterion additive C 1 C 4 imSO 3 , decreased the surface storage, but has the effect of increasing the depth of lithium insertion into the TiO 2 NTs. This increased of the overall capacity of the active electrode in the electrolyte with additive comparing to the electrolyte without additive, both in discharge and charge as shown in Figure 5d for the 100th cycle. The increase in overall capacity, together with the stabilization of the discharge/charge capacity with high Coulombic efficiency and without capacity fade that is shown in Figure 5b,c, ultimately led to the excellent performance of the LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 electrolyte.
After GS cycling, the UV-Vis spectra of the electrolyte solutions were recorded for both tested electrolytes, especially considering the observed change from colorless to yellow in the case of the LiTFSI/C 2 C 2 imTFSI electrolyte. The recorded UV-Vis spectra showed strong absorption by the newly formed compound(s), occurring at very low concentrations but were undetectable with UV-Vis spectroscopy ( Figure S2).
The SEM micrographs showed no significant change in the morphology of TiO 2 before and after galvanostatic cycling in the LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 electrolyte, as shown in Figure 6. both in discharge and charge as shown in Figure 5d for the 100th cycle. The increase in overall capacity, together with the stabilization of the discharge/charge capacity with high Coulombic efficiency and without capacity fade that is shown in Figure 5b and 5c, ultimately led to the excellent performance of the LiTFSI/C2C2imTFSI/C1C4imSO3 electrolyte. After GS cycling, the UV-Vis spectra of the electrolyte solutions were recorded for both tested electrolytes, especially considering the observed change from colorless to yellow in the case of the LiTFSI/C2C2imTFSI electrolyte. The recorded UV-Vis spectra showed strong absorption by the newly formed compound(s), occurring at very low concentrations but were undetectable with UV-Vis spectroscopy ( Figure  S2).
The SEM micrographs showed no significant change in the morphology of TiO2 before and after galvanostatic cycling in the LiTFSI/C2C2imTFSI/C1C4imSO3 electrolyte, as shown in Figure 6.

Vibrational Spectroscopy
The Ti/TiO2 NTA electrodes from both tested electrolytes after GS cycling were further analyzed with FTIR spectroscopy. First, the electrodes were washed with acetone to fully dissolve the ionic liquid C2C2imTFSI. The only sharp band for both electrolytes was located at ~1420 cm −1 (Figure 7a) and could be assigned to S=O stretching of the sulfate group (1415-1380 cm −1 ) rather than to the same vibration of sulfonates (1372-1335 cm −1 ) [47][48][49]. For the SO3 molecule the antisymmetric stretch was at 1391 cm −1 [50], which is also lower than the obtained value. The formation of surface SO4 2− may occur if surface SO3 2− reacts with lattice oxygen [51] when adjacent Ti 4+ is reduced to Ti 3+ during intercalation of Li + -ions. For the electrolyte with ZI additive, which has sulfonate group, such formation of surface SO4 2− seems to occur in a simpler way compared to the electrolyte without additive, where the SO2 2− group inside the TFSI − anion is in a more complex coordination. Most likely, this is the reason why the intensity of the IR band at ~1420 cm −1 was higher in the presence of the ZI additive (Figure 7a). In the IR spectra of both electrolytes, on the high frequency side of the obtained IR band a shoulder exists at ~1480 cm −1 due to the presence of a wide band of lower intensity. Considering that the low intensity of this band composed of more overlapping bands and the existence of a curved base line which all make deconvolution problematic, the assignation of this band was not performed. However, it is important to stress that its intensity follows the intensity of the band at ~1420 cm −1 .

Spectroscopy Measurements after Cycling Vibrational Spectroscopy
The Ti/TiO 2 NTA electrodes from both tested electrolytes after GS cycling were further analyzed with FTIR spectroscopy. First, the electrodes were washed with acetone to fully dissolve the ionic liquid C 2 C 2 imTFSI. The only sharp band for both electrolytes was located at~1420 cm −1 (Figure 7a) and could be assigned to S=O stretching of the sulfate group (1415-1380 cm −1 ) rather than to the same vibration of sulfonates (1372-1335 cm −1 ) [47][48][49]. For the SO 3 molecule the antisymmetric stretch was at 1391 cm −1 [50], which is also lower than the obtained value. The formation of surface SO 4 2− may occur if surface SO 3 2− reacts with lattice oxygen [51] when adjacent Ti 4+ is reduced to Ti 3+ during intercalation of Li + -ions. For the electrolyte with ZI additive, which has sulfonate group, such formation of surface SO 4 2− seems to occur in a simpler way compared to the electrolyte without additive, where the SO 2 2− group inside the TFSI − anion is in a more complex coordination. Most likely, this is the reason why the intensity of the IR band at~1420 cm −1 was higher in the presence of the ZI additive (Figure 7a). In the IR spectra of both electrolytes, on the high frequency side of the obtained IR band a shoulder exists at~1480 cm −1 due to the presence of a wide band of lower intensity. Considering that the low intensity of this band composed of more overlapping bands and the existence of a curved base line which all make deconvolution problematic, the assignation of this band was not performed. However, it is important to stress that its intensity follows the intensity of the band at 1420 cm −1 .
The IR spectra showed that during GS cycling, decomposition of the ZI additive C 1 C 4 imSO 3 and electrolyte LiTFSI/C 2 C 2 imTFSI occurred at the same surface sites of the TiO 2 NTs. In the electrolyte where additive was present, its decomposition blocked surface sites for LiTFSI/C 2 C 2 imTFSI decomposition. This stabilized the electrode capacity after a certain number of cycles and enabled a high Coulombic efficiency of the Li + -ion insertion/extraction process. Without additive, the decomposition of the LiTFSI/C 2 C 2 imTFSI electrolyte continued during all 100 cycles without final stabilization of the electrode capacity which was dropping. The Coulombic efficiency gradually increased, but did not reach the values as in the electrolyte with additive. The IR spectra of the electrodes after an additional wash with water, which removed all molecules from the surface of the TiO 2 NTs, are shown in Figure 7b. The IR spectra showed that during GS cycling, decomposition of the ZI additive C1C4imSO3 and electrolyte LiTFSI/C2C2imTFSI occurred at the same surface sites of the TiO2 NTs. In the electrolyte where additive was present, its decomposition blocked surface sites for LiTFSI/C2C2imTFSI decomposition. This stabilized the electrode capacity after a certain number of cycles and enabled a high Coulombic efficiency of the Li + -ion insertion/extraction process. Without additive, the decomposition of the LiTFSI/C2C2imTFSI electrolyte continued during all 100 cycles without final stabilization of the electrode capacity which was dropping. The Coulombic efficiency gradually increased, but did not reach the values as in the electrolyte with additive. The IR spectra of the electrodes after . FTIR spectrum of Ti/TiO 2 NTA electrodes in electrolytes LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 or LiTFSI/C 2 C 2 imTFSI after GS cycling: (a) after washing with acetone and (b) after additional wash with water.
The Raman spectra of the Ti/TiO 2 NTA electrodes from both electrolytes were also recorded. The Raman modes at 635 cm −1 (E g ), 519 cm −1 (A 1g , B 1g ), 399 cm −1 (E g ), 205 cm −1 (E g ), and 152 cm −1 confirmed the presence of the TiO 2 anatase phase for both electrodes. The low intensity Raman band at 325 cm −1 , observed in the spectrum of the electrolyte with ZI additive was of slightly better quality and was a combination band [52,53] (Figure 8).
recorded. The Raman modes at 635 cm −1 (Eg), 519 cm −1 (A1g, B1g), 399 cm −1 (Eg), 205 cm −1 (Eg), and 152 cm −1 confirmed the presence of the TiO2 anatase phase for both electrodes. The low intensity Raman band at 325 cm −1 , observed in the spectrum of the electrolyte with ZI additive was of slightly better quality and was a combination band [52,53] (Figure  8). The results imply that the sulfonate-based zwitterionic compound might facilitate Li + movement because a partial negative charge that developed at the end of the sulfonate SO3 − group is able to bind with Li + , leading to pronounced kinetic behaviors during the process of de-lithiation [33][34][35]. This could be assigned to the uniformity of the light artificial layer based on sulfonate at the surface of the TiO2 NTs [49], which could be liable for retarding the decomposition of the electrolyte. In some reported papers, previously investigated zwitterionic compounds, such as N,N-dimethylpyrrolidinium methyl sulfonate showed a minimization of electrolyte decomposition resulting in interfacial stability of the used electrodes [38]. In addition, some of zwitterionic compounds [1-(1-Butylpyrrolidinium)butane-4-sulfonate betaine, 1-(Tri-n-butylphosphonium)butane-4-sulfonate betaine [54]], contributed to the stabilization of the Li + ions in the mixtures. When compounds with a similar sulfonate functional group, such as sodium dodecylbenzene sulfonate [55] interact with nanotubes, the self-organization of organic molecules with the carbon nanotubes led to the formation of functionalized molecular structures with advanced properties [55,56]. Some future steps for further development of efficient zwitterionic compounds could lie in the modification of substituents in the imidazole ring (such as a vinyl group, or oxygenated alkyl side chain attached at one N in the imidazole ring) with the same SO3 part from the other side of imidazole ring. The interfaces' structure and the electrode-protecting films' nature are other aspects that need appropriate consideration. Raman spectra of electrode material after cycling in different electrolytes: LiTFSI/C 2 C 2 imTFSI and LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 .
The results imply that the sulfonate-based zwitterionic compound might facilitate Li + movement because a partial negative charge that developed at the end of the sulfonate SO 3 − group is able to bind with Li + , leading to pronounced kinetic behaviors during the process of de-lithiation [33][34][35]. This could be assigned to the uniformity of the light artificial layer based on sulfonate at the surface of the TiO 2 NTs [49], which could be liable for retarding the decomposition of the electrolyte. In some reported papers, previously investigated zwitterionic compounds, such as N,N-dimethylpyrrolidinium methyl sulfonate showed a minimization of electrolyte decomposition resulting in interfacial stability of the used electrodes [38]. In addition, some of zwitterionic compounds [1-(1-Butylpyrrolidinium)butane-4-sulfonate betaine, 1-(Tri-n-butylphosphonium)butane-4sulfonate betaine [54]], contributed to the stabilization of the Li + ions in the mixtures. When compounds with a similar sulfonate functional group, such as sodium dodecylbenzene sulfonate [55] interact with nanotubes, the self-organization of organic molecules with the carbon nanotubes led to the formation of functionalized molecular structures with advanced properties [55,56]. Some future steps for further development of efficient zwitterionic compounds could lie in the modification of substituents in the imidazole ring (such as a vinyl group, or oxygenated alkyl side chain attached at one N in the imidazole ring) with the same SO 3 part from the other side of imidazole ring. The interfaces' structure and the electrode-protecting films' nature are other aspects that need appropriate consideration.

Materials
The preparation of the TiO 2 nanotube arrays (NTAs) proceeded by anodic oxidation of Ti foil that is 0.25 mm thick and 0.5 cm wide (Table 1). Under the constant voltage of 30 V, in 0.7% NH 4 F in glycerol solution by using graphite as a cathode, the anodization was conducted. For a period of 6 h the anodic oxidation was accomplished in order to obtain a thick NT layer [57] for accurate mass determination. The complete electrode preparation was carried out in the manner described in our previous paper [28]. It should be noted that after the galvanostatic cycling experiments (described in Section 2.4.1) the SO 3 group from the C 1 C 4 imSO 3 zwitterionic compound from the electrolyte (detailed electrolyte content is described in next Section 3.1.1) was partially adsorbed at the TiO 2 nanotubes used as the electrode in these experiments.

Synthetic Procedure for the Zwitterionic Compound
The detail specifications of all reagents used in the experimental work are tabulated in Table 1. An equimolar amount of 1,4-butane sultone and 1-methylimidazole were mixed in acetonitrile (Figure 1d), heated, and stirred at temperature around 60 • C for 12 h. Afterwards, the used solvent was removed in vacuo and remaining solid was washed with 8 × 20 mL acetonitrile. It was dried in vacuo to give 1-butylsulfonate-3-methylimidazole (yield, 74%) at room temperature as a white solid. The resulting white solid, C 1 C 4 imSO 3 , was heated under a vacuum and stored in the vacuum desiccator with P 2 O 5 , with a final yield of 90%. The obtained zwitterionic compound was stored in the dry box under an atmosphere of nitrogen (the structure is presented in Figure 1c).
The structural confirmation of the synthesized zwitterion C 1 C 4 imSO 3 was carried out by 1 H, 13 C NMR, and FTIR analyses ( Figures S3 and S4 show the recorded spectra together with assignments in Supporting Information). The purity of the C 1 C 4 imSO 3 was estimated from the 1 H NMR spectra (Table 1).
Before use, the ionic liquid C 2 C 2 imTFSI was stored in a vacuum desiccator and in the dry box under an atmosphere of nitrogen. Using the Metrohm 831 Karl Fischer coulometer, the water content in IL was determined (Table 1).
Lithium salt, LiTFSI, was dried in a vacuum at T = 110 • C. Under inert conditions in an argon-filled glove box, the electrolyte was used to dissolve the LiTFSI in ionic liquid C 2 C 2 imTFSI, obtaining a Li + concentration c(Li + ) of 0.5 mol·dm −3 . For simplicity, the electrolyte 0.5 mol·dm −3 LiTFSI in C 2 C 2 imTFSI was referred as LiTFSI/C 2 C 2 imTFSI in this work.
The other investigated electrolyte was obtained by dissolving 0.5 mol·dm −3 LiTFSI in a previously made mixture of ionic liquid C 2 C 2 imTFSI and ZI compound C 1 C 4 imSO 3 , where 3% of C 1 C 4 imSO 3 was added. NMR. Spectroscopic identification of the synthesized C 1 C 4 imSO 3 using nuclear magnetic resonance (NMR) was recorded in D 2 O solvent at temperature T = 25 • C on a Bruker Advance III 400 MHz spectrometer. For assignation of 13C NMR spectra, the selective decoupling method was used and for 1H homodecoupling the 2D COSY method was used.

Apparatus and Procedures
FTIR. A Shimadzu Fourier Transform Infrared (FTIR) Spectrometer Reflectance with a Universal Diamond ATR Sampling Accessory (MIRacle 10 ATR; Dia/ZnSe) was used to perform the FTIR measurements at room temperature in a range from 500 to 3400 cm −1 . The synthesized C 1 C 4 imSO 3 or the tested samples during the experimental work were placed on the top of the diamond crystal and FTIR was performed under an inert atmosphere.
UV-Vis. In the wavelength range of 250-800 nm, the absorption spectra of the investigated electrolytes in this work were recorded. The UV-Vis measurements were conducted on a Thermo Scientific UV-Vis spectrophotometer Evolution 220 in 1 cm path length quartz cuvettes, at room temperature and under an inert atmosphere.
Raman spectroscopy. Raman spectra of the TiO 2 nanotubes were collected using a solid-state Nd:YAG laser excitation line of 532 nm, with an incident laser power of less than 60 mW to minimize the heating effects on the samples. The measurements were performed at room temperature. A Tri Vista 557 triple spectrometer coupled with a nitrogen-cooled CCD detector was used. The measurement was recorded in the wavenumber range of 1400-50 cm −1 .

Thermal Properties (TG, TG-MS and DSC)
TG measurement. The thermophysical properties of the synthesized C 1 C 4 imSO 3 were analyzed using a simultaneous thermogravimetric analyzer with a differential scanning calorimetry TG/DSC thermal analyzer SDT Q600 (TA Instruments, Milford, Massachusetts, USA) with calorimetric accuracy and precision ±0.2% (based on metal standards) and temperature accuracy ±0.5 • C. The initial mass used for the TG measurements was~2.5 mg and the measurement was conducted in dry nitrogen.
TG-MS measurement. The same thermal analyzer coupled online with a Hiden Analytical HPR-20/QIC mass spectrometer (Warrington, United Kingdom) was used to perform the TG-MS measurements. The sample of newly synthesized zwitterionic compound C 1 C 4 imSO 3 (~2.5 mg) was placed in an open alumina pan, the measurements were carried out in an argon atmosphere (flow rate: 100 cm 3 ·min −1 ), from room temperature to 450 • C, with a heating rate of 10 • C·min −1 . The fragments were monitored between m/z = 1-100 through 30 channels in the Multiple Ion Detection (MID) mode with the electron impact ionization mode using an electron energy power of 70 eV. The collected MS data, RC RGA Analyzer, and MAS soft Manual Set were used for the TG-MS data collection. DSC measurement. The differential scanning calorimetry (DSC) measurements for the zwitterionic compound C 1 C 4 imSO 3 were recorded using a TA Instruments Differential Scanning Calorimeter DSC Q20 (TA Instruments, Milford, Massachusetts, USA) under an atmosphere of nitrogen (flow rate 50 cm 3 ·min −1 ). The initial mass used for the DSC measurement was 5 mg. Specific heat capacity (C p ) calibration was performed with sapphire crystal provided by TA Instruments as the reference material (sapphire heat material for hermetic pans, cylindrical shaped clear disk, 22 mg, 3.2 mm diameter and 0.4 thick). The standard temperature uncertainty was u(T) = ± 0.5 • C. Thermograms were recorded during cooling from from room temperature to −90 • C and during the reheating cycle to 300 • C, at a cooling and heating rate of 10 • C·min −1 .

Electrode Material Characterization (XRD and SEM)
XRD. The crystal structure of the electrode TiO 2 nanotubes was examined by X-ray diffraction (XRD) collected with a Philips PW diffractometer 1050 with Cu-Kα1,2 radiation in a 2θ range between 20 and 80 • with step size 0.05 • and counting time 2 s per step. Figure  S5 shows the XRD Ti foil patterns before and after anodization, where there are anatase (marked with asterisk) and Ti-metal phase reflections.
SEM. The morphology of the obtained TiO 2 NTs surface was investigated using a JEOL JSM 6460LV scanning electron microscope (SEM). From the SEM images presented in Figure S6, we can see that the NTs have a more or less cylindrical shape with an outer diameter of~150 nm, inner diameter of~80 nm, and wall thickness of 35-40 nm.

Flammability Test
The flammability of the LiTFSI/C 2 C 2 imTFSI electrolyte and LiTFSI/C 2 C 2 imTFSI/ C 1 C 4 imSO 3 electrolyte could be examined by directly observing the flame on the surface of the solution for 60 s, using a Digital Thermocouple Thermometer Dual-channel LCD Backlight Temperature Meter with an R-type Thermocouple sensor probe. The details about the mass measurement and burner exposition around 1200 • C are described in Papović et al. [11].

Electrochemical Experiments
Galvanostatic measurement. Galvanostatic (GS) cycling was performed in a twoelectrode bottle-type cell made of Pyrex glass, with a Teflon stopper closed with a double "O" ring. The transparent Pyrex glass cell filled with~3 cm 3 of electrolyte enables the monitoring of a possible change in the color of the electrolyte during cycling. The twoelectrode cell was compiled in a glove box filled with argon. Li-metal foil was used as the counter electrode and the Ti/TiO 2 NTA foil as the working electrode. The studied electrolyte was in contact with 1 cm 2 of surface area of the working electrode. GS cycling was carried out at 25 • C by using the battery testing device Arbin BT 2042. GS cycling of the Ti/TiO 2 NTA electrode was conducted using a current density of 50 µA·cm −2 . The specific capacity was calculated by using the mass of the active electrode material, i.e., TiO 2 NTs, which was obtained by scraping NTs from the Ti foil after all experiments were done.

Conclusions
The zwitterionic (ZI) compound C 1 C 4 imSO 3 was synthesized to investigate its potential stabilizing effect on electrodes in LIBs during galvanostatic cycling, which corresponds to actual battery charging/discharging conditions. The specific thermophysical properties of pure C 1 C 4 imSO 3 were examined by simultaneous TG, TG-MS, and DSC analysis, and showed that the compound has quite high stability. The electrochemical properties of a 0.5 M solution of LiTFSI in the LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 system as a potential electrolyte for LIBs were tested and compared with the electrolyte LiTFSI/C 2 C 2 imTFSI, using a robust anatase TiO 2 NTA electrode as the anode material. Despite the low vapor pressure, non-flammability and room temperature electrochemical stability of the ionic liquid-based electrolytes, indicated their possible use in safer LIBs and the need for their testing with functionalized additives was emphasized. In the case of the LiTFSI/C 2 C 2 imTFSI/C 1 C 4 imSO 3 electrolyte, electrochemical galvanostatic experiments showed a significant improvement in capacity retention and Coulombic efficiency of the Ti/TiO 2 NTA electrode, compared to the electrolyte without the ZI additive. According to IR spectroscopy measurements, the same surface sites of the nanotubes are responsible for the decomposition of the ZI additive and ionic liquid-based electrolyte during galvanostatic cycling. The decomposition of the ZI additive blocked those surface sites of the TiO 2 nanotubes and thus prevented the decomposition of the LiTFSI/C 2 C 2 imTFSI ionic liquid-based electrolyte, which led to the improvement of the electrochemical properties.