An ether-functionalised cyclic sulfonium based ionic liquid as an electrolyte for electrochemical double layer capacitors

(cid:1) A novel ether-functionalised cyclic sulfonium IL, [THT G1 ][TFSI], was synthesised. (cid:1) Transport properties promoted by ether moiety relative to an alkyl-based analogue. (cid:1) Electrochemical/thermal stability [THT G1 ][TFSI] not inhibited by ether moiety. (cid:1) [THT G1 ][TFSI] used as solvent-free electrolyte in EDLC at 2.6 V. A novel cyclic sulfonium cation-based ionic liquid (IL) with an ether-group appendage and the bis {(tri ﬂ uoromethyl)sulfonyl}imide anion was synthesised and developed for electrochemical double layer capacitor (EDLC) testing. The synthesis and chemical-physical characterisation of the ether-group con- taining IL is reported in parallel with a similarly sized alkyl-functionalised sulfonium IL. Results of the chemical-physical measurements demonstrate how important transport properties, i.e. viscosity and conductivity, can be promoted through the introduction of the ether-functionality without impeding thermal, chemical or electrochemical stability of the IL. Although the apparent transport properties are improved relative to the alkyl-functionalised analogue, the ether-functionalised sulfonium cation-based IL exhibits moderately high viscosity, and poorer conductivity, when compared to traditional EDLC electrolytes based on organic solvents (propylene carbonate and acetonitrile). Electrochemical testing of the ether-functionalised sulfonium IL was conducted using activated carbon composite electrodes to inspect the performance of the IL as a solvent-free electrolyte for EDLC application. Good cycling stability was achieved over the studied range and the performance was comparable to other solvent-free, IL- based EDLC systems. Nevertheless, limitations of the attainable performance are primarily the result of sluggish transport properties and a restricted operative voltage of the IL, thus highlighting key aspects of this ﬁ eld which require further attention.


Introduction
Electrochemical double layer capacitors (EDLCs), or supercapacitors, stand among the most promising energy storage technologies today [1]. EDLCs store energy by charge separation at the electrode/electrolyte interface rather than by reversible redox reactions as in metal ion batteries [2]. To date, the major challenge for EDLCs when compared to other energy devices is their limited energy density (<10 Wh kg À1 ) [3], which currently is unable to meet the requirements of high energy density applications. To overcome this challenge, extensive work has been undertaken to increase the energy density of EDLCs which has been the subject of a number of recent reviews [3e7]. As the energy density is proportional to the capacitance and the square of the voltage, increasing the capacitance and/or the cell voltage is an effective way to increase the energy density. This can be achieved through the development of novel carbon materials for electrodes with high capacitance, such as templated carbide-derived carbons [8,9], carbon nanotube-based structures [10,11] and graphene-derived carbons [12]. Furthermore, alternative materials may be selected for the electrode structure to introduce pseudocapacitive charging contributions. These materials may combine transition metal oxides with conductive polymer coatings, such as Co 3 O 4 in polypyrrole [13] and MnFe 2 O 4 in polyaniline [14], which, by incorporating redox active pseudocapacitive processes, may increase the total achievable capacitance. Alternatively, the development of novel electrolytes with increased operating potential windows is an important strategy for improving the energy storage capabilities of devices relying on purely capacitive charge storage. Currently, the majority of commercial EDLCs use organic electrolytes based on acetonitrile or propylene carbonate (with a tetraethylammonium tetrafluoroborate, [Et 4 N][BF 4 ], salt) with a cell operating voltage of 2.5e2.8 V [3,15]. While the utilisations of these electrolyte formulations have been studied in-depth and optimised for application in EDLCs, the ultimate performance may be hindered by the limited solubility of traditional salts in the solvents. A higher solubility salt can be used to form more concentrated electrolytes to promote better electrolyte conductivities and limit the possible electrolyte depletion at the interface [3]. Additionally, higher salt concentrations within the electrolyte can be utilised to reduce the potentially hazardous volatility and flammability of such solvents and, in turn, improve the safety of an operating EDLC [16].
In this regard, ionic liquids (ILs) are presently considered as attractive electrolyte materials for the development of safer EDLCs due to their intrinsic chemical-physical properties including their non-volatility and good (electro)chemical stability. Recent studies indicate that applying ILs as EDLC electrolytes, such as 1-ethyl-3methylimidazolium tetrafluoroborate, [EMIm][BF 4 ], and 1-butyl-1-methylpyrrolidinium bis{(trifluoromethyl)sulfonyl}imide, [Pyrr 14 ][TFSI], operative voltages of up to 3.7 V can be achieved [17e19]. However, as ILs exhibit higher viscosities and lower conductivities than electrolytes based on organic solvents, IL-based EDLCs typically display lower power outputs than conventional EDLCs, especially under ambient operating conditions [20]. In recent years several strategies have been proposed to improve the power of IL-based EDLCs, focusing on the use of mixtures of ILs and organic solvents to be one of the most effective approaches. These mixtures have included conventional solvents, propylene carbonate and acetonitrile [19,21e23], and several reports of novel solvents including a variety of mononitriles [24], adiponitrile (an aliphatic dinitrile solvent) [25,26], gamma butyrolactone [27] and amides [28]. In general, the introduction of the solvent is used to promote the transport capabilities of the electrolyte relative to the neat IL while still maintaining high concentrations of the ionic species and, in turn, improve capacitance of the IL-based EDLCs at high currents under ambient conditions.
Other approaches towards the improvement of ILs as electrolytes for EDLCs and other electrochemical energy storage devices can involve judicious tailoring of the cation-anion functionality to manipulate the chemical-physical properties of the IL. For example, incorporation of ether appendages (e.g. substitution of a butyl side chain for a similarly sized 2-methoxyethyl group) can in some instances contribute to the decrease of viscosity and melting point without drastically impacting on the electrochemical and thermal stability [29e31]. Similarly, careful selection of the structure of the anionic component, for example bulky fluorine-based anions with highly delocalised electron distributions, may be utilised to tailor the chemical-physical properties of the resulting IL [29,31]. For example, in view of an IL consisting of a 1-butyl-1methylpyrrolidinium cation with a [TFSI]anion (bulky, high charge delocalisation, flexible), the melting point and viscosity (254 K and 76 mPa s at 298 K, respectively) is significantly lower than for the analogous IL with a symmetrical, non-flexible [BF 4 ] -anion where the negative charge is more localised and the melting point is 423 K [29]. Alternatively, the application of ILs based on the fluorine-free dicyanamide anions, [N(CN) 2 ] -, in EDLCs has also recently been reported [32]. In combination with pyrrolidinium based cations, the [N(CN) 2 ] --based ILs were reported to exhibit lower viscosity and higher conductivity than the [TFSI] --based analogues and, in turn, displayed very good power capabilities, and good capacity retention over 25,000 cycles when used as solventfree EDLC electrolytes. Nevertheless, the electrochemical stability, and the operative voltage, of the [N(CN) 2 ] --based ILs was reported as lower than the [TFSI]analogues, potentially limiting the available energy density of EDLC based on such an IL electrolyte.
Cations suitable for the implementation of innovative ILs include those based on sulfur which typically display lower viscosities in comparison to their analogous ammonium counterparts [33,34]. For example, at 298 K the reported viscosity and conductivity of [TFSI] --based ILs with an S-butyl-S-dimethylsulfonium cation (39.0 mPa s and 10.0 mS cm À1 at 298 K, respectively) is significantly improved relative to N-butyl-N-triethylammonium [TFSI] -(104.1 mPa s and 6.9 mS cm À1 , respectively); an analogous tetraalkylammonium cation-based IL of approximately similar size [34]. However, despite having a lower viscosity, higher conductivity and lower melting points, ILs based on trialkylsulfonium cations have attracted limited attention compared to their ammonium analogues. Nevertheless, several groups have reported work emphasising the use of cyclic [35e38] and acyclic [39e41] sulfonium ionic liquids as electrolytes for energy devices. Research surrounding the influence of ether functionalisation of the cation on the properties of sulfonium-based ILs has only been studied by Han et al. [42] wherein ILs based on the bis(fluorosulfonyl)imide anion, [FSI] -, and small acyclic sulfonium cations with and without ether functionality were reported to exhibit low viscosities (20e30 mPa s at 298 K), good conductivities (8.2e15.7 mS cm À1 at 298 K) and reasonable electrochemical stability windows (ca. 4.2e4.8 V). Recently, we also reported the use of acyclic aprotic sulfonium ILs as potential electrolytes for EDLC devices [27].
With the aim to further develop and understand the limits related to the use of sulfonium-based ILs in supercapacitors, we investigated the chemical-physical properties of the alkyl and ether functionalised cyclic sulfonium ILs, S-butyltetrahydrothiophenium bis{(trifluoromethyl)sulfonyl}imide, [ The alkylating agent, 1-iodobutane (Sigma-Aldrich, 99%, 190.46 g, 1.035 mol, 1.0 eq.) or 2-bromoethyl methyl ether (Fluorochem, 95%, 143.86 g, 1.035 mol, 1.0 eq.), was added to a flask of tetrahydrothiophene (Sigma-Aldrich, 99%, 91.28 g, 1.035 mol, 1.0 eq.) with a small amount of acetone (Sigma-Aldrich, 99.8%). The reaction mixture was stirred vigorously, covered in aluminium foil and left to react for 72 h at 333 K to give the corresponding halide salts as a white solid. After filtration of the solid, each of the intermediate salts were analysed by 1 H NMR. These sulfonium halide salts were then mixed and stirred with a solution of lithium bis {(trifluoromethyl)sulfonyl}imide (3 M, battery grade) (205.84 g, 0.717 mol, 1.1 eq.) and (233.69 g, 0.814 mol, 1.1 eq.) for the iodide and bromide salts, respectively in 150 cm 3 Figure S1 in the Supporting Information. Prior to all further measurements, the ILs were dried under high vacuum (10 À3 mbar) at elevated temperature (343 K) with stirring for several days. The ILs were subsequently stored in an Ar-filled glove box (<3 ppm H 2 O) to limit water contamination by the ambient atmosphere. 1

Physical measurements
Water content of the dried ILs was analysed by Karl Fischer Coulometric titration using an 899 Coulometer (Metrohm). The resolution of the water content measurements was 0.001 wt/wt % (or 10 ppm) and measurements were completed in duplicate. All ILs displayed water contents lower than 50 ppm H 2 O after vacuum drying. Thermal gravimetric analysis (TGA) of the ILs was undertaken using a Q5000 TGA instrument (TA Instruments). The sample size was 5e10 mg for each IL. Heating was conducted from room temperature to 873 K (±1 K) applying a temperature gradient of 5 K min À1 using nitrogen as purge gas (10 cm 3 min À1 ). Differential scanning calorimetry (DSC) analysis of the ILs was completed using a DSC Q2000 (TA Instruments) using a heating gradient of 5 K min À1 between lower and upper temperature limits of 183.15 K and 323.15 K (±0.1 K), respectively. IL samples for DSC were prepared in hermetically sealed Al pans inside an Ar-filled glove box.
Density measurements were completed at atmospheric pressure using a DM40 (Mettler Toledo, ± 1 Â 10 À4 g cm À3 ) oscillating tube density meter in the range of 293.15e363.15 K (±0.1 K). The DM40 instrument was cleaned using acetone and dried using dehumidified air prior to any measurements. The viscosity of the ILs was measured using a Bohlin Gemini Rotonetic Drive 2 cone and plate rheometer (±1%) from ca. 293.15e363.15 K (±0.01 K) at atmospheric pressure. Density and viscosity measurements were completed under atmospheric conditions and, therefore, samples were sealed in glass vials inside the Ar-filled glove box and only exposed to the ambient atmosphere immediately before completion of the respective measurement to limit water contamination. Conductivity measurements were completed using a sensION þ EC71 benchtop meter with a 3-pole platinum sensIONþ 5070 conductivity probe (<0.5% of range) with an in-built Pt1000 temperature probe (Hach Lange). The conductivity probe was calibrated using aqueous KCl standard conductivity solutions (147 mS cm À1 , 1413 mS cm À1 , and 12.88 mS cm À1 at 298.15 K). The sample was prepared inside an Ar-filled glove box by immersion of the probe into the liquid sample in a glass sample tube and sealing using an O-ring seal and parafilm. The conductivity of the IL was then measured as a function of the temperature within the range of 293.15e363.15 K (±0.2 K). The temperature of the sample was controlled using a small oil bath and the temperature and conductivity of the sample was recorded when the observed values were stable for ca. 30 s.

Electrochemical window at glassy carbon electrodes
Measurements of the electrochemical stability windows by cyclic voltammetry were performed using a VMP multichannel potentiostatic-galvanostatic workstation (BioLogic Science Instruments). Measurements were conducted inside the dry atmosphere of the Ar-filled glovebox using a three-electrode configuration in a three-necked glass cell. The working electrode was a glassy-carbon macro-disk (ALS Co., Ltd., 3 mm diameter). Prior to all experiments, the glassy carbon working electrode was polished using alumina slurries of decreasing grain size (1.0 mm, 0.3 mm, and 0.05 mm) in distilled water. The electrode was then sonicated in distilled water for 2e3 min then dried at ca. 353 K in an oven. The counter electrode was platinum-coiled wire heat-sealed in a glass capillary. The Ag[OTf]/Ag reference electrode consisted of the silver wire immersed in a 0.01 mol dm À3 solution of silver triflate (Ag[OTf]) in 1-butyl-1-methylpyrrolidinium bis{(trifluoromethyl)sulfonyl}imide, [Pyrr 14 ][TFSI], separated from the bulk solution by a glass-frit tip. The Ag[OTf]/Ag electrode potential was referenced vs. an internal ferrocene (Sigma-Aldrich, 98%) couple. The determination of the ferrocene redox couple was completed by cyclic voltammetry after adding a small quantity of ferrocene to the IL after measurement of the electrochemical window.

EDLC measurements
In order to assess the usability of [THT G1 ][TFSI] in EDLC application, the IL was tested with carbon based composite electrodes. For this procedure, electrodes were prepared following the procedure reported in Ref. [43] using activated carbon, AC, as the active material. The final ratio between active material (DLC Super 30, Norit), conductive agent (SuperC65, Imerys) and binder (CMC, Walocel CRT 2000 PPA 12, Dow Wolff Cellulosics) was 90:5:5. The electrode thickness was equal to 80 mm and the electrode diameter was 12 mm. The mass loading of the active material of the electrodes was, on average, 3.9 mg cm À2 . Free standing electrodes with mass loadings over 30 mg cm À2 were prepared using a different binder (polytetrafluoroethylene, PTFE, Sigma Aldrich) with a final composition of 85:10:5 (activated carbon, carbon black, PTFE). The aqueous PTFE dispersion, the activated carbon and the carbon black were dispersed in an excess amount of ethanol and stirred at elevated temperatures until the slurry became a highly viscous, dough-like mass. This mass was put onto a glass plate and rolled out several times before punching out disks with a diameter of 12 mm (area: 1.13 cm 2 ). The obtained electrodes were pre-dried in an oven (ED 115, Binder) set to 353 K, before drying them for 24 h under vacuum at 453 K, in order to remove any residual water.
Electrochemical tests were performed using a VMP multichannel potentiostatic-galvanostatic workstation (Biologic Science Instruments) connected to climatic chambers set to 293 K (KBF 115, Binder). The electrochemical investigations were carried out using Swagelok ® -type cells, which were assembled in an Ar-filled glove box with water and oxygen contents below 1 ppm.
The maximum operative voltage of the electrolyte was determined in a three-electrode cell setup. An AC composite electrode was used as working electrode and a heavy and free standing AC electrode was used as counter electrode. As a reference electrode, an Ag quasi-reference electrode was employed. A Whatman GF/D glass microfiber filter disk (675 mm thickness and 13 mm diameter), soaked with 150 mL of electrolyte, was used as separator. The measurements were performed by cyclic voltammetry (CV), scanning the cell potential with a scan rate of 5 mV s À1 . Starting from a potential of 0.8 V/À0.8 V vs. Ag, the maximum potential was increased/decreased stepwise by 0.1 V until the efficiency of the CVs dropped below 99%.
In order to evaluate the performance of EDLCs with a [THT G1 ] [TFSI] electrolyte, CV and galvanostatic charge-discharge cycling experiments were carried out in a 2-electrode configuration using 100 mL of electrolyte and two AC composite electrodes. An asymmetric configuration with different electrode masses based on Equation (1) was used for this test using values of specific capacitance (C), carbon loading (m) and voltage excursion (DV) obtained by the 3-electrode configuration [44].
Cyclic voltammetry was carried out using scan rates ranging from 5 mV s À1 to 200 mV s À1 . The reported values for specific capacitance correspond to the value achieved at the half of the maximum voltage and were normalised using the mass of the active material of both electrodes. Galvanostatic charge-discharge cycling was carried out using current densities ranging from 0.5 A g À1 to 5 A g À1 . The values of capacitance of the total active material (C), equivalent series resistance (ESR), Coulombic efficiency (h eff ), average energy (E) and average power (P) have been calculated using the formulas indicated in Refs. [43,45].

Results and discussion
The tetrahydrothiophenium bis{(trifluoromethyl)sulfonyl}imide based ILs were synthesised via alkylation of tetrahydrothiophene in moderate yield followed by subsequent metathesis of the resultant halide salts with Li[TFSI] in a dichloromethane/aqueous mixture. The ILs were purified by multiple washings with water to remove the halide and lithium residues followed by extensive drying under high vacuum. The chemical structures of the synthesised ILs, and their respective abbreviations, used in this work are shown in Fig. 1. A physical chemical characterisation of the two ILs was completed and the results are briefly summarised at 298.15 K in Table 1. Electrochemical characterisation and application of the ether functionalised IL in an EDLC in described in the latter stages of this work.

Thermal properties
Differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) of the ILs were used to study thermal phase transitions and thermal stability, respectively. As described in the experimental section, the DSC traces were completed within a range of 183.15 K and 323.15 K at a rate 5 K min À1 . Both ILs exhibited no distinct features associated with freezing or melting within this range at this heating rate. Accurate determination of the liquid range may require further experiments in which a slower heating rate and wider temperature range is utilised. In the low temperature region, both ILs showed evidence of a glass transition during the heating scan in the form of small broad peaks. The temperature at the mid-point of these peaks is used to define the glass transition temperature, T g , for these ILs; ca 191 K for [THT 4 ][TFSI] and 187 K for [THT G1 ][TFSI]. The low temperature region of the DSC traces is shown in Figure S6 in the Supporting Information. The data presented in the traces show that these peaks appear very close to the switching temperature (183.15 K), particularly for the [THT G1 ][TFSI] IL. This factor may limit the accuracy of determination of this peak position, especially since T g peaks are broad. Nevertheless, the relative comparison of the ILs shows that the introduction of ether functionality depresses the T g . This observation is possibly attributed to higher rotational freedom and flexibility of the ether group  M w is the IL molecular weight, T g and T d are the glass transition temperature and thermal decomposition temperature, respectively, r is the density, h is the viscosity, s is the conductivity, L is the molar conductivity and V m is the molar volume. and, for example, has been reported for [TFSI] --based ILs with Nbutyl-N,N-diethyl-N-methylammonium (T g ¼ 180 K) and N,Ndiethyl-N-(2-methoxyethyl)-N-methylammonium (T g ¼ 178 K) [31]. For the TGA experiments, the temperature is increased at a rate of 10 K min À1 giving dynamic information about the sample mass change in response to high temperatures. Results of the TGA analysis are shown in Figure S7 in the Supporting Information. For this analysis, the decomposition temperature (T d ) is the temperature at which a 5% loss in the sample weight is recorded. The measured T d for the alkyl and ether-functionalised ILs is 523 K and 554 K, respectively. The recorded value for [THT 4 ][TFSI] is slightly lower than has been reported previously for the thermal decomposition of a series of alkyl functionalised tetrahydrothiophenium [TFSI]based ILs (which ranged from 545 to 550 K) [37]. This may be the result of several contributing factors; different pan materials, trace level impurities (including water) for example. Nevertheless, these results importantly show that the substitution of the ethereal oxygen into the functional group, does not negatively affect the relative thermal stability of the IL and it is, therefore, sufficiently thermally stable for practical electrochemical energy storage devices.
In addition, the combination of DSC and TGA measurements facilitates the approximation of the lower and upper temperature limits of the liquid ranges of the ILs.  [26], respectively, which are considerably higher than the glass transition temperatures observed for both the alkyl and ether functionalised thiophenium ILs reported here.
While these observations were completed using dynamic techniques, the results suggest the possibility of the application of this class of tetrahydrothiophenium-based ILs as electrolytes in lowtemperature EDLCs.

Density
The densities of the sulfonium-based ILs were measured as a function of the temperature within the range of 293.15e363.15 K.
The temperature dependence on the measured density, r, and the molar volume, V m , (calculated using the density and molar mass) for both ILs is presented in Fig. 2. As is expected for ILs, the density of both samples was found to decrease with a linear dependence with respect to increasing temperature. As such, the data is fitted using the linear fitting equation where T is the sample temperature and A and B are the coefficients of the linear fit. The fitting parameters used in the linear correlation are shown, with the numerical density data, in Table S1 in the Supporting Information. Additionally, a logical increase in the molar volume of both ILs is observed with increasing temperatures. The experimental density of the butyl functionalised IL, [THT 4 ] [TFSI], at 298 K (1.4558 g cm À3 ) is in excellent agreement with previously reported values (1.46 g cm À3 and 1.456 g cm À3 ) at the same temperature [36,37]. Substitution of the butyl chain with the similarly sized 2-methoxyethyl group creates an IL with similar molecular weight but leads to tighter packing and a more dense IL. This trend has been observed previously for alkyl/etherfunctionalised imidazolium, ammonium and phosphonium-based ILs [29e31, 51,52].
Furthermore, the isobaric coefficient of thermal expansion, a p , of the sulfonium ILs was calculated using Equation (3).

For [THT 4 ][TFSI] and [THT G1 ][TFSI]
, the a p values obtained at 298 K were 6.59 Â 10 À4 K À1 and 6.64 Â 10 À4 K À1 , respectively. Since the observed temperature dependence of density and molar volume appears to be linear across the studied temperature range, minimal deviation in the calculated a p values is found. These calculated values are in good accordance with a p values previously

Viscosity and conductivity
The viscosity and conductivity of EDLC electrolytes is of particular importance for the high-power capabilities demanded by the application. Since viscosity is a property linked to the movement of the constituent molecules of the liquid, low viscosity electrolytes are required to enable good conductivity and rapid response of the ionic species to applied polarisation at the electrode interface. Both the viscosity and conductivity of an IL electrolyte contribute significantly to the maximum specific power, P (W$kg À1 ), of a given EDLC, as shown in Equation (4) [3]: where U max is the maximum operative potential range in volts, m is the mass of the EDLC in kg and the ESR is the equivalent series resistance in ohms. The ESR is a summation of the different contributions to resistance of the EDLC, including electrode/electrolyte interfacial resistances and ionic resistance (reciprocal of the electrolyte ionic conductivity). Therefore, low viscosity and high ionic conductivity electrolytes are favoured for minimizing excessive contributions to the ESR. Since the operative potential range of ILs is characteristically high, it is the typically high viscosity and poor conductivity which may limit the application of pure IL electrolytes. The viscosity and conductivity of the ILs was measured as a function of temperature within the range of 293.15e363.15 K. The temperature dependence of the viscosity and conductivity for the two tetrahydrothiophenium ILs is presented in Fig. 3. Both ILs exhibited a non-linear reduction in the measured viscosity with respect to an increase in the sample temperature. In accordance with the typical inverse proportionality observed between viscosity and conductivity, the measured conductivity of both ILs is found to increase with a non-linear dependence on the sample temperature. Correlation of the temperature dependence of the data was completed using the Vogel-Tamman-Fulcher, VTF, equations for viscosity, h (Equation (5)), and for conductivity, s (Equation (6)): s are fitting parameters for the viscosity and conductivity correlation, respectively, and T is the experimental temperature. The VTF equations are typically utilised for correlation of the temperature dependencies of conductivity and viscosity for glassy-forming liquids which do not obey Arrheniustype dependence [54]. The correlation parameters, and the  (5) and (6)).
numerical viscosity and conductivity data, are shown in Tables S2  and S3 in the Supporting Information. Across the full temperature  range, the viscosity of the ether-functionalised IL, [THT G1 ][TFSI], is found to be lower than for the butyl-functionalised analogue. As discussed previously, the substitution of a short alkyl substituent chain on an IL cation centre with a similarly sized ether group has been previously reported to result in a reduction of the IL viscosity [29e31, 52,55]. In accordance with the reduction in the viscosity, the measured conductivity of the ether-functionalised IL is higher than for the alkyl-functionalised analogue. This observation is expected due to the similarities between the nature and the size of the IL structures. . It is common for significant degrees of variation to occur between reported viscosity and conductivity values possibly owing to several factors including experimental technique and low-level impurities. One such impurity, which may significantly contribute to differing reports of ILs transport properties, is the water content (listed in Table 2 where available in the respective referenced articles). Higher water content in an IL may reduce the observed viscosity and, in turn, promote the conductivity of the solution. For the viscosity measurements reported in this work, and the comparative references, experimental measurements were conducted under ambient atmosphere without any moisture control. Therefore, the actual water content at the time of measurement would vary depending on the conditions of the atmosphere and the exposure time. As described in the experimental section, the IL samples were sealed in a glass vial with a septum inside an Ar-filled glove box and only exposed to atmosphere when added via a syringe and needle to the cone and plate rheometer. Taking this factor into consideration, the variation in the apparent viscosity measurements follows an approximate trend in-line with the IL water content. Conversely, no such trend may be inferred for the conductivity data and, in particular, the highest reported value of 3.2 mS cm À1 is in combination with the highest measured viscosity at the lowest recorded temperature which ultimately appears paradoxical. For the conductivity measurements reported in this work, the IL sample and probe were sealed inside a glass sample tube using an O-ring seal and parafilm while inside the Ar-filled glove box. This method has been tested previously using a range of different IL samples and no observable changes to the measured conductivity occurred over a period of 48 h.

Ionicity
To further probe the ionic behaviour and the inverse relationship between the exhibited viscosities and conductivities of the two tetrahydrothiophenium ILs, the ionicity of the ILs may be considered [57]. The degree of ionicity, or degree of (dis)association of the ionic species, is a description of the extent to which the ionic species within the IL behave as free and dissociated ions. A simple inspection of this feature is made using a Walden plot of Log 10 (h À1 ) vs. Log 10 (L), where the reciprocal viscosity (or fluidity), h À1 , is given in Poise À1 and the molar conductivity, L (where L ¼ s (S$cm À1 )/ concentration (mol$cm À3 )), is given in S$cm 2 $mol À1 . The concentration was calculated using measured density and the molar mass of the IL. The Walden plot for the two tetrahydrothiophenium ILs in the temperature range of 293.15e363.15 K is shown in Fig. 4. Specific data points on the Walden plot are calculated using the correlation parameters of the density, viscosity and conductivity and the molecular weight of each IL. The solid line represents the ideal Walden behaviour of a 0.01 mol dm À3 KCl aqueous solution, a strong electrolyte solution in which the ionic species are known to be fully dissociated and equally mobile [58]. The magnitude and direction of deviation away from this ideal KCl line is said to describe the degree of ionicity of the studied materials. For example, an IL which lies well below the KCl line will exhibit a much lower-than-expected conductivity for the given viscosity according to the Walden rule. Where this deviation, DW (where DW represents the magnitude of the vertical displacement from the ideal KCl line) is greater than an order of magnitude lower than the ideal KCl line (DW > 1), these ILs have been described as poorly ionic as a result of various interionic associations [57,59,60].
Conversely, ILs which lie close to the ideal KCl line, and DW is small,  [56] a Value in brackets corresponds to an estimation of the reported viscosity of the IL at 298 K based on the provided Arrhenius correlation (h ¼ A·exp(E a /RT)) parameters given in the respectively referenced article (E a ¼ 36.44 kJ mol À1 and A ¼ 3.9 Â 10 À5 mS cm À1 ). b ¼Not reported. The number of significant figures provided for the literature values is exactly as provided by the authors of the respective articles. show more dissociated ionic character in which the constituent ions are more independently mobile. Both the alkyl and etherfunctionalised tetrahydrothiophenium based ILs reported in this work show small deviations away from the ideal line (DW > 0.3) across the full temperature range representing 'good ionic liquids' in which the ionic species are mostly dissociated and mostly move independently. Nevertheless, both of the studied tetrahydrothiophenium-based ILs display a decrease in the calculated Walden-like product (L·h) with increasing temperature ranging from ca. 60 Â 10 À7 S N$s$mol À1 at 293.15 K to 53 Â 10 À7 S N s mol À1 at 363.15 K for [THT 4 ][TFSI] and 60 Â 10 À7 S N s mol À1 to 52 Â 10 À7 S N s mol À1 for [THT G1 ][TFSI] over the same range. The magnitude of these values is in good agreement with other previously reported [TFSI] --based ILs [61] including [Pyrr 14 ][TFSI] (L·h ¼ 56 S N s mol À1 at 298 K). This variation in calculated L·h is additionally in accordance with a slight increase in the observed DW over the same temperature range and demonstrates a lower gradient of both apparent linear dependencies relative to the ideal KCl line. While this variation is minimal over the studied temperature range, the observations imply that the conductivity of both ILs does not increase at the rate expected from the measured reduction in IL viscosity. Furthermore, the slight difference in the magnitude of DW for the two ILs may be discussed in terms of the different cation structures. Firstly, the only difference in IL structure is the substitution of the butyl group of the cation for the ether group which results in a slight increase in DW for the latter example. In comparison to the alkyl group, the ether moiety is more electron donating and, in turn, is more able to donate electron density towards the positively charged sulfur centre of the cation. As such, the charge distribution of the [THT G1 ] þ cation is expected to be more localised around the sulfur centre relative to the [THT 4 ] þ cation. As an effect of this relatively small difference, a more localised positive charge would favour stabilisation through Coulombic interaction with the anionic species forming ion pairs and, in turn, reducing the apparent ionicity of the IL.

Electrochemical window at a glassy carbon electrode
One of the primary benefits of utilising ILs for electrochemical energy storage devices, in addition to the safety aspects related to non-volatility and non-flammability, is the typically wide electrochemical stability. Electrochemical stability of electrolytes is defined by the positive and negative potentials, in volts, at which the onset of oxidative or reductive decomposition reactions, respectively, occur at the electrode/electrolyte interface. In turn, the potential window defined by these upper and lower potential limits to give the electrochemical window of the electrolyte. The electrochemical window of the [THT 4 ][TFSI] and [THT G1 ][TFSI] was measured by cyclic voltammetry at 2 mV s À1 at a glassy carbon working electrode and the resulting voltammograms are presented in Fig. 5. For the purpose of comparison of the specific potentials, the Ag[OTf]/Ag reference potential was normalised vs. the formal redox potential of an internal ferrocene couple (Fc þ /Fc). Additionally, low current boundaries of ±0.03 mA cm À2 to define the onset potential of electrolyte decomposition.
As seen in Fig. 5, both ILs exhibit one large negative current response at approximately À2.0 to À2.2 V vs. Fc þ /Fc associated with reductive decomposition of the ILs, most likely reduction of the cation species. In the anodic oxidative region, both ILs exhibit a significant oxidative current at approximately 2.25e2.35 V vs. Fc þ / Fc. In addition to this, a small positive current response is observed at ca.   [62], the electrochemical windows of the cyclic analogues reported in this work show wider and cleaner electrochemical stability windows (where the term cleaner refers to the absence of significant current responses in addition to the two primary decomposition peaks). In comparison with other IL families commonly utilised for electrochemical applications, the reductive stability is approximately 1 V more positive than typically reported for quaternary ammonium [52,63e65] [64] and P-pentyl-P,P,Ptriethylphosphonium-[TFSI] [68], respectively). Conversely, the reported magnitude of DE for the tetrahydrothiophenium ILs is on a similar scale to that reported for imidazolium [TFSI] --based ILs [51,69]. Nevertheless, the results presented in Fig. 5 importantly show that the introduction of ether-functionality to the cyclic sulfonium cation does not negatively affect the electrochemical stability relative to the alkyl-functionalised analogue.

EDLC measurements
After identifying the electrochemical stability window, the operative voltage of the [THT G1 ][TFSI] IL was determined via completion of CV measurements using a 3-electrode setup with high surface area, EDLC-type electrodes. The CV curves as well as the corresponding efficiencies are presented in Fig. 6. The positive and negative potential limits were defined by the highest and lowest potential limits, at which the efficiency of the chargedischarge remained higher than 99%. Within this threshold, the IL ILs at a glassy carbon working electrode vs. a Ag[OTf]/Ag reference electrode (top Xaxis) and normalised vs. an internal reference potential of the ferrocene redox couple (bottom X-axis). Scan rate was 2 mV s À1 . Table 3 Cathodic reduction potential (E c ), anodic oxidation potential (E a ) and electrochemical windows (DE) of the alkyl and ether functionalised tetrahydrothiophenium bis{(trifluoromethyl)sulfonyl}imide ILs at a planar glassy carbon working electrode. electrolyte is considered stable and no significant loss of Coulombic efficiency, via undesirable Faradaic electron transfer decomposition reactions, is observed. As seen in Fig. 6, when used with the AC electrodes, this operative potential range is between À0.85 V vs. Ag and 1.85 V vs. Ag. The positive potential limit of 1.85 V vs. Ag corresponds to the anodic stability of the [TFSI]anion and is in line with other potential limits determined for this anion [45]. The maximum potential of the [THT G1 ] þ cation of À0.85 V vs. Ag is low, particularly when compared with the apparent cathodic reduction potential of the IL (À2.15 V vs. Fc þ /Fc) as determined at a planar glassy carbon electrode. This difference in potential limits can most certainly be attributed to the differences in electrochemical setups. The electrochemical stability window is derived using inert, planar and nonporous electrodes, which feature high electrical conductivity and very low specific surface areas (like the used glassy carbon and Pt-wire) [70]. Conversely, EDLC electrodes are constructed using porous activated carbon with varying amounts of surface groups, which are able to interact in different ways with the electrolyte species. This variation in electrode structure ultimately effects the determination of electrolyte stability and is responsible for the difference in potential limits obtained by the different methods. This is another example of how the applied method influences the assessment of the stability of the electrolyte, which one has to keep in mind for the determination of the operative voltage for full cells [70,71]. Nevertheless, the actual maximum cathodic stability of the [THT G1 ] þ cation at the EDLC-type electrodes (À0.85 V vs. Ag) is very similar to previously reported literature values for three [TFSI] -based ILs with acyclic S-alkyl-S,S-diethylsulfonium cations (where the term alkyl refers to a methyl, ethyl and propyl group) [41].

IL cation
To further evaluate the electrochemical behaviour, EDLCs with two AC-based composite electrodes were constructed. Using the potential limits determined previously, a maximum cell voltage of 2.6 V was selected for these tests. Due to the difference in voltage excursion on the positive and negative electrodes, an asymmetrical cell configuration with different electrode masses was selected based on Equation (1). Fig. 7a shows an exemplary CV for the investigated electrolyte completed at a scan rate of 20 mV s À1 . The typical rectangular shape for capacitive behaviour of an EDLC is observed and no additional current peaks, which would indicate unwanted Faradaic reactions, are present. Using this scan rate, a specific capacitance of 21 F g À1 is reached for the whole device. This value of capacitance is comparable to that observed for similar electrodes in pyrrolidinium-based and trialkylsulfonium-based IL electrolytes [41,45]. Fig. 7b shows the development of the specific capacitance with an increasing scan rate. At a scan rate of 200 mV s À1 , 33% of the initial capacitance at 5 mV s À1 is left. As discussed previously, at 298 K the [THT G1 ][TFSI] IL has a conductivity of 3.01 mS cm À1 and a viscosity of 70.38 mPa s, which leads to high resistance and, therefore, poorer performance at higher rates.
The electrochemical performance of the IL was further investigated by galvanostatic cycling experiments using current densities ranging from 0.5 A g À1 to 2 A g À1 with a maximum applied operative voltage of 2.6 V (Fig. 8). The voltage profiles of the different current densities are presented in Fig. 8d; a comparison of the voltage profile of the investigated [THT G1 ][TFSI] IL with a standard organic electrolyte (1 mol dm À3 Et 4 NBF 4 in PC) using a current density of 1 A g À1 is presented in Figure S8. In the cycling experiments, the Coulombic efficiency reaches values close to 100%. Furthermore, the EDLC is very stable upon cycling at the different current densities, indicating that no electrolyte degradation is taking place. Both the ESR and the specific capacitance show a minor decrease in the first 1000 cycles stabilizing at values of 28.5 U cm 2 and 18.7 F g À1 , respectively. At 2 A g À1 the specific capacitance decreases to 14.3 F g À1 , while at 5 A g À1 the system becomes completely polarized. Again, the relatively high viscosity of the IL at room temperature limits the possible electrochemical performance of the EDLC at higher current densities. The voltage profiles show a linear increase/decrease of the voltage upon charging/discharging, thus, no contribution of faradaic processes are involved in the storage processes of the investigated combination of materials. Upon reversing the applied current, a large  Ohmic drop is visible attributed to the high resistance of the EDLC due to the high viscosity of the IL at room temperature. Compared to an organic electrolyte, 1 mol dm À3 Et 4 NBF 4 in PC ( Figure S8

Conclusions
In this work, we have reported the synthesis and chemicalphysical properties of the alkyl and ether-functionalised cyclic sulfonium ILs, [THT 4 ][TFSI] and [THT G1 ][TFSI], respectively. An assessment of the thermal properties revealed both ILs exhibit wide liquid range temperatures, particularly in sub-ambient temperatures. Physical characterisation of the density and transport properties, viscosity and conductivity, of the ILs showed similarities between commonly used, and similarly sized, ILs like those based on the pyrrolidinium cation with the [TFSI]anion. Additionally, these measurements demonstrated how small structural modifications, like the introduction of an ethereal oxygen onto the cation, may be used to improve or fine-tune some desired characteristics. Specifically, the introduction of the ether-appendage was found to reduce the measured viscosity and increase the conductivity of the IL, relative to the alkyl-functionalised analogue. Moreover, this structural modification was found to have no negative implications on the thermal stability of the IL nor the electrochemical stability window of the IL, as measured by CV at a glassy carbon working electrode.
Initial EDLC testing using the novel ether-functionalised IL, [THT G1 ][TFSI], as a solvent-free electrolyte with AC-composite electrodes was completed to inspect the potential application of this material. The electrochemical performance of these EDLCs was found to be comparable to that of other solvent-free EDLC systems, i.e. using an electrolyte based on the pyrrolidinium-cation. However, the operative voltage of the investigated devices appears lower than that of many other solvent-free EDLCs and power outputs are limited by sluggish transport properties at ambient temperatures. Taking these points into account, the increment of the operative voltage and improvement of transport properties appear as the main goals to achieve in order to take full advantage of the favourable properties of ILs as electrolytes. Furthermore, also the behaviour at low temperature of these innovative solvent-free EDLCs should be investigated.