Optimizing protic ionic liquid electrolyte for pre-intercalated Ti₃C₂T_x MXene supercapacitor electrodes

Improving energy storage devices such as batteries and electrochemical supercapacitors is crucial to reducing our dependence on fossil fuel-based energy. While supercapacitors typically have lower specific energy than batteries, they have the advantage of fast charging and discharging times, which gives them excellent specific power.^1,2 Therefore, it is essential to engineer supercapacitors with high specific energy and power for various applications, such as transportation and grid energy storage.³ The capacitor’s energy density (E) can be calculated using the equation E = ½ CV² where C is the specific capacitance, a property of electrode materials, and V is the operational voltage window, determined by the properties of electrolytes. Increasing the specific capacitance and voltage window is essential to unlocking high specific energy in supercapacitors.

MXenes are transition metal carbides and/or nitrides, first discovered in 2011, which have many properties that make them attractive electrode materials for electrochemical energy storage devices.^4,5 They are electrically conductive and rich with surface functional groups, therefore allowing tunable chemistries and the ability to host many ions.⁶ MXenes have already shown great promise as supercapacitor electrodes in aqueous systems.⁷ It was found that nanoengineered⁸ and pre-intercalated⁹ Ti₃C₂T_x achieve a higher capacitance in sulfuric acid electrolytes. Unfortunately, the narrow breakdown voltage window of water limits the overall capacitance and, thus, the specific energy of aqueous supercapacitors. The utilization of MXenes in non-aqueous electrolytes, including room-temperature ionic liquid (RTIL) electrolytes with high operational voltage windows,¹⁰ has been minimal. A study to intercalate RTILs between MXene layers showed that, similar to the MXene aqueous system where water molecules reside in the inter-stack gaps,¹¹ the ionic liquid ions do not intercalate between the MXene layers.¹² Instead, it gets trapped in the larger-sized MXene inter-stack gaps. It has been concluded that the d-spacing in typical MXenes is not wide enough for the ions of an ionic liquid, which are too large to intercalate. Even under applied potential, ionic liquid ions did not intercalate between the layer, both in pristine and in the presence of acetonitrile in different concentrations. However, an enhancement in specific capacitance in the presence of 25% acetonitrile has been reported.¹³ Therefore, in one of our previous studies, we pre-intercalated MXene with alkylammonium (AA) cations with alkyl chains containing 6 (C6) to 16 (C16) carbon atoms. Pre-intercalated MXene with 12 (C12) carbon atoms in the alkyl chain (referred to as Ti₃C₂T_x-C12), with an interlayer spacing of ∼2.2 nm, allowed for RTIL cation electrochemical intercalation.¹⁴ As a result, it showed a significant improvement in electrochemical behavior both in a neat 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide (EMIMTFSI) and in a 1M EMIMTFSI in acetonitrile, with the localized motion of the ionic liquid ions in confinement.¹⁴ 

Although they have a large operational voltage window, aprotic ionic liquids struggle to form solvation shells and stabilize ions.^15,16 Meanwhile, protic ionic liquids, which have a considerable stability window and labile protons, have superior capacitive reactions compared to the large bulky cations of the aprotic ionic liquids.¹⁷ Furthermore, protic ionic liquids can add aqueous solution-like conductivity and other advantageous characteristics of ionic liquids to energy storage applications,¹⁸ where the ion transfer mechanism plays a significant role. Two different transport mechanisms, the vehicle mechanism, which is a slow diffusion-controlled process, and a Grotthuss hopping mechanism of fast transport, have been suggested for systems involving ionic liquids.^19,20 In a protic ionic liquid with a labile proton that can be transferred easily, the Grotthuss-like mechanism of structural diffusion results in higher conductivity. Higher ionic conductivity observed in phosphoric acid, which has labile protons, has been found to demonstrate a hopping mechanism of fast proton transport.²¹ Therefore, protic ionic liquids have the potential to serve as an electrolyte for fast ion transfer. One way to catalyze the proton hopping behavior in an ionic liquid is by using cations with more than one basic site. Having more basic sites allows more spots for proton transfer and favors such a mechanism over the slower matrix-assisted transport. A second consideration is synthesizing protic ionic liquid electrolytes by maximizing fluidity, particularly by exploiting different mixtures involving ionic liquids and salts.²² Here, we have examined the electrochemical performance together with ions dynamics of the protic ionic liquid formed using different compositions of BuIM and HNTf₂ (Scheme 1) confined in Ti₃C₂T_x-C12 MXene. Using cyclic voltammetry together with quasi-elastic neutron scattering, we have found an optimal electrolyte ratio of 1-butyl imidazole (BuIM): bis(trifluoromethanesulfonyl)imide (HNTf2) of 0.8:0.2 at which the Ti₃C₂T_x-C12 MXene electrode system provides the optimal specific capacitance. This higher capacitance comes from accumulating larger numbers of BuIMH⁺ on the surface of the electrodes, thereby giving more room for the solvent molecules to move in the space away from the surface, as revealed from the higher diffusivity obtained from the dynamics measurements.

We started by examining the protic ionic liquid’s capacitive behavior, which was calculated using the coarse-grained molecular model as detailed in Ref. 23. The normalized differential capacitance of the protic ionic liquid confined in a 2 nm slit pore as a function of the protic ionic liquid, BuIM, concentration is presented in Fig. 1. There is a maximum capacitance (four times more than pure ionic liquid) at 80 mol % of BuIM in its mixture with HNTf₂, suggesting the significant impact of the composition of the protic ionic liquid on electrochemical performance. Using a coarse grain model, Gallegos and Wu have attributed such concentration-dependent capacitance maximum to the interplay between the solvent molecules (BuIM), coions (BuIMH⁺), and counterions (NTf₂⁻) in the mixture.²³ They predicted a significant change in the number of BuIm molecules, resulting in maximum differential adsorption of the counterions inside the pore when 20 mol % of HNTf2 is present. Based on their calculations, they concluded that at this particular composition, an optimal balance results from a trade-off between the higher number density of ions and the non-electrostatic repulsion among ions (cation and anion).

We evaluated the electrochemical performance of Ti₃C₂T_x-C12 MXene (material characterization details in Fig. S1 of the supplementary material) in electrolytes with different BuIM:HNTf₂ ratios (details regarding electrode and electrolyte preparation are presented in the supplementary material). Figure S2 presents cyclic voltammograms (CVs) at the scan rates of 1, 2, 5, 10, 20, 50, and 100 mV/s for pre-intercalated Ti₃C₂T_x-C12 MXene at BuIM:HNTf₂ molar ratios of 0.9:0.1, 0.8:0.2, 0.5:0.5, and 0.2:0.8. The CVs for 0.9:0.1 BuIM:HNTf₂ electrolytes [Fig. S2(a)] exhibit a resistive behavior with negligible capacitance even at a low scan rate of 1 mV/s. As shown in Fig. S2(b), a significant improvement in the electrochemical performance was noticed by changing the ratio to 0.8:0.2 BuIM:HNTf₂. A specific capacitance of ∼154 F/g was measured at 1 mV/s. Fitting the relationship between CV’s current (i_p) and sweep rate (v) according to the power law i_p = av^b, where a and b are constants, results in b values ∼0.52 for 0.8:0.2 BuIM:HNTf₂, suggesting that a semi-infinite diffusion-controlled step is the rate-limiting step. By increasing the HNTf₂ content to a ratio of 0.5:0.5 BuIM:HNTf₂ [Fig. S2(c)], a significant drop in the capacitance was noticed with a specific capacitance of ∼28 F/g. This capacitance did not change noticeably by further increasing the HNTf₂ content to 0.2:0.8 BuIM:HNTf₂ even though the shape of the CV changed [Fig. S2(d)]. The b values were found to be ∼0.68 for 0.5:0.5 and ∼0.64 for 0.2:0.8 BuIM:HNTf₂, indicating the existence of a surface-controlled limiting step in addition to diffusion-controlled. The capacitance results are shown in Fig. 2. The higher capacitance for the 0.8:0.2 BuIM:HNTf₂ (normalized to the capacitance obtained from the 0.5:0.5 mixture at 1 mV/s) indicates that many of the ions from the protic ionic liquid are attached to the surfaces of the Ti₃C₂T_x-C12 MXene electrode in this composition. This composition, which gave a maximum capacitance in electrochemical measurements, is consistent with the normalized differential capacitance presented in Fig. 1 without applied voltage. Although the specific capacitance drops significantly at higher scan rates, the 0.8:0.2 composition still yields a higher capacitance even at higher scan rates among all the compositions. This observation indicates that there is a loss of cations with a corresponding gain in the number of anions when the scan rates increase, leading to a decrease in the electrochemical performance as predicted by classical density functional theory (cDFT).²³ 

CDFT also predicted a higher capacitance in a system with a larger pore size. Analogously to our previous experiment,¹⁴ where Ti₃C₂T_x-C12 with a d-spacing of 2.2 nm resulted in a higher capacitance while using 1M EMIMTFSI in acetonitrile solution, we believe that the combined effect of the layering of the ions together with the large d-spacing of the MXene results in the higher capacitance at 0.8:0.2 composition of BuIM:HNTf₂. The CVs for all four electrolyte compositions studied are shown in Fig. S2. All the compositions had a similar voltage window, from −0.7 to 0.7 V vs Ag wire. The 0.9:0.1 ratio mixture [Fig. S2(a)] showed very poor capacitance and high resistivity. The 0.5:0.5 and 0.2:0.8 electrolyte ratios showed similar values for capacitance at all scan rates (Fig. 2), but the shapes of the CVs are quite different. The CV shape of the 0.5:0.5 ratio [Fig. S2(c)] suggests that it is primarily double layer capacitance, while the 0.2:0.8 ratio [Fig. S2(d)] does have a redox peak visible at 0.1 V vs Ag wire during the charging portion of the CV, but the peak during the discharging is not very prominent. The optimal ratio, 0.8:0.2, has a redox peak at −0.2 V vs Ag wire during charging, but the peak during discharge is less prominent and appears at −0.1 V. These redox peaks suggest that the protic ionic liquid electrolyte is able to participate in redox as well as double layer capacitance. To understand the mechanism of higher capacitance at this BuIM:HNTf₂ = 0.8:0.2 composition, we performed energy-resolved elastic and quasi-elastic neutron scattering experiments using the high flux backscattering (HFBS)²⁴ and backscattering silicon (BASIS)²⁵ spectrometers, respectively.

Figures 3(a) and 3(b) display the mean square displacements (MSDs) (more details in the supplementary material) obtained using energy-resolved elastic neutron scattering from the ionic liquid BuIM in bulk and confined states (pristine and its mixture with HNTf₂), respectively. The MSD, ⟨u²⟩, of scattering particles was obtained from the Q-dependence of elastic scattering measured as a function of temperature using the fixed window scan. This Q-dependence of the elastic incoherent scattering can be related to ⟨u²⟩ as $I(T=T)I(T=4K)=exp−Q23⟨u2⟩$⁠, where the left-hand side of the equation is the elastic intensity at a given temperature (T) normalized to the intensity at the lowest temperature measured. In our case, the lowest temperature was 4 K. The MSD is calculated from the line slope obtained after plotting the natural logarithm of the normalized intensity vs Q². In a harmonic system, the Q-dependence of ⟨u²⟩ with temperature is approximated with the Debye–Waller factor arising due to thermal vibration, giving a monotonic decay of the intensity as a function of temperature.²⁶ When the time scale of the scattering particle’s motion is faster than that of the instrument’s resolution, a deviation from linearity arises, both in elastic intensity and in MSD as a function of temperature. The MSD of the pristine BuIM is much larger than in mixtures. Mixing HNTf₂ (salt) in BuIM lowers the MSD of the liquid. There is a systematic decrease in MSD when the salt concentration is increased [Fig. 3(a)]. This reduction in MSD is attributed to a possible complex formation between the ionic liquid and the salt. This decrease becomes much smaller when the concentration of BuIM reaches 50% and lower. At a certain concentration level, when there is an equilibrium between HNTf₂ and the two other ionic species, BuIMH⁺ and NTf₂⁻, further addition of HNTf₂ does not contribute much to reducing the MSD of the protic ionic liquid. However, when those mixtures are confined in Ti₃C₂T_x-C12 MXene, the MSD obtained from the mixture with 80% of BuIM is the highest compared to all the other compositions, where the MSDs are nearly the same [Fig. 3(b)]. This higher mobility is also reflected in the diffusion coefficients extracted from the QENS measurements. The half-width at half maxima (HWHM) obtained from the QENS spectra [Figs. S3(a) and S3(b)] of 0.8:0.2 (BuIM:HNTf₂) mixture are higher compared to all other compositions, both at 300 K [Fig. 4(a)] and at 360 K [Fig. 4(b)].

The diffusion coefficients obtained from a simple Fickian model give the maximum value [Figs. 4(c) and 4(d)] at this 0.8:0.2 (BuIM:HNTf₂) composition at both the measured temperatures. Similarly, besides the current protic ionic liquid system, a higher diffusivity at 0.8:0.2 (BuIM:HNTf₂) ratio of two different aprotic ionic liquid mixtures in confinement has been reported.²⁷ In that ionic liquid system, the higher diffusivity of the cation was attributed to the structural modification of the ions at the surface of the matrix (onion-like carbon) due to the replacement of a larger anion by a smaller one, resulting in excess adsorption of cation on the surface. This excess adsorption allows other cations to move away from the surface in the less crowded environment, thus exhibiting a higher diffusivity. When more cations are attached to the surface of the matrix, a higher elastic incoherent scattering fraction (EISF) is obtained. However, in a 0.8:0.2 (BuIM:HNTf₂) mixture system, even though the diffusivity is higher at this composition in confinement, the EISF (Fig. S4) decreases gradually as the concentration of BuIM is decreased in the mixture, without showing a maximum. This decrease in EISF must be due to the progressive displacement of the cations from the MXene surface with the addition of salt. So, when more salt is in the electrolyte, more cations go off the walls (hence a smaller EISF). Since the MXene surface is rich with surface functional groups,²⁸ we anticipate the formation of an electrical double layer (EDL), as illustrated by cDFT under applied potential, on its surface.²³ Furthermore, the density functional theory (DFT) also suggests that the presence of confined BuIM molecules in pores changes significantly at 20 mol % HNTf₂.²³ Therefore, the formation of EDL, comprising NTf₂⁻ counterions and BuIMH⁺ coions, results in fewer BuIM molecules away from the surfaces, which exhibit a higher diffusivity. However, there is a monotonic decrease in EISF, with no maximum at 0.8:0.2 (BuIM:HNTf₂) composition in the confined system. Even though it requires further investigation, the lower EISF observed in the current study is associated not with lower diffusivity in the middle of the pores (as it usually is) but with higher diffusivity in the middle. This unusual behavior can be attributed to the interactions between these three chemical species and the solvent molecules (BuIM), which progressively decrease the number of immobile hydrogen-bearing species on the surface of the MXene, resulting in a BuIM concentration-dependent EISF. However, the effect of temperature on the EISF [Fig. S4(b)] is larger when the composition is higher in BuIM, suggesting the presence of more immobile ions when the BuIM remains as a solvent. Note that the QENS spectra [Fig. S5(a)] of the pristine Ti₃C₂T_x-C12 MXene also show some temperature-dependent quasi-elastic broadening. The HWHM of the peaks at 300 and 360 K are found to be Q-independent [Fig. S5(b)], which means that we are probing the localized motion of the tetradodecylammonium salt, which was used as an intercalant in this Ti₃C₂T_x MXene.

In summary, we have probed the microscopic dynamics and the electrochemical performance of a protic ionic liquid confined in pre-intercalated Ti₃C₂T_x-C12 MXene employing quasi-elastic neutron scattering and electrochemical measurements. We found an optimal composition of an ionic liquid, BuIM, with a salt, HNTf₂, in the ratio of 0.8:0.2 (BuIM:HNTf₂), giving an enhanced capacitance and a higher diffusivity. The higher capacitance results from the accumulation of more of the BuIMH⁺ on the surface of the MXene electrode, replacing more of the solvent (BuIM) molecules. This replacement provides more freedom for BuIM molecules to move due to the conversion of more of the solvent molecules to protic ionic liquid. This study’s findings will help design electrolyte systems involving ionic liquids and salts forming protic ionic liquids, resulting in fast ion transport while using pre-intercalated MXene as an electrode in the supercapacitor configuration.