Ionic liquids behave as dilute electrolyte solutions

Abstract

We combine direct surface force measurements with thermodynamic arguments to demonstrate that pure ionic liquids are expected to behave as dilute weak electrolyte solutions, with typical effective dissociated ion concentrations of less than 0.1% at room temperature. We performed equilibrium force–distance measurements across the common ionic liquid 1-butyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide ([C₄mim][NTf₂]) using a surface forces apparatus with in situ electrochemical control and quantitatively modeled these measurements using the van der Waals and electrostatic double-layer forces of the Derjaguin–Landau–Verwey–Overbeek theory with an additive repulsive steric (entropic) ion–surface binding force. Our results indicate that ionic liquids screen charged surfaces through the formation of both bound (Stern) and diffuse electric double layers, where the diffuse double layer is comprised of effectively dissociated ionic liquid ions. Additionally, we used the energetics of thermally dissociating ions in a dielectric medium to quantitatively predict the equilibrium for the effective dissociation reaction of [C₄mim][NTf₂] ions, in excellent agreement with the measured Debye length. Our results clearly demonstrate that, outside of the bound double layer, most of the ions in [C₄mim][NTf₂] are not effectively dissociated and thus do not contribute to electrostatic screening. We also provide a general, molecular-scale framework for designing ionic liquids with significantly increased dissociated charge densities via judiciously balancing ion pair interactions with bulk dielectric properties. Our results clear up several inconsistencies that have hampered scientific progress in this important area and guide the rational design of unique, high–free-ion density ionic liquids and ionic liquid blends.

Ionic liquids are fluids composed solely of ions (1, 2). Much of the recent scientific interest surrounding ionic liquids derives from the fact that ionic liquids have been demonstrated for numerous applications, such as safe, high-efficiency electrochemical storage devices (3, 4), self-assembly media (5), and lubrication (6). A key paradigm within ionic liquids research is that the physical properties of ionic liquids can be controlled to an unprecedented degree through the judicious design of cation–anion pairs (1–9). Thus, ionic liquids are known as “designer” solvents/materials (1). However, fully realizing this advantage requires the development of a comprehensive framework that can be used to rationalize the relationship between an ionic liquid’s molecular structure and its bulk and interfacial behavior and properties, which are governed by a complex interplay of coulomb, van der Waals, dipole, hydrogen bonding, and steric interactions (10, 11).

Recent work has greatly progressed a fundamental understanding of ionic liquids at charged interfaces, but there are currently inconsistencies between experiment and theory (4, 7–9, 12–23); this is particularly true for the ranges of surface-induced ordering and electrostatic screening. For example, a comparison of the values obtained from ionic conductivity and ion diffusion coefficient measurements with ionic liquid electrolytes led to the postulation that ionic liquids behave as highly dissociated electrolytes with an immense concentration of free ions in solution, where typical ionic liquids are expected to exhibit effective free-ion concentrations on the order of 50–80% of the maximum ion density (24, 25). With such high effective free-ion concentrations, the Debye lengths for typical ionic liquids, κ⁻¹, should be on the order of 0.1 Å (4, 9), which is at least 1–2 orders of magnitude smaller than the molecular dimensions of even the smallest ionic liquid cations and anions. This discrepancy is typically provided as evidence that ionic liquids should completely electrostatically screen charged surfaces within several bound (Stern) ion layers (1–2 nm), beyond which charged surfaces no longer impact ion density or ordering (9, 21–23). This screening behavior has been observed in theoretical studies involving simplified model ionic liquids (9, 21–23).

In contrast, X-ray reflectivity, atomic force microscopy (AFM), and surface forces apparatus (SFA) measurements have shown that strong ion–surface interactions can induce ion ordering that extends up to 10 nm into solution (per surface) (12–20). Notably, the sign and magnitude of the charge density intrinsic to solid surfaces are crucial determinants for the ranges of the surface-induced ordering and for the templating of the specific structure formed by the bound ion layers (19, 20). For example, a combined X-ray reflectivity–molecular dynamics study on ionic liquid–solid interfaces demonstrates that replacing a neutral graphene surface with a negatively charged muscovite mica surface leads to a substantial increase in the range of surface-induced ion ordering and changes the structure of the bound ion layers from a mixed cation–anion layering motif to an alternating cation–anion layering structure (20).

Furthermore, a study of the differential capacitance of ionic liquids at electrode surfaces indicates that ionic liquids electrostatically screen charged surfaces through the formation of electric double layers consisting of distinct inner (bound/compact) and outer (diffuse) double layers (26). This screening mechanism is consistent with proposed models, analogous to the Gouy–Chapman–Stern model for dilute electrolyte solutions (27, 28). Nevertheless, the prevailing picture is that ionic liquids should behave as highly concentrated, but largely dissociated, electrolyte solutions.

Here, we report quantitative equilibrium force–distance measurements of two dissimilarly charged surfaces: single-crystalline muscovite mica and molecularly smooth, polycrystalline gold (0.2 nm rms roughness), interacting across an ionic liquid—1-butyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide ([C₄mim][NTf₂]) (Fig. 1B)—using the electrochemical (EC) SFA technique described in Fig. 1A. These results were quantitatively modeled using the van der Waals and electrostatic double-layer forces of the Derjaguin–Landau–Verwey–Overbeek (DLVO) theory with an additive monotonic repulsive steric (entropic) ion–surface binding force. Additionally, we propose a general, molecular-scale framework that can be used to predict the temperature-dependent effective free-ion density for ionic liquids, provided that the ion dissociation energy (interaction potential) in vacuum, E_d, and the low-frequency relative permittivity, ε, of an ionic liquid can be independently determined.

Fig. 1.

(A) Schematic of the electrochemical surface forces apparatus (EC SFA). The electrochemical potential of the top gold electrode surface was controlled through a custom-built electrochemical SFA attachment based on a three-electrode setup (32, 43). When immersed in [C₄mim][NTf₂], muscovite mica is known to acquire a negative surface charge through the dissociation of surface-bound K⁺ ions; the resultant K⁺ concentration within the bulk ionic liquid is negligible (17, 20). The mica surface potential, Ψ_M, was assumed to be independent of applied potentials, ΔU. The distance between the two surfaces, D, was defined with respect to the contact of the gold and mica surfaces in the absence of [C₄mim][NTf₂], where D = 0. Equilibrium (static) force measurements were performed at constant applied potentials, ΔU, and the open circuit potential (OCP), where the potential was sensed to be −140 mV. Degradation peaks in postexperimental cyclic voltammetry measurements were used to account for small shifts in potential referencing across experiments (Fig. S2). All results described here were measured at applied potentials, ΔU, that did not induce faradaic reactions at the gold electrode surface. (B) Molecular structure and dimensions of the 1-butyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide ([C₄mim][NTF₂]) cation and anion. Conformations and dimensions were calculated through the conjugate gradient energy minimization of the chemical structures using the universal force field (44).

Our results further show that the relative charges and potentials of confining surfaces, as well as the chemical nature of each of the two ion species in single-component ionic liquids, can impact the qualitative nature of surface–surface interactions across nanoconfined ionic liquids by resulting in the presence (or absence) of long-range electrostatic interactions. These results provide insights into current inconsistencies regarding the ranges of electrostatic screening and surface-induced ordering in ionic liquids, and thus have important implications for the utilization of ionic liquids in numerous applications, including as electrolytes in electrochemical storage devices, self-assembly media, and lubricants.

Results and Analysis

Equilibrium Force–Distance Measurements.

Representative force–distance profiles, F(D), for the open circuit potential (OCP) (the potential was sensed) and applied potentials of ΔU = −500, 0, and +500 mV are shown in Fig. 2. For the measurements that were performed at constant ΔU, the system was allowed to equilibrate for at least 5 min at each potential before beginning the equilibrium F(D) measurements. Immediately, after applying each potential, the magnitudes of the current densities measured at the gold electrode were found to decay to a negligible baseline current density within several seconds (on the order of ∼10–50 nA/cm²), consistent with the charging of an electric double layer (i.e., no faradaic currents were observed).

Fig. 2.

Points correspond to experimental data and solid lines correspond to forces calculated using Eq. 1. (A) Representative equilibrium force–distance, F(D), profiles measured between mica and gold on approach. Measurements were taken while monitoring the open circuit potential (OCP) (red) and at three different applied potentials, ΔU, across the ionic liquid [C₄mim][NTf₂]. Inset shows the same data for distances of D < 10 nm. Equilibrium F(D) profiles measured on retraction are reversible and are shown in Fig. S1. (B and C) Distance dependence of the individual and total calculated interaction potentials for (B) ΔU = OCP and (C) ΔU = +500 mV. Calculations for ΔU = −500 and 0 mV can be found in Fig. S3.

Static (equilibrium) F(D) measurements were taken by using a piezoelectric crystal to move the gold surface in constant voltage steps of ∼1.5–3.0 V, which corresponds to constant distance, D, steps of between 3 and 5 nm. The surfaces were allowed to equilibrate at each point for 45 s before moving to the next D. The voltage–distance calibration for the piezoelectric crystal was determined at surface separation distances, D, where no interaction forces were measured. This procedure allowed us to determine the equilibrium interaction forces across the surfaces at each measured D, and the experimental error of the measured data points are F = ±0.01 mN/m and D = ±0.2 nm. All equilibrium F(D) measurements were seen to be reversible, as long as the applied forces did not result in significant deformations of the surfaces (Fig. S1).

Force Curve Analysis.

The measured force–distance profiles exhibit a long-range interaction regime (D out to 35 nm) that is solely attributable to electrostatic forces. The forces measured in this regime asymptotically decay with an exponential decay length of 11 ± 2 nm and are well-fitted by equations describing overlapping diffuse electric double layers. For surface separations D < 3 nm, short-range forces, arising from a nontrivial interplay of van der Waals forces, ion–surface binding forces, and forces induced by ion structuring begin to dominate the force profiles. The solid lines overlaying the experimental points in Fig. 2 were fitted using Eq. 1:

where

and where A_H (in joules) is the Hamaker constant, D (in meters) is the surface separation, ε (11.6) is the relative permittivity of [C₄mim][NTf₂] (4), κ⁻¹ (in meters) is the Debye screening length, Ψ_M (in volts) is the outer Helmholtz plane (OHP) mica potential, σ_Au (in coulombs per square meter) is the OHP gold charge density, λ₀ (in meters) is a steric decay length, W₀ (in joules per square meter) is a fitted steric prefactor, T (295 K) is the temperature, e (in coulombs) is the elementary charge, n (in meters⁻³) is the number density of dissociated ions, k (in joules per kelvin) is Boltzmann’s constant, and ε₀ (in farads per meter) is the permittivity of free space. The parameter values can be found in Table 1 and Table S1 with a detailed discussion of the fitting procedure provided in SI Text.

Table 1.

Measured, calculated, and fitted parameters using Eq. 1

Eq. 1 parameters	Applied potential, ΔU; −500 mV	Sensed potential, OCP; −140 mV	Applied potential, ΔU; 0 mV	Applied potential, ΔU; +500 mV
Gold surface charge density, σAu; mC/m2	0 ± 0.1	0 ± 0.1	0.4 ± 0.1	3.6 ± 0.1
Mica surface potential, ΨM; mV	−198 ± 8	−191 ± 8	−196 ± 8	−193 ± 5
Debye length, κ−1; nm	13 ± 2	12 ± 2	12 ± 2	10 ± 2
Total dissociated ion concentration, Ci; M	0.8 ± 0.3 × 10−4	0.9 ± 0.3 × 10−4	0.9 ± 0.3 × 10−4	1.4 ± 0.3 × 10−4

Eq. 1 represents a “DLVO-type” interaction potential—a well-established equation for colloidal interactions—where the first term is the mica–gold van der Waals interaction, the second is a linearized Poisson–Boltzmann equation for overlapping electric double layers (29, 30), and the third is an empirical steric repulsion potential arising from surface-bound ions. For dissimilar surfaces, the electric double-layer interactions can be attractive for all separations, repulsive for all separations, or change sign from attractive to repulsive (or vice versa) at finite distances. Fig. 2 B and C shows the relative magnitude and distance dependence of each of the three calculated force contributions, as well as the total calculated force–distance potential (the contributions are additive) for ΔU = +500 mV and OCP. Importantly, although the electrochemical potential that is applied to the gold surface, ΔU, ultimately determines the OHP charge density of the isolated gold surface, σ_Au, there is no equation that directly relates the two values; the value of σ_Au must be determined by fitting the measured force–distance profiles.

The long-range interaction forces are well fitted by a Debye screening length of κ⁻¹ = 11 ± 2 nm for all applied potentials, which agrees with the asymptotic form of Eq. 1 that decays as e^−κD, where κ⁻¹ is the Debye length. This long-range effect indicates that the surface-bound ions (i.e., the ions within the inner Helmholtz plane or Stern layer) are not able to fully screen the charged surface. The level of agreement between Eq. 1 and the measured force–distance profiles (Fig. 2) indicates that the “residual” potential propagating past the bound ion layers (two charged surfaces) is screened by an exponentially decaying diffuse electric double-layer–type mechanism (Fig. 3). Because ionic liquids are solvent-free, the diffuse double layer must consist of an effectively neutral, coordinated cation–anion network (i.e., like “solvent molecules”) that exists in equilibrium with a small fraction of effectively dissociated ions (Fig. 3)—analogous to the double-layer electrode screening found for dilute solvent–salt mixtures (31, 32), where the plane of origin for this electrostatic interaction is the OHP. For [C₄mim][NTf₂], the OHP is shifted out from the surfaces by between 0 ± 0.1 nm (OCP, ΔU = 0 and +500 mV) and 1.5 ± 0.1 nm (ΔU = −500 mV) per surface, depending on the relative surface potentials.

Fig. 3.

Diagram of the diffuse electric double layers formed by [C₄mim][NTf₂] ions for ΔU = +500 mV. Analogous electric double layers are formed for all other potentials. Effectively dissociated pairs of ions are generated via the equilibrium dissociation reaction above. Light gray shading in the interface outside the outer Helmholtz plane (OHP) represents a dielectric fluid consisting of an effectively neutral cation–anion network. The lines overlaying the diagram indicate the relative concentration gradient of dissociated ions: blue represents cations; red, anions; and green, C_cation = C_anion = 1/2 C_i,bulk. The bound and diffuse electric double layers are enriched in cations (blue) at the negative mica surface and anions (red) at the positive gold surface.

Strikingly, the long-range portions of the measured equilibrium force–distance profiles for applied potentials, ΔU, that are more positive than the OCP are fundamentally different from the profiles measured for applied potentials that are more negative than the OCP (Fig. 2). For ΔU > OCP, the force profiles exhibit a long-range attraction that is consistent with a positive OHP charge at the gold surface and a negative OHP potential at the mica surface. Therefore, the bound ion layers formed at both the mica and gold surfaces do not fully electrostatically screen the surfaces.

For ΔU ≤ OCP, the force profiles exhibit a weaker long-range attraction, whereas an electrostatic repulsion would be expected for the case of negative gold charging. Shifting the OHP out by 3 nm for ΔU = −500 mV, hence accounting for the fact that a thicker layer of bound ions is required to screen the increasingly negative gold surface charge density, brings the range and magnitude of this electrostatic attraction into quantitative agreement with the ΔU = OCP (Fig. 2): both force–distance profiles exhibit κ⁻¹ = 12 ± 2 nm, Ψ_M of approximately −195 mV, and σ_Au = 0 ± 0.1 C/m². Thus, the bound ion layer at the gold surface fully screens the electrode for ΔU = OCP and −500 mV, whereas the bound ion layer at the mica, again, does not fully screen the surface.

From these results, we discern that bound ion layers enriched in [C₄mim]⁺ cations electrostatically screen polycrystalline gold surfaces more effectively than bound ion layers enriched in [NTf₂]⁻ anions. This contrast in screening efficiency likely arises from the substantial differences between the molecular structures of the [C₄mim]⁺ and [NTf₂]⁻ ions: [C₄mim]⁺ cations are largely planar, contain well-defined polar and nonpolar regions, and exhibit a lower degree of conformational flexibility compared with the bulkier [NTf₂]⁻ anions (Fig. 1B). These observations are consistent with studies of ionic liquids at the free vacuum interface, where the size, polarity, and shape asymmetry of ionic liquid cations and anions are shown to differently impact their structuring at the free vacuum interface, with ionic liquids forming increasingly charge-separated electric double layers at the ionic liquid–vacuum interface as the degree of cation–anion asymmetry is increased (33, 34).

The short-range portions (D < 3 nm) of the force–distance profiles are fitted by the superposition of a mica–gold van der Waals attraction and a monotonic steric “hydration-type” repulsion (decay length, 0.6 ± 0.15 nm) corresponding to a characteristic length of the adsorbed ions (Fig. 1B). The superposition of these two short-range force contributions qualitatively captures the transition from the force regimes where the long-range diffuse double-layer forces are the sole contributor to the measured force–distance profiles, to the regimes where the force profiles are dominated by short-range, specific ion–ion and ion–surface interactions. This empirical monotonic repulsion is, however, unable to fully describe the complex, nonmonotonic features of the short-range portions of the measured force–distance profiles. In particular, the force–distance profiles measured at the OCP (Fig. 2) exhibit nonmonotonic behavior that is indicative of complex ion ordering. As a result, the agreement between the measured and theoretically modeled forces progressively diverges as the surface–surface separation distance, D, decreases from D = 1 to D = 0 nm. The physical origins of the forces measured in the short-range portions of the profiles are extremely complex because continuum theories of colloidal interactions no longer apply, and the trends exhibited at D < 3 nm will be addressed in a future publication.

Thermodynamic Model for the Effective Free-Ion Concentration in Ionic Liquids.

Our results indicate that the general electrostatic screening behavior of [C₄mim][NTf₂] (long-range) is consistent with a diffuse electric double-layer mechanism when the Stern layers, “solvent molecules,” and dissociated ions (Fig. 3) are properly defined. Remarkably, the 11 ± 2-nm Debye length, κ⁻¹, for this system indicates that the concentration of freely dissociated ions in [C₄mim][NTf₂] at 295 K is C_i = 1.1 ± 0.5 × 10⁻⁴ M (mol/L), corresponding to 0.003% dissociation. Therefore, the dissociation constant is K_d = 7.3 ± 0.7 × 10⁻¹⁰ (pK_d = 9.14).

The value for K_d appears to be explicable in terms of the energetics of thermally dissociable ions interacting in a dielectric medium, providing further evidence for electric double-layer formation by ion dissociation: as a first approximation, the equilibrium mole fraction (solubility) of dissociated (uncoordinated) pairs of ions (for electroneutrality) in a dielectric medium of associated (electrically neutral) [C₄mim][NTf₂] ions, X_d, can be calculated from the Boltzmann distribution as follows:

where Δµⁱ (in joules) is the chemical potential change for ions going from the associated to dissociated state in a dielectric medium, E_d (in joules) is the ion dissociation energy (interaction potential) in vacuum, k (in joules per kelvin) is Boltzmann’s constant, T (295 K) is the temperature, and ε (11.6) is the relative permittivity of [C₄mim][NTf₂] (4). Although Eq. 2 only accounts for low-frequency dielectric screening of the interaction potential, coulombic interactions typically account for greater than 90% of the total interaction energy between ionic liquid ions (7, 35). Electronic structure calculations show that E_d = 315.26 kJ/mol (128.50 kT per pair of ions) for [C₄mim][NTf₂] (35), and this value yields a dissociated ion concentration of C_i = 1.2 × 10⁻⁴ M (mol/L), agreeing with the experimental results to within 10%.

Discussion and Conclusions

Although our results may, at first glance, appear to stand in contradiction to previous SFA and AFM studies on ionic liquids that found only short-range structural forces (13–19), we discuss why our higher-resolution results not only are consistent with the existing scientific literature but also provide insights into the apparent discrepancies that exist regarding the ranges of electrostatic screening and surface-induced molecular ordering in ionic liquids.

Notably, results from AFM measurements with in situ electrochemical control of the substrate surface indicate that the near-surface structuring of the ionic liquid ions depends on the magnitude and sign of the applied electrochemical potentials (19). The authors of this study concluded that electrode surface potentials are entirely screened by adsorbed ion layers, and ruled out screening by a diffuse electric double layer (19). However, the forces arising from the overlap of two diffuse electric double-layers scales as the radius of the AFM tip (R of ∼10⁻⁹ m) for the sphere-on-flat geometry used in AFM studies. The crossed-cylinder geometry used in the SFA leads to the same local sphere-on flat scaling for the measured forces, where the radius of the sphere, R, is equal to the geometric mean of the curvature of the cylinders (R of ∼10⁻² m). As a result, the magnitude of the diffuse double-layer forces that we measured in this series of experiments, which are on the order of 10⁻⁵ N after accounting for the radius normalization (well within the experimental accuracy of the SFA technique), would correspond to a force of 10⁻¹² N for a nanoscale AFM tip, which is 3 orders of magnitude lower than the 10⁻⁹ N sensitivity reported for current AFM studies on ionic liquids (18, 19).

Additionally, previous SFA studies on ionic liquids used symmetric ceramic surfaces to explore the electric double layers formed by ionic liquids (13–17). The force profiles that have been consistently measured across ionic liquids between symmetric interfaces are repulsive and oscillatory in nature (13–17). The measurement of oscillatory force–distance profiles across fluids arises from the presence of layered, phase-separated regions within confined fluid films (36–38). However, short-range oscillatory forces, when present, are typically only one contribution to the forces that are measured across confined liquid films (36, 37). For example, in SFA measurements of two symmetric mica surfaces interacting across a 1.0 × 10⁻³ M aqueous KCl solution, oscillatory forces were shown to be superimposed on a steep, short-range monotonic double layer and/or hydration repulsion at surface–surface separation distances, D, of less than 2 nm (36). The oscillatory force profiles measured across ionic liquids also appear to be superimposed on monotonically increasing repulsions that are consistent, in both range and magnitude, with the electric double-layer forces measured in the current study (13–17).

In our system, we expect the isolated single-crystalline mica surfaces to induce alternating cation–anion layering motifs that extend into the [C₄mim][NTf₂], as has been observed previously (17, 20). However, the ion structures templated by the polycrystalline gold surfaces will necessarily be more disordered than the structures induced by single-crystalline mica surfaces. Because the presence of even a subnanometer scale surface roughness can be sufficient to smear out the discrete oscillatory character of force–distance profiles (36, 39), we infer that the polycrystalline nature of the gold surfaces used in our experiments is responsible for the suppression of pronounced discrete oscillatory instabilities. Nevertheless, the trends in the short-range repulsions that arise from the overlap of the strongly bound ion layers is consistent with previous work on ionic liquids and dilute electrolyte solutions (13–17, 31, 32, 36).

We do, however, note that the room temperature effective dissociated ion concentration on the order of 50–80% of the total ionic density that has been postulated by comparing ionic conductivity to diffusivity measurements for ionic liquids (i.e., the “ionicity” approach) (24, 25) stands in stark contrast to the freely dissociated ion concentration of 0.003% calculated from our equilibrium force measurements. However, conductivity and diffusivity measurements inherently include kinetic (dynamic transport) effects, and thus are not directly related to the equilibrium (static) association and dissociation (i.e., the K_d value) of pairs of ions (40). Furthermore, the ionicity approach to predicting the effective concentration of dissociated ions in ionic liquids (24, 25) never explicitly considers the energetics of dissociating ionic liquid ions. Therefore, although the ionicity approach has great utility with regards to studying the dynamic properties of ionic liquids, we reason that these ratios do not appear to be directly proportional to the equilibrium effective ionic concentration that contributes to electrostatic screening in ionic liquids.

In conclusion, our results show that the ionic liquid [C₄mim][NTf₂] screens charged surfaces through the formation of both bound (Stern) and diffuse electric double layers, where the diffuse double layer is composed of effectively dissociated [C₄mim]⁺ and [NTf₂]⁻ ions. Additionally, we used the energetics of thermally dissociating ions in a dielectric medium to quantitatively predict the equilibrium for the effective dissociation reaction of [C₄mim][NTf₂] ions, in excellent agreement with the measured value (Debye length). The theories that we have used during our analysis (the theory of electric double-layer forces, the Boltzmann distribution for thermally dissociating ions in a dielectric medium) are general, thermodynamically rigorous theories; thus, we expect these observations to be widely applicable to other ionic liquids. We therefore deduce that each combination of ionic liquid ions should have a unique (temperature-dependent) dissociation constant, K_d, that can be used as a design parameter to help rationalize and tune both the bulk and interfacial interactions of the liquid. We also suggest that K_d can be calculated with reasonable accuracy by independently knowing ε and E_d for each pair of ions, which makes comparing the effective dissociated ion concentration and electrostatic screening lengths (and resultant double-layer forces) for a wide range of ionic liquid ion pairs a straightforward task.

Materials and Methods

The ionic liquid [C₄mim][NTf₂] was synthesized, purified, dried, and characterized using a previously established procedure for the preparation of electrochemical-grade ionic liquids (41). The [C₄mim][NTf₂] was clear and colorless, and NMR and UV-Vis spectroscopy were used to establish that the [C₄mim][NTf₂] was of electrochemical-grade purity. Raman spectroscopy found that the ionic liquid has a negligible fluorescence background, further confirming its purity, as ionic liquid impurities are often strongly fluorescent. After synthesis and drying, Karl Fischer titration was used to determine that the (trace) water content of the [C₄mim][NTf₂] was 7 ppm. The [C₄mim][NTf₂] was stored under vacuum until use in the EC SFA experiments.

Before each EC SFA experiment, the [C₄mim][NTf₂] was dried a second time by heating in a vacuum oven under vacuum at 373 K for at least 48 h. Following drying, the ionic liquid was immediately injected into a gas-tight SFA through an injection port under dry nitrogen purge conditions—the nitrogen pressure flowing through the SFA was higher than ambient pressures. After injection, the SFA was resealed, and all experiments were performed in the presence of a reservoir of the desiccant phosphorus pentoxide (P₂O₅) that was placed in the SFA chamber. Each experiment lasted between 12 and 36 h, and the P₂O₅ that was removed from the box was unsaturated at the conclusion of each experiment. Additionally, postexperimental cyclic voltammetry measurements showed that the [C₄mim][NTf₂] exhibited an electrochemical window in excess of 4 V (Fig. S2), and previous work has shown that the electrochemical window of [C₄mim][NTf₂] is 4.3 V when dried, but only 2.9 V when saturated with water (42).