Influence of humidity on performance and microscopic dynamics of an ionic liquid in supercapacitor

We investigated the influence of water molecules on the diffusion, dynamics, and electrosorption of a room temperature ionic liquid (RTIL), $\left[\mathrm{BMI}{\mathrm{m}}^{+}\right]\left[\mathrm{T}{\mathrm{f}}_2{\mathrm{N}}^{-}\right]$, confined in carbide-derived carbon with a bimodal nanoporosity. Water molecules in pores improved power densities and rate handling abilities of these materials in supercapacitor electrode configurations. We measured the water-dependent microscopic dynamics of the RTIL cations using quasielastic neutron scatting (QENS). The ionic liquid demonstrated greater mobility with increasing water uptake, facilitated by the nanoporous carbon environment, up to a well-defined saturation point. We concluded that water molecules displaced RTIL ions attached to the pore surfaces and improved the diffusivity of the displaced cations. This effect consequently increased capacitance and rate handling of the electrolyte in water-containing pores. Our findings suggest the possible effect of immiscible co-solvents on energy and power densities of energy storage devices, as well as the operating viability of nonaqueous supercapacitor electrolytes in humid environments.

Figure 1

Gas sorption isotherms of the oxidized bimodal CDC showing the available surface area along with the total pore volume (a) before filling the IL and (b) after filling the IL

Figure 2

Electrochemical performance of $\left[\mathrm{BMI}{\mathrm{m}}^{+}\right]\left[\mathrm{T}{\mathrm{f}}_2{\mathrm{N}}^{-}\right]$ in CDC as a function of ${\mathrm{H}}_2\mathrm{O}$ exposure time. (a) Rate handling comparison for CDCs over the entire 0.5–1000 mV/s sweep rate range. (b) Three-electrode quasireference measurements with symmetrical electrodes, carried out at 10 mV/s (summarized in Table 2). (c) Nyquist plot. The inset is the zoomed in region of high-frequency impedance spectra. (d) Two-electrode cyclic voltammograms measured using a 1.0 mV/s sweep rate, as compared to those measured at (a). (e) 50 mV/s sweep rate. (f) Bode impedance plot that compares the impedance phase angle as a function of oscillating frequency.

Figure 3

(a) Elastic neutron scattering intensity of the $\left[\mathrm{BMI}{\mathrm{m}}^{+}\right]\left[\mathrm{T}{\mathrm{f}}_2{\mathrm{N}}^{-}\right]$ confined in bimodal CDC as a function of temperature. (b) Mass normalized $S$($Q, E$) of all samples showing a drop (as marked by a dotted ellipsoid) in intensity after water exposure. QENS spectra at (c) $Q=0.68{\AA }^{-1}$ and (d) $Q=1.16{\AA }^{-1}$ at 290 K. The symbols correspond to the data collected at different exposure times as indicated in the figure. Solid lines are fit using a Lorentzian function and the dotted lines (black) show the instrument resolution measured at 4 K. Data has been normalized to the highest intensity for the purpose of comparison and truncated at 0.1 arbitrary units for clarity. Error bars throughout the text represent one standard deviation.

Figure 4

(a) FWHMs extracted from a single Lorentzian fit of $S$($Q$, $E$) of the confined $\left[\mathrm{BMI}{\mathrm{m}}^{+}\right]\left[\mathrm{T}{\mathrm{f}}_2{\mathrm{N}}^{-}\right]$ as a function of ${\mathrm{D}}_2\mathrm{O}$ exposure time are fitted using a jump diffusion model [represented by solid lines (red)]. (b) Corresponding diffusion coefficients of cations (left $y$ axis) and average elastic incoherent scattering fraction (right $y$ axis). The dotted lines are a guide to the eye.

Figure 5

Dynamic susceptibility (intensity divided by Bose factor) of QENS spectra at a representative $Q$ of $1.16{\AA }^{-1}$ showing a raising tail of the peak originating from the cation dynamics visible in the accessible dynamic range.

Figure 6

Residence times extracted from a single Lorentzian fit of $S$($Q, E$) of the confined $\left[\mathrm{BMI}{\mathrm{m}}^{+}\right]\left[\mathrm{T}{\mathrm{f}}_2{\mathrm{N}}^{-}\right]$ as a function of ${\mathrm{D}}_2\mathrm{O}$ exposure time form the model fit.

Figure 7

Visualizations of representative snapshots of MD simulations of confined humid $\left[\mathrm{BMI}{\mathrm{m}}^{+}\right]\left[\mathrm{T}{\mathrm{f}}_2{\mathrm{N}}^{-}\right]$. Carbon atoms making up pore walls are silver spheres and water molecules are drawn with conventional CPK coloring. For simplicity, ionic liquid molecules were represented with a surface model; they are not explicitly shown, but the area in which they fill is outlined with a transparent gray surface. Water composition in each image are (a) 0.2, (b) 0.48, and (c) 0.60 mole fraction with corresponding 1, 3.8, and 6.1 wt. %. At low water concentrations (a), some water is found in the bulk and those in the pore tend to prefer the pore wall. At moderate concentration (b), more water is found throughout, including filling the middle of the pore. Comparison of these images shows a stronger preference of water molecules to reside in the pore (both near the walls and in the middle), which implies that confinement increases the solubility of water in IL. The length of these MD simulations is sufficient for water molecules to sufficiently explore both regions of the simulation box. These results therefore imply that the solubility of water in confined IL exceeds that of bulk IL.

Figure 8

Comparison of water densities in regions of pore simulation boxes compared to bulk densities in independent bulk simulations. To facilitate comparison, initial fluid compositions between compared simulations were equal. Throughout, the density of water inside the pore was greater than the bulk-like region of the pore simulations.

Figure 9

Number density profiles of RTIL ions and water across the width of the model pore used in MD simulation at different mole fractions ($X_w$). Inset in the figure shows the relationship between cation diffusivity and water concentration. The energetic coefficient of the water-carbon interaction (${\varepsilon }_{\mathrm{water}-\mathrm{wall}}$) has been doubled in each simulation. As water molecules are added to the simulations, they settle on the pore walls before filling the entire pore.