Novel Na+ Ion Diffusion Mechanism in Mixed Organic–Inorganic Ionic

Article pubs.acs.org/JPCC

Novel Na+ Ion Diffusion Mechanism in Mixed Organic−Inorganic Ionic Liquid Electrolyte Leading to High Na+ Transference Number and Stable, High Rate Electrochemical Cycling of Sodium Cells. Maria Forsyth,*,† Hyungook Yoon,† Fangfang Chen,† Haijin Zhu,† Douglas R. MacFarlane,‡ Michel Armand,§ and Patrick C. Howlett† †

*ARC Centre of Excellence for Electromaterials Science (ACES), Institute for Frontier Materials (IFM), Deakin University, Burwood, Victoria 3125, Australia ‡ School of Chemistry, Monash University, 3800 Melbourne, Victoria Australia § CIC Energigune, 01510 Vitoria, Spain S Supporting Information *

ABSTRACT: Ambient temperature sodium batteries hold the promise of a new generation of high energy density, low-cost energy storage technologies. Particularly challenging in sodium electrochemistry is achieving high stability at high charge/ discharge rates. We report here mixtures of inorganic/organic cation fluorosulfonamide (FSI) ionic liquids that exhibit unexpectedly high Na+ transference numbers due to a structural diffusion mechanism not previously observed in this type of electrolyte. The electrolyte can therefore support high current density cycling of sodium. We investigate the effect of NaFSI salt concentration in methylpropylpyrrolidinium (C3mpyr) FSI ionic liquid (IL) on the reversible plating and dissolution of sodium metal, both on a copper electrode and in a symmetric Na/Na metal cell. NaFSI is highly soluble in the IL allowing the preparation of mixtures that contain very high Na contents, greater than 3.2 mol/kg (50 mol %) at room temperature. Despite the fact that overall ion diffusivity decreases substantially with increasing alkali salt concentration, we have found that these high Na+ content electrolytes can support higher current densities (1 mA/cm2) and greater stability upon continued cycling. EIS measurements indicate that the interfacial impedance is decreased in the high concentration systems, which provides for a particularly low-resistance solid-electrolyte interphase (SEI), resulting in faster charge transfer at the interface. Na+ transference numbers determined by the Bruce−Vincent method increased substantially with increasing NaFSI content, approaching >0.3 at the saturation concentration limit which may explain the improved performance. NMR spectroscopy, PFG diffusion measurements, and molecular dynamics simulations reveal a changeover to a facile structural diffusion mechanism for sodium ion transport at high concentrations in these electrolytes.

■

performance.6−11 For example, our previous research9 proposed an electrolyte composed of sodium bis(trifluoromethanesulfonyl)amide (NaNTf2) in C3mpyrFSI IL as a possible electrolyte for the sodium battery demonstrating stable cycling of Na, even at elevated temperatures. Ding et al. also reported sodium battery performance with a 0.8 mol kg−1 NaFSI in C3mpyrFSI electrolyte (approximately 20 mol % NaFSI).12 While they subsequently also demonstrated that a wider concentration range of NaFSI salt in C3mpyrFSI can support cycling in a sodium battery, they found that lower concentrations (20−25 mol %) were favored at ambient temperatures, which they ascribed to the lower ionic conductivity and higher viscosity for the higher NaFSI concentration electrolytes.12,13 However, more recent work by Hagiwara et al. showed that the Na+ transference number

INTRODUCTION Safe, inexpensive energy storage solutions are crucial for distributed or household based renewable energy sources such as solar and wind, or indeed for greater penetration of renewables into the electricity grid. Further advances in energy storage technologies are also required for transitioning to electric transportation technologies, including buses, trams, cars, and motorcycles. While lithium-ion batteries are currently the most advanced and implemented, the room temperature sodium battery has recently been investigated by many researchers seeking an alternative to lithium due to the possibility of substantial cost reductions and the greater abundance of sodium.1−4 Most room temperature sodium battery research has focused on the development of new sodium ion intercalating cathode and anode materials while, typically, propylene carbonate-fluoroethylene carbonate based electrolytes have been used to evaluate their electrochemistry.4,5 Ionic liquid (IL) based electrolytes have also been investigated for sodium batteries to improve safety and © 2016 American Chemical Society

Received: December 1, 2015 Revised: February 3, 2016 Published: February 3, 2016 4276

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C

nium bis(trifluoromethanesulfonyl)amide. A platinum pseudoreference was used for some experiments. This voltammetry was performed with a scan rate of 20 mV/s with potentiostatic control provided by Biologic SP-200TM controlled with ECLab (ver. 10.38) software as described previously.9 Symmetrical Cell Preparations. A sodium metal rod (Sigma-Aldrich) immersed in Paraffin oil was mechanically pressed to form a plate having approximately 100 μm thickness and cleaned with n-hexane. 1.2 cm diameter discs were then cut from the plate. These discs were attached to both cases of 2032 type coin cell compartments to be used for a sodium symmetrical cell. A 30 μm thick fiber-glass filter was used as a separator. Polarization was done with Neware battery tester BTS3000TH for 10 μA/cm for 20 cycles, 50 μA/cm for 20 cycles, 100 μA/cm for 20 cycles, 500 μA/cm for 20 cycles, 1000 μA/ cm for 20 cycles, and finally 10 μA/cm for 20 cycles. Transference Number Measurement. The transference number of each electrolyte was measured using a Na/Na electrochemical cell comprising two Na metal electrodes and according to the method described by Bruce and Vincent,18 and using the following equation introduced by these authors:

still increases at higher concentration despite the decrease in conductivity.14 Furthermore, our prior work with LiFSI in C3mpyrFSI systems15,16 showed that a lithium ion concentration around 50 mol % produced the best cell performance for lithium battery applications at ambient temperatures, providing the highest rate performance and stability compared to lower LiFSI concentrations. Yamada et al. also recently reported enhanced rate capability of graphite anodes in highly concentrated LiFSI ether based electrolyte systems, following a similar strategy of employing “superconcentrated” FSI based electrolytes.17 Thus, in this paper, we further investigate concentrated NaFSI−C3mpyrFSI IL electrolytes, studying the concentration dependence of the electrochemical properties, in the context of sodium electrochemistry and use NMR spectroscopy and MD simulations to understand their behavior. In particular, we isolate the performance of anode and cathode from one another in these electrolytes−previous publications having been primarily focused on different cathode materials.12,13 In this context we employ cyclic voltammetry, and Na| Na symmetrical cell cycling and record the EIS of the cells. We find that the highest concentration of NaFSI (50 mol %) produces the most stable, high-rate Na cycling. The Na+ transference number is also determined as a function of concentration, and NMR spectroscopy and PFG diffusion measurements are used to investigate transport in these electrolyte systems. Molecular dynamics (MD) simulations show that, at these high alkali salt concentrations, significant ion clustering occurs that leads to a facile, structural-rearrangementbased sodium ion diffusion mechanism, explaining the high transference number.

TNa + =

Here, ΔV is the applied voltage across the cell less (than 10 mV), I0 and Is are initial and steady-state current, and R0 and Rs are initial and steady state surface impedance of the measured cell. NMR Measurements. All the PFG-NMR and single pulse excitation (SPE) experiments were carried out on a Bruker Avance III 500 MHz wide bore spectrometer (with proton Larmor frequency of 500.07 MHz) equipped with a 5 mm diff50 pulse-field gradient probe. Each sample was packed to a height of 50 mm in a 5 mm Schott E NMR tube in an Ar filled glovebox and sealed with Teflon tape and a cap. The pulsedfield gradient stimulated echo (PFG-STE) pulse sequence was used to obtain the diffusion coefficients.19 The maximum strength of the gradient field is 29.454 T/m. The 1H, 19F NMR signals were used for the determination of the diffusion coefficients of the IL cation and anion species, respectively, and the diffusion coefficients were calculated using the following equation:19

■

EXPERIMENTAL SECTION Materials. C3mpyrFSI (99%) and NaFSI (99.9%) were obtained from Solvionics and used without further purification. Sample preparation and impurity characterization were performed as described in our previous publication.9 Eight different samples ranging from 0.17 to 3.24 mol kg−1 were prepared by dissolution of the salt in the IL at 50 °C. These concentrations are equivalent to 5 mol % to 50 mol % of NaFSI salt in the C3mpyrFSI IL as shown in Table 1. It should be Table 1. Electrolyte Compositions Employed in This Study mol % of NaFSI

mol % of C3mpyrFSI

molality (mol kg−1)

5 10 15 20 25 30 40 50

95 90 85 80 75 70 60 50

0.17 0.36 0.57 0.81 1.08 1.39 2.16 3.24

Is(ΔV − I0R 0) I0(ΔV − IsR s)

⎛I⎞ ⎛ δ⎞ ln⎜ ⎟ = − DNMR γ 2⎜Δ − ⎟δ 2g 2 ⎝ 3⎠ ⎝ I0 ⎠

Here I and I0 are the signals in the presence and absence of the gradient, respectively, γ is the gyromagnetic ratio of the nuclear studied, Δ is the interval between the gradient pulses, δ is the length of the gradient pulse, and g is the magnitude of the gradient pulse. In the present study, Δ was varied from 5 to 10 ms, δ was set between 1 and 4 ms, and g was optimized to a suitable strength range from 0.3 to 29.4 T/m according to the diffusion coefficients. Recycle delays were set to 5 s for all the diffusion experiments. 23Na chemical shifts were obtained from single pulse excitation (SPE) experiments with a recycle delay of 10 s. Each sample was measured from 0 to −70 °C. The sample temperatures for the variable temperature experiments were calibrated by using the relative chemical shift separation between the OH resonance and CH3 resonance in dry methanol.20

noted that the concentrations written in mol kg−1 indicates moles of NaFSI salt per kilogram of C3mpyrFSI (molality). The water contents of the samples after preparation were determined by Karl Fischer and shown to be 0.3) is consistent with the higher Na+ diffusion coefficient seen in Figure S1. Cyclic Voltammetry. Cyclic voltammograms acquired for the 5 mol % and 50 mol % electrolytes are shown in Figure 3, parts a and b, and other CVs over a range of concentrations and temperatures are presented in Supporting Information, Figures S2−S5. Sodium metal deposition is evident at about −3.5 V vs Ag|AgNTf2 and a corresponding stripping peak at about −3.3 V. The deposition and stripping processes exhibit several overlapping peaks, which implies that the deposit formed was not a single uniform layer. The deposition and stripping peak current intensity is low when the salt concentration is 0.17 mol kg−1 (5 mol %), less than 3 mA cm−2, and the cycling stability after the first cycle was also low (Figure 3a) for this NaFSI concentration for 25 °C and also at higher temperatures. (Supporting Information, Figures S2−S5). Furthermore, while the Coulombic efficiency in these cases starts at over 70% in the initial cycle, it reduces

electrolyte systems using the Walden plot and ionicity analysis.16 Therefore, we postulate that the effective fraction of active sodium ions at these higher concentrations, where ion clustering is likely, is greater than in dilute electrolytes. This could result from rapid exchange of the coordination environment of the sodium ion, effectively leading to a transport mechanism that arises from structural rearrangement (akin to the Grotthus mechanism in proton conductors25) of the system rather than vehicular diffusion of the sodium ion. Given that the preferred average coordination number of Na+ is between 4 and 5, at 50 mol % NaFSI, the only means by which this can be achieved is by a substantial fraction of anions being shared by more than one Na+. Diffusion of the sodium ions can then occur through their facile exchange along available anion sites. To test this hypothesis, molecular dynamics simulations were undertaken for both high (50 mol %) and low concentrations (10.2 mol % was used in simulations). Figure 2a presents a snapshot of the simulated partial ion environment for 50 mol % NaFSI in C3mpyrFSI. It is clear from this picture that, for highly concentrated NaFSI electrolytes, the FSI anion can either be coordinated to a single Na+ ion via one or two oxygens and it may be coordinated to multiple Na+ cations. At lower NaFSI concentrations there are fewer Na+ ions, but sufficient anions for full coordination with each Na+; this greatly reduces the chance of FSI being shared by multiple Na+ ions. This is supported by the RDFs of Na−NFSI arrangement in Figure 2b. The averaged coordination number of Na+ surrounding a FSI anion (CNNa) significantly increases at 50 mol %. The Na−oxygen environment is also presented in Figure 2c. We see that, independent of concentration, the oxygen nearest neighbors (or coordination number) is between 5 and 6. However, the way in which this is achieved on average for each Na+ cation is highly dependent on concentration, with the desired Na−O coordination being achieved with only 4.3 4279

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C

Figure 4. Cyclic voltammograms of 2nd cycles of various NaFSI concentrations in C3mpyrFSI IL: at 25 °C (left); at 75 °C (right). Conditions: Cu working electrode, Ag wire counter electrode, 10 mmol kg−1 AgNTf2 in C4mpyrNTf2 reference electrode, 20 mV s−1.

Figure 5. Na|Na symmetrical cell polarization profiles at 25 °C. (a) 5 mol % NaFSI in C3mpyrFSI, (b) 25 mol % NaFSI in C3mpyrFSI, (c) 50 mol % NaFSI in C3mpyrFSI, and (d) overpotential trend at various current densities.

The voltammograms also exhibit an additional broad deposition peak at about −2.5 V vs Ag|AgNTf2, the peak current of which increases with increasing temperature as shown in figures in the Supporting Information. This relatively small and broad peak, which was also observed in our previous TFSI/FSI mixed system,9 does not appear to change with salt concentration. It is also observed in Figure 3b in the 50 mol % electrolyte although it is not as clearly visible due to the high current of the main deposition peak. The CV cycling data for the entire composition range and as a function of temperature are given in the Supporting Information and show that for intermediate compositions, e.g., 25 mol %, the peak deposition and stripping currents are significantly increased to 10 mA cm−2, and the cycling stability is improved compared with the lowest concentration. The

rapidly to 0% after the fourth cycles (Figure 3c), suggesting that at this concentration of solute, the electrolyte, while having lower viscosity and higher conductivity, is unable to sustain stable cycling of sodium. At the highest concentration of NaFSI, i.e. 50 mol %, we see significantly improved cycling for the sodium metal, and in particular the Na dissolution peak remains stable with extended cycling (Figure 3, parts b and d). The relatively low efficiencies observed from the CV (30%) are commonly seen even in lithium systems and are due to the other possible reactions that occur, in particular in the cathodic scan where some reduction of the IL is possible. The symmetric cell testing described below is a better test of the cycling performance. Nevertheless, the CV data gives a strong indication of the likely reversibility and stability of the electrochemical reactions. 4280

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C

Figure 6. Na|Na symmetrical cell EIS at 25 °C with various salt concentrations upper left, the fitting model in upper right and the fitting results; R1 at lower left and R2 + R3 at lower right.

highest deposition peak current observed at 25 °C was when the salt concentration was 30 mol %. The somewhat lower peak currents with further increase in concentration may be due to the increasing viscosity (and lower ionic conductivity),12 which affects the slope of the deposition peak and possibly means that the peak deposition is not reached. We have previously observed this with highly concentrated LiFSI systems, despite the electrolyte exhibiting excellent device performance.15 The effect of temperature on cycling and Coulombic efficiency was also investigated, Figure 4, and this suggests that there is a strong interplay between concentration and temperature, with optimum performance being observed at higher concentrations for the higher temperatures. On the basis of this CV data, the optimum concentration of NaFSI in C3mpyrFSI for a sodium battery electrolyte at 25 °C (Figure 4a) would be located at between 25 mol % and 30 mol %, and over 40 mol % at 75 °C (Figure 4b). This data is somewhat in contrast with the full cell data presented by Dinget al.13 for Na/NaCrO2 cells which exhibited no obvious concentration dependence of capacity at elevated temperatures and a maximum in capacity around 20 mol % NaFSI at room temperature and below. This electrochemical behavior is a reflection of the nature of the speciation and a favoring of the structural rearrangement and/or hopping based diffusion mechanism with increased temperature and at the highest NaFSI concentrations, as discussed above. These mechanisms can be considered analogous to the rapid H+ transport in water and acidic

media.25 Early theories to account for the unusually fast H+ transport in water suggested that it arose from either (i) breaking and reformation of hydrogen bonds with rapid transport of the H+ at the periphery of a cluster (denoted structural diffusion) and the rate-determining step being the reformation of the hydrogen bond, or (ii) H+ ion hops from one water molecule to another (or to an OH− defect) where this hop is the rate-determining step. Both types of mechanism are possible, with respect to Na+ motion, in the NaFSI/ C3mpyrFSI electrolytes discussed here, in particular at the high concentrations where the MD simulations indicate significant clustering, with the tendency for one FSI− ion to be coordinated by two Na+ ions within a cluster, and hence a structural diffusion of Na+ ions through such a cluster can be envisaged. This site exchange may also be observed in the NMR data described below when we consider the temperature dependence of the line width and chemical shift data. Na|Na Symmetrical Cell Polarization. Cycling of symmetric Na|Na cells is a useful method to determine the stability of the negative electrode in a sodium based device in a given electrolyte and allows us to isolate the performance of the different electrolyte compositions at the Na electrode, independent of any cathode. Previous work by Ding et al.12,13 focused on the full battery performance. Symmetric cell cycling was performed with the following currents densities from 10, 50, 100, 500, and 1 mA cm−2 (each for 8 min in one direction and 2 min resting) followed by a return to 10 μA cm−2, for 20 cycles at each value. The polarization profiles for each 4281

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C

decreases even though the electrolyte impedance contribution is increasing with concentration. This is an important finding for the development of stable cycling of sodium metal based devices and indicates that a high ionic conductivity of the electrolyte alone, is not necessarily the most important parameter in their performance. We can only speculate at this time that the species present at the electrode interface with increasing concentration favors the formation of a more homogeneous and compact SEI layer thus leading to improved electrochemical characteristics. NMR Characterization. NMR spectroscopy and pulsed field gradient diffusion measurements have been used to characterize the mobility of the various ionic species in the NaFSI/C3mpyrFSI electrolytes as a function of composition and temperature. 23Na diffusion could not be measured in these samples due to the very short transverse relaxation time (T2), however, 1H and 19F PFG NMR were used to measure diffusion of the IL cation and anion respectively (Figures 7 and 8).

concentration are given in Figure S6 in Supporting Information. Figure 5a presents symmetric cell performance for the 5 mol % electrolyte and shows the first 20 cycles at 10 μA cm−2, the cycling requiring approximately 0.3 V polarization potential, which progressively increases with increased current density. Failure occurs, presumably due to dendrite formation, when the currents approaches 500 μA cm−2 as indicated by the unstable cycling profile and final short circuit at 500 μA cm−2. Interestingly, for the 25 and 50 mol % electrolyte (Figure 5, parts b and c), the polarization potential at 10 μA cm−2 is significantly reduced compared to the lower concentrations of NaFSI, despite the increased electrolyte viscosity and decreased bulk conductivity.12,14 This correlates with the CV data discussed above and also with the reduced surface impedance shown below. With increasing current density, the polarization potential increases, as for the lower NaFSI concentration, although still remaining below the values for the equivalent current densities in the 5 mol % sample. Figure 5d summarizes the data presented in Figure S6 for each of the cells with different NaFSI concentrations, showing clearly that the cell polarization potential is decreasing at the higher salt concentrations, enabling higher polarization currents; this will enable high C rate charging and discharging in a full battery cell. Figure S6 more clearly demonstrates the affect of NaFSI concentration on cell stability at higher current densities with indicators of short circuits at all concentrations other than 50 mol %. In summary, these results clearly show that, at 25 °C, the highest concentration electrolytes allow stable sodium metal cycling at higher current densities despite their higher viscosity. A comparison of various compositions and temperatures can be found in Figure S7 and S8 in Supporting Information. Significantly, at moderate temperatures (50−75 °C) the 50 mol % electrolyte provides enhanced rate capability with lower polarization potentials with respect to current density. In this instance the 30 and 40 mol % NaFSI electrolytes still supported stable cycling at 1 mA/cm2 whereas the lower concentrations (≤25 mol %) short circuit at lower current densities (100−500 μA/cm2). This further highlights the importance of NaFSI salt concentration in dictating cell performance. Electrochemical Impedance Spectroscopy of Na|Na symmetrical cells. To better understand the cause of this reduced polarization potential, EIS measurements for each cell were undertaken and the data fitted to calculate the surface impedance. This was carried out using an SP-200 with EC lab version 10.38 with the frequency range of 100 kHz to 0.1 mHz, with 10 mV amplitude (PEIS). The 10 mHz points are shown on each curve. The points in Figure 6a are measured EIS data; these have been fitted with the equivalent circuit shown in the figure, where R1 corresponds to the electrolyte resistance and R2 and R3 correspond to each of the two semicircles, presumably surface related. The fitted data show clearly that, with increasing NaFSI content, the electrolyte resistivity (R1) (shown in Figure 6b) increases, as expected. In contrast, (R2 + R3) (Figure 6c) corresponding to the interfacial impedance, clearly decreases by over an order of magnitude with increasing concentration. A clear relationship with this polarization potential trend and the surface impedance can thus be seen by comparison with the data in Figure 5. This can be interpreted as the surface impedance being a more dominant contribution to cell performance than the bulk electrolyte impedance (or ionic conductivity) for the high NaFSI containing electrolyte, hence the overall cell impedance

Figure 7. Cation (1H) and anion (19F) diffusion coefficients as a function of NaFSI concentration at 25 °C.

Figure 7 presents the diffusion coefficients as a function of NaFSI salt concentrations at 25 °C, showing a continuous decrease in both cation and anion diffusion coefficient with increased NaFSI concentration, which correspond to the increasing viscosity as previously reported.12 The anion diffusion coefficient is slightly higher than the cation which is likely due to its smaller size. Parts a and b of Figure 8 shows the temperature dependence of the diffusivity of the neat C3mpyr FSI and the NaFSI containing electrolytes. The 5 mol % concentrated electrolyte shows slightly lower cation and anion diffusivity than the neat C3mpyr FSI at lower temperatures, but slightly higher diffusivity at higher temperatures. In the highest concentration NaFSI electrolyte (50 mol %) both cation and anion diffusivity are over a magnitude lower than in the neat IL and 5 mol % electrolyte (Figure 8). Interestingly, the diffusivity of both the NaFSI containing electrolytes show an apparently stronger temperature dependence than the neat sample, which would correspond to a larger activation energy. This will be discussed in the context of the 23Na chemical shift data below. The full width at half-maximum (fwhm) of a solution-state NMR spectrum is generally inversely related to the T2 relaxation time, which is mainly influenced by the mobility of a given species although it can also affected by the exchange rate between different physical/chemical sites. The fwhm of the 4282

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C

Figure 8. (a) Cation and (b) anion diffusion coefficients, determined from 1H and 19F PFG NMR measurements, as a function of temperature.

Figure 10. 23Na peak width (fwhm) for all NaFSI concentrations as a function of temperature.

often being a more localized motion, except in the extreme line narrowing regime.26 The fwhm data can be fitted using the BPP theory to extract a more quantitative mobility parameter, a correlation time τ, which relates to the local motion for the Na+ ion. (Δν)2 =

⎛2⎞ 2 ⎜ ⎟δω tan−1(τ Δν) ⎝π ⎠ 0

where Δv is the line width, δω0 is the line width in the rigid limit and τ is the correlation time.27 As we do not have the complete sigmoidal curve to determine the true rigid lattice fwhm, the data need to be considered at best qualitative. Nevertheless, the correlation time versus temperature shown in Figure 11 suggests that the 50 mol % electrolyte has a stronger temperature dependence than that of the lower concentration samples. The correlation time is inversely related to the mobility of the ion and thus at higher temperatures and higher NaFSI content, the sodium ion mobility could significantly increase. This may be another reason for the improved cycling performance observed at elevated temperatures for the 50 mol % electrolyte compared with the lower concentrations. Fits were performed for all concentrations and the activation energies are shown in Table 2 together with the correlation time at 298 K as obtained from the fits. Interestingly the activation energy increases systematically with increasing concentration. The chemical shift of the sodium ion is indicative of its chemical environment. Our previous work with a mixed anion NaTFSI in C3mpyrFSI ionic liquid showed that, with increasing

Figure 9. 23Na spectra for (a) 5 mol % and (b) 50 mol % as a function of temperature and (c) concentration dependence at 298 K.

23

Na resonance is shown in Figure 9, parts a and b, for the lowest and highest NaFSI concentration as a function of temperature, as well as Figure 9c for the concentration dependence at room temperature. In general, the fwhm decreases with increasing temperature and decreasing salt concentration (Figure 10), which suggests that the sodium ions are more mobile at higher temperature and lower salt concentration, consistent with the trends of C3mpyr cation and FSI anion diffusivity. However, it is important to stress that diffusivity, as measured by PFG and fwhm (or T2) measurements, may characterize different motional modes, with T2 4283

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C

represent the Na+ ions, the stick molecules indicate FSI and the lighter blue contours represent the molecular surface of the pyrrolidinium cations. Clustering and increased density of Na+ and FSI− ions is clearly shown in the high concentration species, which would account for the overall decrease in the diffusivity of the ionic liquid ions, and the multiple environments for the Na+ ion are also seen. One can envisage a percolating pathway along which the Na+ ions can exchange which would account of the higher D(Na+) calculated from the MSD data (Figure S1) as well as the fwhm and chemical shift data discussed here. More extensive simulation results and further discussion of the effect of alkali ion concentration on structure, ion dynamics and transference numbers will be presented in future paper.

Figure 11. 23Na correlation time determined from fitting of the fwhm data using the BPP equation.

■

CONCLUSIONS

We have shown that a Na/C3mpyrFSI electrolyte can exhibit a transference number TNa+ of greater than 0.3, which is comparable to commercial lithium ion electrolyte performance and is consistent with recent reports from the Hagiwara group.14 The EIS results show that higher sodium ion concentration decreases the electrode surface impedance dramatically, accounting for the lower polarization potential electrochemical cycling of Na metal, particularly at 50−75 °C, despite a lower overall ionic conductivity. This cycling behavior appears similar to a lithium FSI system previously reported which suggests a similar structural diffusion mechanism also exists in those mixed ion electrolytes.15 The data also confirms that higher salt concentrations result in a lower surface impedance leading to a lower polarization potential and enabling a relatively high current charging of 1 mA cm−2 compared to previous reports which favored lower concentrations of NaFSI. It appears that only the highest salt concentrations leads to stable symmetric cell cycling at higher current densities and strongly indicates that cyclic voltammetry alone cannot predict the performance of a given electrolyte in a device configuration and that the interfacial layer that forms in the highest concentration electrolytes is more important than the ionic conductivity of the electrolyte itself, significantly improving the sodium cycling behavior. MD simulations indicate that for a highly concentrated sample (50 mol %) clustering of Na+ and FSI− ions occur with one FSI anion able to coordinate with two Na+ ions. This then enables a site exchange and/or structural diffusion mechanism for the Na+ ion that accounts for its higher transference number. The NMR data also support rapid exchange for the Na + ion between different coordination environments, particularly at higher concentrations and higher temperatures. A more detailed MD simulation investigation on the concentration dependence of structure and mobility of highly concentrated alkali metal salt mixtures with ionic liquids will be presented in a future manuscript. This work indicates a way forward to increase Na + transference numbers, an important criterion for high performance battery devices. The results also suggest that highly concentrated or “mixed electrolytes” improve interfacial properties of Na metal anodes thereby leading to better

23

salt concentration, the Na resonance occurred at a more negative chemical shift value. This is also observed for the single FSI anion system in Figure 12a. For an equivalent ion concentration (around 15 mol %) sodium salt at 20 °C, the TFSI anion system (Figure 12b) is significantly more negative (−12.2 ppm) compared with the FSI system (−11 ppm). We have previously suggested that the coordination of the Na+ cation with the TFSI anion leads to stronger shielding possibly due to stronger interactions. Note that the maximum solubility of the NaTFSI in the same IL was only 1.35 mol/kg compared with 3.24 mol/kg in the FSI system studied here. Interestingly, at these higher NaFSI concentrations the 23Na resonance occurs at an even more negative chemical shift suggesting strong interaction between Na+ and the FSI anions (or higher density of electronic charge and hence stronger shielding of the Na+ cation). This is consistent with the ionic clustering and coordination seen from the MD simulations described above. With increasing temperature at these higher NaFSI concentrations, we observe a rapid increase in the chemical shift (as observed in all samples for NaTFSI) until a maximum temperature beyond which the chemical shifts for all concentrations is practically identical. In the case of the mixed anion systems, we previously showed that the maximum correlated with the increasing Tg with increasing Na salt content and reflected changes in ionic environment and exchange between different sites; this is also likely to be valid for the NaFSI system here and at higher temperatures the exchange may be rapid enough that the environments appear identical, independent of concentration. In the FSI only situation, the exchange may be occurring between different species, e.g., Na(FSI)32−, Na(FSI)2−, Nax(FSI)yz‑, etc..., where the FSI anion may be either coordinating through two oxygens or only one, and may be coordinated to a single Na+ or two Na+ ions (as shown by the MD results above). These different states for the Na+ ion are likely to be both concentration and temperature dependent. To highlight this clustering phenomenon and visualize how ion dynamics may be affected by alkali salt concentration, a snapshot of the 10 and 50 mol % NaFSI systems from the MD simulations is shown in Figure 13, to more clearly distinguish Na-FSI coordination environments. The dark blue spheres

Table 2. Activation Energy for 23Na Correlation Times Obtained from Fitting Line Widths to the BPP Equation NaFSI concentration (mol/kg) activation energy (kJ/mol)

0.17 21.2

0.36 22.0

0.57 22.8 4284

0.81 23.3

1.08 24.6

1.39 24.7

2.16 28.8

3.24 30.9

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C

Figure 12. 23Na chemical shift as a function of concentration and temperature in NaFSI in C3mpyrFSI (a) in comparison with NaTFSI in C3mpyrFSI (b).

Figure 13. MD simulation images for 10 mol % (left) and 50 mol % NaFSI (right) in C3mpyrFSI showing explicitly the Na+ (dark blue ball) and FSI− (stick) ions and the molecular surface of the C3mpyr cations so as to indicate the difference in the typical Na coordination environment between the low and high concentration systems.

■

performance and stability of ionic liquid electrolytes for next generation sodium electrochemical devices.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.5b11746. Details of the MD simulation methodology, with Figure S1 showing the mean squared displacement data from one simulation run and, in addition, a complete set of cyclic voltammetry and symmetric cell cycling data, Figures S2−S8 (PDF)

■

REFERENCES

(1) Komaba, S.; Murata, W.; Ishikawa, T.; Yabuuchi, N.; Ozeki, T.; Nakayama, T.; Ogata, A.; Gotoh, K.; Fujiwara, K. Electrochemical Na Insertion and Solid Electrolyte Interphase for Hard-Carbon Electrodes and Application to Na-Ion Batteries. Adv. Funct. Mater. 2011, 21, 3859−3867. (2) Goodenough, J. B. Evolution of Strategies for Modern Rechargeable Batteries. Acc. Chem. Res. 2013, 46, 1053−1061. (3) Ong, S. P.; Chevrier, V. L.; Hautier, G.; Jain, A.; Moore, C.; Kim, S.; Ma, X.; Ceder, G. Voltage, Stability and Diffusion Barrier Differences between Sodium-Ion and Lithium-Ion Intercalation Materials. Energy Environ. Sci. 2011, 4, 3680−3688. (4) Oh, S.-M.; Myung, S.-T.; Hassoun, J.; Scrosati, B.; Sun, Y.-K. Reversible Nafepo4 Electrode for Sodium Secondary Batteries. Electrochem. Commun. 2012, 22, 149−152. (5) Slater, M. D.; Kim, D.; Lee, E.; Johnson, C. S. Sodium-Ion Batteries. Adv. Funct. Mater. 2013, 23, 947−958. (6) Fukunaga, A.; Nohira, T.; Kozawa, Y.; Hagiwara, R.; Sakai, S.; Nitta, K.; Inazawa, S. Intermediate-Temperature Ionic Liquid NafsaKfsa and Its Application to Sodium Secondary Batteries. J. Power Sources 2012, 209, 52−56. (7) Chagas, L. G.; Buchholz, D.; Wu, L.; Vortmann, B.; Passerini, S. Unexpected Performance of Layered Sodium-Ion Cathode Material In ionic Liquid-Based Electrolyte. J. Power Sources 2014, 247, 377−383. (8) Monti, D.; Jónsson, E.; Palacín, M. R.; Johansson, P. Ionic Liquid Based Electrolytes for Sodium-Ion Batteries: Na+ Solvation and Ionic Conductivity. J. Power Sources 2014, 245, 630−636. (9) Yoon, H.; Zhu, H.; Hervault, A.; Armand, M.; MacFarlane, D. R.; Forsyth, M. Physicochemical Properties of N-Propyl-N-Methylpyrrolidinium Bis(Fluorosulfonyl)Imide for Sodium Metal Battery Applications. Phys. Chem. Chem. Phys. 2014, 16, 12350−12355.

AUTHOR INFORMATION

Corresponding Author

*(M.F.) E-mail: [email protected]. Telephone: +61 3 92446821. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank the ARC (Australian Research Council) for funding through DP130101652. M.F. and D.R.M. also acknowledge the ARC for support via their ARC Laureate fellowships. M.A. gratefully acknowledges Deakin University for travel funding through the “Thinker in Residence” program. 4285

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286

Article

The Journal of Physical Chemistry C (10) Wang, C.-H.; Yeh, Y.-W.; Wongittharom, N.; Wang, Y.-C.; Tseng, C.-J.; Lee, S.-W.; Chang, W.-S.; Chang, J.-K. Rechargeable Na/ Na0.44mno2 Cells with Ionic Liquid Electrolytes Containing Various Sodium Solutes. J. Power Sources 2015, 274, 1016−1023. (11) Wongittharom, N.; Wang, C.-H.; Wang, Y.-C.; Yang, C.-H.; Chang, J.-K. Ionic Liquid Electrolytes with Various Sodium Solutes for Rechargeable Na/Nafepo4 Batteries Operated at Elevated Temperatures. ACS Appl. Mater. Interfaces 2014, 6, 17564−17570. (12) Ding, C.; Nohira, T.; Kuroda, K.; Hagiwara, R.; Fukunaga, A.; Sakai, S.; Nitta, K.; Inazawa, S. NaFSA-C1C3pyrFSA Ionic Liquids for Sodium Secondary Battery Operating over a Wide Temperature Range. J. Power Sources 2013, 238, 296−300. (13) Ding, C.; Nohira, T.; Hagiwara, R.; Matsumoto, K.; Okamoto, Y.; Fukunaga, A.; Sakai, S.; Nitta, K.; Inazawa, S. Na[FSA][C3C1pyr][FSA] Ionic Liquids as Electrolytes for Sodium Secondary Batteries: Effects of Na Ion Concentration and Operation Temperature. J. Power Sources 2014, 269, 124−128. (14) Matsumoto, K.; Okamoto, Y.; Nohira, Y.; Hagiwara, R. Thermal and Transport Properties of Na[N(SO2F)2]− [N-Methyl-Npropylpyrrolidinium][N(SO2F)2] Ionic Liquids for Na Secondary Batteries. J. Phys. Chem. C 2015, 119, 7648−7655. (15) Yoon, H.; Howlett, P. C.; Best, A. S.; Forsyth, M.; MacFarlane, D. R. Fast Charge/Discharge of Li Metal Batteries Using an Ionic Liquid Electrolyte. J. Electrochem. Soc. 2013, 160, A1629−A1637. (16) Yoon, H.; Best, A. S.; Forsyth, M.; MacFarlane, D. R.; Howlett, P. C. Physical Properties of High Li-Ion Content N-Propyl-NMethylpyrrolidinium Bis(Fluorosulfonyl)Imide Based Ionic Liquid Electrolytes. Phys. Chem. Chem. Phys. 2015, 17, 4656−4663. (17) Yamada, Y.; Yaegashi, M.; Abe, T.; Yamada, A. A Superconcentrated Ether Electrolyte for Fast-Charging Li-Ion Batteries. Chem. Commun. 2013, 49, 11194−11196. (18) Evans, J.; Vincent, C. A.; Bruce, P. G. Electrochemical Measurement of Transference Numbers in Polymer Electrolytes. Polymer 1987, 28, 2324−2328. (19) Tanner, J. E.; Stejskal, E. O. Restricted Self-Diffusion of Protons in Colloidal Systems by Pulsed-Gradient Spin-Echo Method. J. Chem. Phys. 1968, 49, 1768. (20) Van Geet, A. L. Calibration of Methanol Nuclear Magnetic Resonance Thermometer at Low Temperature. Anal. Chem. 1970, 42, 679−680. (21) Smith, W.; Forester, T. R. Dl_Poly_2.0: A General-Purpose Parallel Molecular Dynamics Simulation Package. J. Mol. Graphics 1996, 14, 136−141. (22) Canongia Lopes, J.; Pádua, A. H. Cl&P: A Generic and Systematic Force Field for Ionic Liquids Modeling. Theor. Chem. Acc. 2012, 131, 1−11. (23) Chen, F.; Pringle, J. M.; Forsyth, M. Insights into the Transport of Alkali Metal Ions Doped into a Plastic Crystal Electrolyte. Chem. Mater. 2015, 27, 2666−2672. (24) Fujii, K.; Seki, S.; Fukuda, S.; Takamuku, T.; Kohara, S.; Kameda, Y.; Umebayashi, Y.; Ishiguro, S. Liquid Structure and Conformation of a Low-Viscosity Ionic Liquid, N-Methyl-N-PropylPyrrolidinium Bis(Fluorosulfonyl) Imide Studied by High-Energy XRay Scattering. J. Mol. Liq. 2008, 143, 64−69. (25) Eigen, M.; de Maeyer, L. Proc. R. Soc. London, Ser. A 1958, 247, 505−533. (26) Yao, S.; Babon, J. J.; Norton, R. S. Protein Effective Rotational Correlation Times from Translational Self-Diffusion Coefficients Measured by Pfg-Nmr. Biophys. Chem. 2008, 136, 145−151. (27) Bloembergen, N.; Purcell, E. M.; Pound, R. V. Relaxation Effects in Nuclear Magnetic Resonance Absorption. Phys. Rev. 1948, 73, 679− 712.

4286

DOI: 10.1021/acs.jpcc.5b11746 J. Phys. Chem. C 2016, 120, 4276−4286