7750
J. Phys. Chem. C 2009, 113, 7750–7754
Dissolved Argon Changes the Rate of Diffusion in Room Temperature Ionic Liquids: Effect of the Presence and Absence of Argon and Nitrogen on the Voltammetry of Ferrocene Laura E. Barrosse-Antle,† Leigh Aldous,‡ Christopher Hardacre,‡ Alan M. Bond,§ and Richard G. Compton*,† Department of Chemistry, Physical and Theoretical Chemistry Laboratory, Oxford UniVersity, South Parks Road, Oxford OX1 3QZ, United Kingdom, School of Chemistry and Chemical Engineering/QUILL, Queen’s UniVersity Belfast, Belfast, Northern Ireland BT9 5AG, United Kingdom, and School of Chemistry, Monash UniVersity, P.O. Box 23, Victoria 3800, Australia ReceiVed: February 20, 2009; ReVised Manuscript ReceiVed: March 16, 2009
This work explores the effects of argon and nitrogen, two electrochemically and chemically inert gases frequently used in sample preparation of room temperature ionic liquid (RTIL) solutions, on the electrochemical characterization of ferrocene (Fc) dissolved in the RTIL 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide ([C2mim][NTf2]). Remarkably, chronoamperometrically determined diffusion coefficients of Fc in [C2mim][NTf2] are found to increase from 4.8 ((0.2) × 10-11 m2 s-1 under vacuum conditions to 6.6 ((0.5) × 10-11 m2 s-1 in an atmosphere of 1 atm Ar. In contrast, exposing a vacuum-purified sample to an atmosphere of 1 atm N2 resulted in no significant change in the measured diffusion coefficient of Fc. The effect of dissolved argon on diffusion transport is unexpected and has implications in electrochemistry and elsewhere. Fc was found to volatilize under vacuum conditions. We propose, however, that evacuation of the cell by vacuum prior to electrochemical measurements being carried out is the only way to ensure that no contamination of the sample occurs, and use of an in situ method of determining the diffusion coefficient and concentration of Fc dispells any ambiguity associated with Fc depletion by vacuum. 1. Introduction Since they began to attract interest in the 1970s, room temperature ionic liquids (RTILs), salts that are liquid at 25 °C, have been explored as solvents and electrolytes over a range of fields and applications. The ability of RTILs to be tailored to specific purposes by intelligent choice of the component anions and cations has led to the creation of a diverse spectrum of these compounds.1 In addition to the increasing number of these liquids overall, the ever-increasing intensity with which the physical properties of these liquids are now being studied has led to the realization that it is difficult to make accurate generalizations about the class as a whole.2,3 Often, however, RTILs share properties of negligible vapor pressure, high thermal stability, high polarity, large electrochemical windows, intrinsic conductivity, and high viscosity.3-5 In the last year, RTILs have been studied for use as separation media for gas mixtures,6 as electrode modifications for biosensors,7 and as electrolytes for batteries,8 to name only a few areas of exploration. In recent works, we reported changes in voltammetry with the individual addition of sulfur dioxide and carbon dioxide to the RTIL 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide ([C2mim][NTf2]; Figure 1).9,10 These two gases, SO2 in particular, interact strongly with RTILs,11 so it was not surprising that, in both cases, saturation of the RTIL with gas increased the limiting currents and diffusion coefficients. These observations led us to question the assumed benign character of other gasessAr and N2swith respect to RTILs. This is * Corresponding author. E-mail:
[email protected]. Telephone: +44(0) 1865 275, 413. Fax: +44(0) 1865 275 410. † Oxford University. ‡ Queen’s University Belfast. § Monash University.
Figure 1. Structures and naming conventions of the cation, anion, and electroactive species used in this study.
addressed in the present paper with the remarkable conclusion that dissolved Ar significantly influences the voltammetry of ferrocene in [C2mim][NTf2], indeed to a greater extent than dissolved CO2. It is important to understand the interaction of dissolved gases with RTILs in order to accurately characterize the physical properties of RTILs themselves as well as systems that employ RTILs as solvents. There currently exists in the literature a large number of conflicting reports of diffusion coefficients for Fc in various RTILs. For example, the diffusion coefficient of small concentrations of Fc (5-10 mM) in the RTIL [C2mim][NTf2] ranges in its reported values (over the literature surveyed) from 6.3 × 10-12 m2 s-1 12 to 4.60 × 10-11 m2 s-1 at room temperature.13 Most importantly, the dependence of DFc on Fc concentration is also disputed, with several groups observing a dependence of D on c at high concentrations,14-16 while others submit evidence of no dependence of D on c at any Fc concentration.17,18 These disagreements are not likely to be resolved in the absence of information on the physical effects of various solutes on the samples in question. The susceptibility of RTILs to impurities is well documented and one of the major challenges of using RTILs as solvents in electrochemical systems. The presence of water, for example, changes the viscosity, potential window size, and conductivity
10.1021/jp9015849 CCC: $40.75 2009 American Chemical Society Published on Web 04/02/2009
Effect of Argon on Ionic Liquids of many RTILs.4,5,19-22 Even hydrophobic ionic liquids absorb water from the atmosphere and so must be stored and used in a manner that minimizes that solute’s effect on the physical properties of the solvent.20,21 Different groups employ different methods for reducing the amount of water in RTILs under study. Frequently, dried nitrogen gas is either used to blanket the setup (i.e., a glovebox) or bubbled through the RTIL immediately prior to measurements being taken.14,16,23,24 Other methods include using an argon-filled glovebox,25,26 using helium or a nitrogen/oxygen combination as an inert blanketing atmosphere,27,28 and employing an airtight electrochemical cell that can be evacuated by vacuum.22,29-33 In some cases, while the storage conditions or the sample preparation method of the RTIL solutions are explicitly reported, the measurement conditions are not, and no indication of the time scale between storage/preparation and measurement is available.12,15,34 Though the methods involving inert atmospheres are designed to limit the amount of water present in the samples, they do not address the possibility of the blanketing atmosphere itself affecting the observed electrochemical characteristics of the system. Argon and nitrogen are generally assumed to be benign gases, inert with respect to electrochemical systems, or else they would not be used as blanketing atmospheres in the cases reported above. Remarkably, however, the absorption of Ar by [C2mim][NTf2] is shown in this work to increase the diffusion coefficient of Fc to a greater degree than the absorption of CO2. In contrast, N2-saturation of this RTIL appears to have no signigicant effect on the physical properties of the [C2mim][NTf2]/ Fc system. 2. Experimental Section 2.1. Reagents and Instrumentation. 1-Ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide ([C2mim][NTf2]) was prepared using a standard literature procedure.35 Ferrocene (Aldrich, 98%), tetra-n-butylammonium perchlorate (TBAP, Fluka, Puriss electrochemical grade, > 99%), and acetonitrile (Fischer Scientific, dried and distilled, >99.99%) were used as received without further purification. Cyclic voltammetry (CV) was performed using a type II µAutolab (Eco Chemie, Utrecht, Netherlands), which was interfaced with a PC using GPES (version 4.9) software for Windows. Measurements were performed using a two-electrode cell consisting of a platinum working ultramicroelectrode (10 µm diameter) and a silver wire quasi-reference electrode (0.5 mm diameter). The electrodes were housed in a glass “T-cell” specially designed to control the environment of the RTIL.22,29 The microelectrode was modified using a portion of a disposable plastic micropipet tip to create a reservoir in which 20 µL of ionic liquid was placed. Unless otherwise noted, all electrochemical measurements were taken at 25 °C. Nitrogen (BOC Gases, >99.99%) and argon (BOC Gases, >99.99%) were dried prior to introduction to the T-cell. Gas was bubbled through concentrated sulfuric acid (AnalaR) and then directed through a silica gel trap before flowing into the system via one arm of the T-cell and being removed through the other. The gas was then directed into the fume cupboard. The system took up to 100 min after first introduction of the gas to give a consistent voltammetric response. 2.2. Electrode Preparation. Before use, the microelectrode was polished on soft lapping pads (Kemet Ltd., U.K.) using 1.0 and 0.3 µm aqueous alumina slurries (Buehler, Illinois). The radius of the microdisk electrode was electrochemically calibrated by analyzing the steady state voltammetry of a 2 mM
J. Phys. Chem. C, Vol. 113, No. 18, 2009 7751 ferrocene solution in acetonitrile, which contained 0.1 M TBAP as a supporting electrolyte. The diffusion coefficient value used was 2.3 × 10-9 m2 s-1 at 25 °C.36 2.3. Chronoamperometric Experiments. Chronoamperometric transients were achieved using a sample time of 0.01 s. The pretreatment step consisted of holding the potential at 0 V for 20 s, followed by a 2 s equilibration period. The potential was stepped to the required value, and the current was measured for 10 s. The nonlinear curve fitting function in Origin 7.0 (MicroCal Software Inc.) following the Shoup and Szabo37 approximation as employed by Evans et al.38 was used to fit the experimental data. The equations used in this approximation describe the current response within an accuracy of 0.6% and are given below:
f(τ) ) 0.7854 + 0.8863τ-1/2 + 0.2146 exp(-0.7823τ-1/2) I ) -4nFDcrdf(τ) τ)
4Dt rd2
(1) (2) (3)
where n is the number of electrons transferred, F is the Faraday constant, D is the diffusion coefficient, c is the bulk concentration of parent species, rd is the radius of the microdisk electrode, and t is the time. The value of the electrode radius was fixed, having been previously calibrated. The software performed up to one hundred iterations on the data, stopping when the experimental data had been optimized. A value for the diffusion coefficient, D, and the product of the number of electrons and the concentration of the parent species, nc, was thus obtained. During the fitting routine for the transients obtained from the Ar-saturated solution, cFc was held constant at the value ascertained for that solution prior to the addition of gas. 3. Results and Discussion 3.1. Depletion of Ferrocene in RTIL Solution due to Vacuum Conditions. The following study was carried out using Fc solutions made by adding solid Fc directly to [C2mim][NTf2] and stirring for no fewer than 60 min.39 An advantage of this method of sample preparation is that baseline measurements of the 15 mM Fc solution in [C2mim][NTf2] under atmospheric conditions could be obtained. When the chronoamperometrically determined concentration (c) and diffusion coefficient (D) values of the undried solution (0 min under vacuum conditions) were compared to their counterparts after 90 min spent under vacuum conditions, a drastic decrease in the Fc concentration was noted. Cyclic voltammetric and chronoamperometric measurements were obtained at intervals while the T-cell was under vacuum to ascertain the time dependence of the decrease in Fc concentration. As shown in Figure 2a, a logarithmic decrease in limiting current was observed over a little more than 100 min. Diffusion coefficients over a period of vacuuming were found to drop initially and then hold steady (Figure 2b). The initial drop in D was expected, as the effect of water and other solutes contained in the atmosphere is to decrease the viscosity of the ionic liquid. When vacuum conditions were applied, those solutes were removed and the viscosity of the solution increased, resulting in a slower D. Approximately 20 min of vacuum were sufficient for D to stabilize. The concentration value of Fc decreases with time spent under vacuum, driving the observed decrease in limiting current. Over the course of the 90 min corresponding to Figure 2b, c drops from 14.9 ((0.1) mM to
7752
J. Phys. Chem. C, Vol. 113, No. 18, 2009
Barrosse-Antle et al.
Figure 4. Variation of the log of the limiting current of ferrocene oxidation over the time scale of an experiment in which the solution was alternately exposed to vacuum and saturated with argon. Open circles represent data measured when the solution was under vacuum, and filled squares represent data measured when the solution was exposed to 1 atm Ar. The inset shows a comparison of the cyclic voltammetry of ferrocene in vacuumed and Ar-saturated samples (10 mV s-1 scan rate). The data points corresponding to the inset voltammetry are circled.
Figure 2. Concentration of ferrocene in [C2mim][NTf2], which is shown to decrease with the amount of time that the solution spends in vacuum conditions, as supported by (a) the logarithmic decrease in the oxidative limiting current of ferrocene and (b) the lack of change in the diffusion coefficient of ferrocene over time.
Figure 3. Experimental (O) and fitted theoretical (s) chronoamperometric transients for Fc oxidation in [C2mim][NTf2] prior to the sample being held under vacuum. The potential was stepped from 0 to 0.5 V and held for 10 s.
4.2 ((0.1) mM. A chronoamperometric transient showing the oxidation of Fc in a sample prior to vacuum, fit using the Shoup Szabo approximation, can be seen in Figure 3. Perusal of the literature revealed a recent paper that observed similar volatilization of Fc under vacuum conditions.18 In addressing the purported concentration dependence of D for Fc in RTILs,14-16 Vorotyntsev et al. report a decrease in Fc concentration when a stirred solution of Fc in [C4mim][NTf2] is placed under vacuum. This supports their hypothesis, suggested previously by Rogers et al.,17 that the literature reports of dependence of D on c for Fc in RTILs could be at least partially due to a disperity in the assumed and actual concentra-
tions of Fc in solution. Vorotyntsev et al. use UV-vis spectral data to determine Fc concentration independently of electrochemical signal to mitigate this problem. Fitting chronoamperometric transients with the Shoup Szabo approximation is an equivalent method, as it simultaneously determines the D and c values of the solution in situ. Though both methods serve the same purpose, to provide an accurate measure of Fc concentration in the solution at the time of the measurement, the advantage of the electrochemical method should be clearsall measurements can be performed in the same cell without additional preparation or equipment. 3.2. Effect of Argon and Nitrogen Saturation on the Fc Signal in [C2mim][NTf2]. The oxidation of Fc in [C2mim][NTf2] was investigated using cyclic voltammtery and chronoamperometry at a 10 µm diameter platinum electrode. So that appropriate background scans were obtained, chronoamperometric transients and limiting currents for the solution (initially 15 mM Fc in [C2mim][NTf2]) were measured at approximately 15-min intervals for the duration of the time spent under vacuum. Immediately prior to gas entering the system, D was determined to be 4.8 ((0.2) × 10-11 m2 s-1. The concentration of Fc at that point varied between repititions, as the time spent under vacuum was not always exactly the same. However, as discussed in the previous section, D remained constant once water and other atmospheric solutes were extracted from the ionic liquid, leading to a reportable value. After 45-50 min of exposure to vacuum, the cell was closed, the vacuum tube removed, and gas inlet and outlet lines attached to the arms of the T-cell. Argon gas was allowed to flow through the cell immediately afterward, creating a 1 atm atmosphere in the cell. As when the system was under vacuum, limiting current data (10 mV s-1 scan rate) was obtained approximately once every 15 min. Figure 4 plots the log of limiting current versus time from the start of the experiment (defined as the start of vacuum conditions). The empty circles represent data taken during vacuum, and the squares represent data taken during Ar flow. Sixty minutes after the start of the experiment and 15 min after Ar started flowing into the cell, the limiting current was seen to decrease. However, the next point (75 min) shows an increase in current. The upward trend continues until the limiting current of Fc attains a steady value at 150 min. There is a net
Effect of Argon on Ionic Liquids
J. Phys. Chem. C, Vol. 113, No. 18, 2009 7753 TABLE 1: Solubilities of Various Gases in [C2mim][NTf2] at 1 atm and Diffusion Coefficients of Fc in Gas-Saturated [C2mim][NTf2]a gas
solubility/mM
vacuum N2 CO2 Ar SO2
n/a
