Determination of Water in Room Temperature Ionic Liquids by

Feb 28, 2012 - Determination of Water in Room Temperature Ionic Liquids by Cathodic Stripping Voltammetry at a Gold Electrode. Chuan Zhao*†, Alan M...
14 downloads 18 Views 1MB Size
Download PDF
Article pubs.acs.org/ac

Determination of Water in Room Temperature Ionic Liquids by Cathodic Stripping Voltammetry at a Gold Electrode Chuan Zhao,*,† Alan M. Bond,*,‡ and Xunyu Lu† †

School of Chemistry, The University of New South Wales, Sydney, NSW 2052, Australia School of Chemistry and ARC Centre for Green Chemistry, Monash University, Clayton, Victoria 3800, Australia

‡

S Supporting Information *

ABSTRACT: An electrochemical method based on cathodic stripping voltammetry at a gold electrode has been developed for the determination of water in ionic liquids. The technique has been applied to two aprotic ionic liquids, (1-butyl-3ethylimidazolium tetrafluoroborate and 1-butyl-3-methylimidazolium hexafluorophosphate), and two protic ionic liquids, (bis(2-hydroxyethyl)ammonium acetate and triethylammonium acetate). When water is present in an ionic liquid, electrooxidation of a gold electrode forms gold oxides. Thus, application of an anodic potential scan or holding the potential of the electrode at a very positive value leads to accumulation of an oxide film. On applying a cathodic potential scan, a sensitive stripping peak is produced as a result of the reduction of gold oxide back to gold. The magnitude of the peak current generated from the stripping process is a function of the water concentration in an ionic liquid. The method requires no addition of reagents and can be used for the sensitive and in situ determination of water present in small volumes of ionic liquids. Importantly, the method allows the determination of water in the carboxylic acid-based ionic liquids, such as acetate-based protic ionic liquids, where the widely used Karl Fischer titration method suffering from an esterification side reaction which generates water as a side product.

W

reagents is required for monitoring the water concentration in ionic liquids, even during the course of an experiment. The above considerations describe the qualitative impact of water in ILs when present as an adventitious impurity. A quantitative knowledge of water concentration in ionic liquid is required in fundamental and practical studies related to ionic liquid polarities,9 solution theromodynamics,10 liquid−liquid phase separation11 extraction,12 and synthesis of nanoparticles.13,14 Importantly, when present in ionic liquid media, water is not necessarily as chemically inert as is the case when introduced into molecular solvents. Distinctly different molecular structure and reactivity have been observed when water is dissolved in ionic liquids relative to the situation in pure liquid water, as a result of the dramatic difference in intermolecular interactions derived from van der Waals bonds, hydrogen bonds, and electrostatic interactions.9,15−24 The modified molecular structure inevitably affects the chemical properties of both the ionic liquid and the bound water. Recently we discovered that water, when present as an impurity in an ionic liquid, can be photooxidized to oxygen and thus serve as an electron donor for photochemical synthesis of semiconducting silver-tetracyanoquinodimethane nanowires.13 Enhanced photooxidation of water also has been achieved in

ater is one of the most significant impurities present in room temperature ionic liquids (RTILs). It can originate in RTILs from the synthesis or from absorption from air during storage or handling prior to or during the course of an experiment conducted under benchtop laboratory conditions.1,2 Whereas many aprotic ionic liquids such as [BMIM][BF4],3 protic ionic liquids,4 and distillable ionic liquids5 are totally miscible with water, some so-called “hydrophobic” RTILs such as [BMIM][PF6] and [EMIM][Ntf2] are actually reasonably hygroscopic up to their water concentration saturation level.6 The presence of water in RTILs can substantially alter their physicochemical properties; e.g. decrease the viscosity, increase the conductivity, narrow the electrochemical potential window.1,2,7 The water concentration in ionic liquids is strongly dependent on the experimental environment. Studies with ionic liquids undertaken under benchtop laboratory conditions will almost certainly contain much higher concentrations than those present under glovebox or vacuum conditions. Recently, we have noted that nitrogen purging, a popular protocol to remove dissolved oxygen from aqueous, molecular solvent and ionic liquid media, also removes water from “wet” ionic liquids.8 All these effects are highly likely to alter the water levels present in ionic liquids and thus introduce significant variation in results from laboratory to laboratory. In this content, an in situ technique that does not require addition of © 2012 American Chemical Society

Received: November 29, 2011 Accepted: January 31, 2012 Published: February 28, 2012 2784

dx.doi.org/10.1021/ac2031173 | Anal. Chem. 2012, 84, 2784−2791

Analytical Chemistry

Article

method, the growth of surface oxide on gold electrodes has been studied in detail for the first time in an ionic liquid medium.

ionic liquid media when the polyoxometalate anion [P2W18O62]6‑ is irradiated with light.25 These new forms of chemical reactions are attributed in part to the modified structure of water in ionic liquids, which is strongly concentration dependent. Currently, water determinations in ionic liquids are routinely carried out by the Karl Fischer (KF) titration method. This analytical approach is based on the modified Bunsen Reaction between I2 and SO2 in a nonaqueous medium containing a primary alcohol (typically methanol) as the solvent, and a base (e.g., pyridine, imidazole) as the buffering agent. The alcohol reacts with SO2 and base to form an intermediate alkylsulfite salt, which is then oxidized by iodine to an alkylsulfate salt in the presence of water (eq 1)26

■

EXPERIMENTAL SECTION Ionic Liquids. The aprotic ionic liquids, 1-n-butyl-3methylimidazolium tetrafluoroborate ([BMIM][BF4]), 1butyl-3-methylimidazolium hexafluorophosphate ([BMIM][PF6]), and 1-n-butyl-3-methylimidazolium chloride from Merck (high purity) were used as provided by the manufacturer. The protic ionic liquids, bis(2-hydroxyethyl)ammonium acetate ([DEA][Ac]) and triethylammonium acetate ([TEtA][Ac]), were synthesized and characterized as described elsewhere.4 Preparation of Water Standard Solutions in Ionic Liquids. Water is present as an adventitious impurity in almost every ionic liquid. However, to prepare an ionic liquid solution with a known low water concentration is not straightforward. Conventional addition of water into an ionic liquid with micropipets is not always suitable because the concentration added will be too high for analytical applications. For cost reasons, it is impractical to use a large volume of ionic liquids and make a series of dilutions, as can be done with aqueous solutions. For many ionic liquids, especially hydrophobic ones such as [BMIM][PF6], even the addition of a single small drop of water can exceed the solubility limit giving rise to the formation of two immiscible phases. In addition, ionic liquids can rapidly take up water from the atmosphere during the course of experiments and thereby introduce errors in the preparation of standard solutions. In this study, ionic liquid/water standard solutions are prepared by a procedure which is based on water uptake from the gas phase.2 A custom designed electrochemical cell shown in Figure S1 (Supporting Information) was devised. Typically 1 mL of ‘dried’ ionic liquid e.g. [BMIM][BF4] was transferred inside a nitrogen filled glovebox into a cell where all the electrodes have been placed for following electrochemical measurements. The gas inlet and outlet were initially sealed with parafilm. The sealed cell was then transferred from the drybox into the laboratory and weighed on an analytical balance. The weight of the water absorbed from the atmosphere was obtained by subtraction of the initial weight. The water concentration in wt.%, g mL−1, or ppm was then calculated. To prepare very ‘wet’ water/ionic liquid standard solutions, a stream of nitrogen, presaturated with water by bubbling through a wash bottle, was passed over the surface of the ionic liquid. A magnetic stirrer was used to agitate the ionic liquid during this step in order to accelerate the rate of water uptake (e.g., 60 h is need to saturate [BMIM][PF6] with water by absorption from water saturated gas phase) and to adequately mix the water with the ionic liquid. Nitrogen may also dissolve into the ionic liquid but typically has only a very small solubility and also is voltammetrically ‘inert’. No effect of nitrogen was observed in the electrochemical experiments. Electrochemical Instrumentation and Procedures. Karl Fischer titration measurements were carried out with a 831 KF Coulometer (Metrohm, Herisau, Switzerland). Voltammetric experiments were undertaken at (20 ± 1) °C with a BAS 100B/ W electrochemical workstation (Bioanalytical System, West Lafayette, IN). The uncompensated resistance values (Ru), typically from 1000−10000 Ohm were measured by applying a potential step (ΔE = 50 mV) in a potential region where no

In the KF method, the ionic liquid is diluted with methanol before titration, and the amount of water present in the sample is calculated based on the quantity of iodine consumed in the titration. In the case of very low level concentration of water (5 °C) and continuously absorbs water from water saturated nitrogen gas phase (Figure 4a). The initial rate of uptake of water is rapid but then decreases. A water content of 10% is achieved in 2 h. With an increase of water content in the ionic liquid, cyclic voltammograms exhibit a decrease in the anodic potential limit (Figure 4b), as reported in other studies.2,7 Concurrently, the gold oxide formation and reduction peak current magnitude increase, while peak potentials shift to more negative values.

increasing Epol from 1.3 to 1.6 V (Figure 2b), the reductive stripping peak current (ipred) magnitude increases and the peak potential (Epred) shifts first to less positive potentials by 37 mV, consistent with growth of a thin oxide film whose thickness is increasing as Epol becomes more positive. On further increasing Epol from 1.6 to 2.0 V, Epred shifts back to more positive potentials by 60 mV. This reversal is believed to be associated with the development of the higher oxide Au2O3 under more positive polarization potentials.30 The influence of polarization time (tp) on the gold oxide reduction was investigated in [BMIM][BF4] by using linear sweep voltammetry (not shown). The initial potential was held at 1.8 V for tp values of 2−30 s, before the potential was swept negatively to −0.2 V at a scan rate of 0.05 V s−1. An increase of tp results in an increase of gold oxide stripping peak current without significant change of peak potential. Figure 3a shows the influence of temperature on the gold oxide growth and reductive stripping in [BMIM][BF4] when Ep = 1.8 V. An increase in T from 295 to 333 K leads to an increase in the magnitude of both the oxidation peak and reduction peak current. Concomitantly, the oxide formation peak potential shifts to more negative values by 40 mV, whereas the oxide reduction peak shifts positively (Figure 3b). The absolute values of peak potentials are uncertain because the temperature dependence of the Ag QRE is unknown. However, the decrease of hysteresis (potential difference between anodic peak and 2787

dx.doi.org/10.1021/ac2031173 | Anal. Chem. 2012, 84, 2784−2791

Analytical Chemistry

Article

increase in water content, with peak potentials shifting to more positive values. Figure S4a (see the Supporting Information) provides data on the water adsorption profile of [TEtA][Ac] as function of time. This much less viscous ionic liquid absorbs water far more quickly and absorption of 20% water is now achieved in less than 2 h, although the rate of increase at longer times was also slow. The voltammograms in this IL show similar trends as in the other three ionic liquids. Thus, an increase of the oxide reduction stripping peak current with water concentration was observed although the stripping process is broader and less well-defined (Figure S4b). Determination of Water in Aprotic Ionic Liquids and Protic Ionic Liquids by Cathodic Stripping Voltammetry. The above results show that the development of surface oxides (correspondingly the stripping peak current or surface charge) depends strongly on the water concentration in ionic liquids and also the properties of ionic liquids themselves, such as hydrophobicity and viscosity, and the electrochemical detection conditions such as polarization potentials, polarization time, and temperatures. Cyclic voltammograms obtained at various scan rates also show that the stripping peak current increases with scan rate (not shown). Therefore, for quantification purposes, the analytical protocol needs to be optimized for each ionic liquid. Cathodic linear sweep stripping voltammetry (LSSV) using a gold electrode was applied to establish a sensitive method for water determination. Cathodic stripping voltammetry also has been applied in ionic liquids for determination of trace chloride using a silver electrode.33 Typically, after exposing the ionic liquid samples to air, water is adsorbed into the ionic liquids. The gold macrodisk working electrode was then held at 1.8 V vs Ag QRE for 120 s. Then a 5 s rest period was applied before applying a linear potential sweep from 1.8 to 0 V at a scan rate of 100 mV s−1. Figure S5 (see the Supporting Information) shows the linear sweep stripping voltammograms obtained for [BMIM][BF4] and [DEA][Ac], following the addition of water by use of the method of standard additions. The sharp stripping peaks increase with addition of water, while the peak potentials shift with water concentration, in accordance with expectations associated with formation of thicker oxide films. The voltammograms also show that the background current prior to the stripping process is very small, allowing a very high signal-to-background ratio to be achieved in the method. Quantification of low water content in ionic liquids was performed by using the standard addition method. In this case, the dried ionic liquid samples were exposed to air, and the LSSV signals were recorded with increasing absorbed water content. Figure 5 shows the standard addition curves established for [BMIM][BF4], [BMIM][PF6], [DEA][Ac], and [TEtA][Ac] using the stripping peak current with respect to the water concentration (Table 1). Table 1 also summarizes the LOD and LOQ data for the four ionic liquids using the criteria of the lowest value in the linear plot as the limit of quantification (LOQ) and the signal-to-background ratio as limit of detection (LOD).34 Among the four ionic liquids, hydrophobic [BMIM][PF6] has the highest LOD and LOQ values. This could be partly attributed to the higher hydrophobicity and/or higher viscosity of [BMIM][PF6]. Among the three water-miscible ionic liquids, the LOQ and LOD appear to decrease with decreasing viscosity. The LOQ and LOD for [DEA][Ac] are found to be higher than [BMIM][BF4], and the least viscous [TEtA][Ac] has the

Figure 4. (a) Time dependence of absorption of water from a watersaturated nitrogen gas phase into the hydrophilic aprotic ionic liquid [BMIM][BF4] and (b) cyclic voltammograms recorded at a scan rate of 0.1 V s−1 at a gold electrode showing the decrease in anodic potential limit and the increase of surface oxide with increase in water concentration.

[BMIM][PF6], in contrast to [BMIM][BF4], is hydrophobic and only partially miscible with water. Figure S2a (see the Supporting Information) shows that the rate of uptake of water with [BMIM][PF6] is also initially fast but approaches a plateau after 1 h. The total water absorbed after 2 h is ca. 1.8%, which represents close to a saturated solution. Nevertheless, the voltammograms obtained in [BMIM][PF6] (Figure S2b) exhibit analogous trends to those seen in [BMIM][BF4] with regard to the increase of gold oxide formation and negative potential shifts. [DEA][Ac] and [TEtA][Ac] are both fully miscible with water at room temperature. However, [DEA][Ac] has a very much larger viscosity (336 mPa s, 25 °C) than [TEtA][Ac] (11 mPa s, 25 °C).4 It is therefore interesting to establish if viscosity plays a role in the water uptake rate. In [DEA][Ac], a steady increase in water content was observed with 20% water being absorbed into the ionic liquid after 4 h of exposure to water saturated nitrogen (Figure S3a, Supporting Information). Cyclic voltammograms again exhibit the narrowing of the anodic potential window with increase in water content in this protic ionic liquid (Figure S3b). The gold oxide formation process on the anodic scans is less pronounced than in aprotic [BMIM][BF4] and [BMIM][PF6]. However, the oxide reduction stripping peaks were clearly detected on the reverse scan, and the peak current magnitude increased steadily with an 2788

dx.doi.org/10.1021/ac2031173 | Anal. Chem. 2012, 84, 2784−2791

Analytical Chemistry

Article

Figure 5. Dependence of stripping peak current on concentration of water absorbed into (a) [BMIM][BF4], (b) [BMIM][PF6], (c) [DEA][Ac], and (d) [TEtA][Ac]. The cathodic stripping parameters for the [BMIM][BF4] are as follows: polarization potential 2.2 V, polarization time 120 s, rest period 5 s, linear sweep rate 0.1 V s−1, for [BMIM][PF6]: polarization potential 2.2 V, polarization time 5 s, rest period 5 s, linear sweep rate 0.1 V s−1, for DEAA: polarization potential 1.2 V, polarization time 120 s, rest period 30 s, linear sweep rate 0.1 V s−1, and for [TEtA][Ac]: polarization potential 1.2 V, polarization time 120 s, rest period 30 s, linear sweep rate 0.1 V s−1.

Table 1. Linear Calibration Ranges, Sensitivity, Limit of Quantification (LOQ), and Limit of Detection (LOD) of Cathodic Linear Sweep Stripping Voltammetrya for the Determination of Water in [BMIM][BF4], [BMIM][PF6], [DEA][Ac], and [TEtA][Ac] ionic liquids [BMIM][BF4] [BMIM][PF6] [DEA][Ac] [TEtA][Ac] a

linear range/ppm 370−2350 640−2640 500−2500 50−3450

sensitivity ip/A ip/A ip/A ip/A

= = = =

−6

−8.07 × 10 6.83 × 10−7− −5.52 × 10−7 −3.26 × 10−6

A− 2.43 A− A−

−9

3.52 × 10 ([H2O]/ppm) × 10−9([H2O]/ppm) 2.48 × 10−10([H2O]/ppm) 3.40 × 10−10([H2O]/ppm)

R2

LOQ/ppm

LOD/ppm

0.9980 0.9926 0.9970 0.9968

370 640 500 50

140 490 150 10

The LSSV conditions are as in Figure 5.

regions are formed. The first (lower water content) linear region is always more sensitive (higher slopes) than the second ones (higher water content). Clearly, the standard addition method cannot be applied to very ‘wet” ILs and reference to a calibration curve will be needed. Peak areas are also used instead of peak height for quantification. In this case, similar trends are observed when using peak area, however, with decreased linearity in some cases (not shown) so this approach is not recommended. Results with the proposed method were compared with KF titration data. For this purpose, 1 mL of [BMIM][BF4] was collected from a bottle that has been exposed to the laboratory benchtop condition. The cathodic stripping voltammetric measurement using the following parameters, polarization

smallest LOQ and LOD. The viscosity of the ionic liquids appears to play a significant role in the method, which is not surprising as the formation of the gold oxide is mass transport dependent. The proposed method has been applied for the determination of water in very ‘wet’ ionic liquids, where the KF method suffers due to prolonged titration time and excessive reagent consumption. In this case, the dried ionic liquid samples were exposed to water saturated nitrogen flow under stirred conditions in order for rapid water uptake and to achieve complete homogenization. Table 2 summarizes the dependence of stripping peak current with respect to the concentration of added water in very ‘wet’ ionic liquids. For all four ionic liquids, two linear stripping peak current versus water concentration 2789

dx.doi.org/10.1021/ac2031173 | Anal. Chem. 2012, 84, 2784−2791

Analytical Chemistry

Article

Table 2. Linear Range and Sensitivity of Cathodic Linear Sweep Stripping Voltammetrya for Determination of Water in Very “Wet” [BMIM][BF4], [BMIM][PF6], [DEA][Ac], and [TEtA][Ac] ionic liquids [BMIM][BF4] [BMIM][PF6] [DEA][Ac] [TEtA][Ac]

linear range/ppm of added water 1.77 × 10 4.42 × 104 1.604.42 × 4.81 × 105 1.40 × 104 9.02 × 104 2.01 × 104 1.97 × 104 4

− 4.42 × 10 − 1.084.42 × 105 105 − 3.50 × 105 − 1.61 × 106 − 6.67 × 104 − 2.65 × 105 − 2.75 × 104 − 2.07 × 105 4

Δ linear range/ppm 2.64 6.40 1.90 1.13 5.27 1.75 2.55 1.88

× × × × × × × ×

R2

sensitivity

4

ip/A ip/A ip/A ip/A ip/A ip/A ip/A ip/A

10 104 105 106 104 105 104 105

= = = = = = = =

−6

−10

1.22 × 10 A − 1.75 × 10 ([H2O]/ppm) −4.72 × 10−6A − 4.31 × 10−11([H2O]/ppm) −2.15 × 10−6 A − 1.93 × 10−11([H2O]/ppm) −8.01 × 10−6 A − 4.42 × 10−12([H2O]/ppm) 4.37 × 10−7 A − 5.88 × 10−11([H2O]/ppm) −2.35 × 10−6 A − 2.34 × 10−11([H2O]/ppm) −8.31 × 10−7 A − 1.25 × 10−10([H2O]/ppm) −5.39 × 10−6 A − 1.45 × 10−11([H2O]/ppm)

0.9977 0.9950 0.9984 0.9914 0.9922 0.9936 0.9986 0.9934

a

The LSSV conditions for [BMIM][BF4] and [BMIM][PF6]: polarization potential 1.8 V, polarization time 5 s, rest period 5 s, linear sweep rate 0.1 V s−1; and for [DEA][Ac] and [TEtA][Ac]: polarization potential 1.2 V, polarization time 120 s, rest period 5 s, linear sweep rate 0.1 V s−1.

potential 2.0 V, polarization time 120 s, rest period 5 s, linear sweep rate 0.1 V s−1, suggested that the sample contains 1200 ± 100 ppm of water. Subsequent use of KF titration measurement indicated 1340 ± 80 ppm of water, which is in reasonable accordance with the cathodic stripping method. Halides, practically chloride, are common impurities in ionic liquids and often originate from synthetic methods involving the use of relevant organic halide salts in a metathesis reaction. Since halides can be relatively easily removed from ionic liquids by washing procedures, the level of halides in commercially available ionic liquids is usually 1000 ppm. This level of chloride is much larger than that usually present in ionic liquids, suggesting good selectivity of the proposed method against this impurity. Other impurities may also be present in an ionic liquid. If they can be oxidized during the preaccumulation step or compete with the gold oxide reaction via formation of an insoluble gold compound, then this should lead to changes in stripping voltammograms via introduction of a shoulder, a shift in peak position, the appearance of a new process, or a change in waveshape. It is therefore recommended to routinely measure the shapes (peak width at half wave height) and peak potentials of the stripping voltammograms against those encountered in the pure ionic liquid, if available, when a calibration curve method is used, or by comparing these characteristics before and after deliberate addition of water when using a standard addition method. Employment of this kind of protocol into the analytical procedure will provide a simple cross check against the possible presence of interference and hence enhance the fidelity of the voltammetric method for water determination. Interferences arising with respect to the presence of easily oxidized compounds also may occur in the Karl Fisher method but are not amenable to this simple kind of detection. Comparison with the Karl Fisher Titration Method. Karl Fischer (KF) titration, being based on coulometry, is an absolute method of analysis that should determine all water present in the sample without the need for a standard or calibration curve. This represents a major difference between

the KF method and the proposed cathodic stripping electrochemical one where calibration is needed for each ionic liquid. Data in Table 1 show that the proposed cathodic stripping electrochemical method is not as sensitive as the KF one, although a lower detection limit may be achieved by increasing the preconcentration time, stirring the solution during the preconcentration stage, increasing the stripping scan rate, and using pulse related cathodic stripping techniques. Indeed, the linear ranges and the LOQ are restricted by the smallest water concentration that can be accurately introduced into the ILs reported in this study. Nevertheless, the stripping method provides excellent performance over commonly encountered water concentration ranges that are found when working under normal laboratory benchtop conditions. This outcome follows naturally, since our experimental design essentially mimics the water adsorption process from ambient atmosphere. Moreover, it can be used to detect water in very “wet” (>5%) ionic liquids where the KF method becomes impractical. Furthermore, the method offers several significant advantages over the KF method: (i) Foremost, it allows the detection of the water concentration in carboxylic acid-based protic ionic liquids without the interference of esterification side reactions. This is important particularly with recent rapid development and applications of protic ionic liquids. (ii) The method requires only very small sample volumes (usually