1 Introduction

Room temperature ionic liquids [RTILs] may be used as solvents in many applications, as described in references [1–6], including electrochemistry [7–10]. In the latter case a reference system is necessary to compare potentials measured in different solvent-free RTILs. Different reference systems have been proposed, including pseudo-reference electrodes (metal immersed in an electrolyte, usually Pt or Ag), according to references [11, 12]. In such a case, the surface of the pseudo-reference system must be much larger in comparison to the working electrode; consequently, the current density at the reference is much lower. Half-cells based on the Ag|Ag+ or Ag|AgCl|Cl− systems (where a salt, being a source of Ag+ or Cl− ions, is dissolved in a mixture of acetonitrile with ionic liquid) have also been recommended as references for RTILs [13–15] and also for high temperature ILs (LiCl–KCl–CaCl2) [16]. Silver wire, immersed in a solution of silver(I) salt and cryptand 222 in molecular liquids [17] and ionic liquids [18], has been proposed as a stable reference system.

Finally, electrode potentials may be expressed versus an inner reference organometallic redox system, which consists of a large cation and its reduced form. The oxidized and reduced forms, both of large radius, may be assumed to be solvated similarly in different solvents. This leads directly to the assumption that redox potentials of such couples should be comparable in different solvents. Consequently, a given redox couple may be regarded as a universal potential reference. Organometallic redox couples such as bis(biphenyl)chromium(0)/(I) (BCr|BCr+), ferrocene|ferrocenium (Fc|Fc+), and cobaltocene|cobaltocenium (Cc|Cc+) have been investigated in various RTILs as described in the literature [10, 19–34], including a study on the applicability of cobaltocenium reduction as a reference for ionic liquids [30] and a detailed study on the simultaneous presence of both (Cc|Cc+) and (Fc|Fc+) couples in RTILs [32]. The general aim of the present study was to conduct a systematic investigation of the Cc/Cc+ redox reference system in various aprotic ionic liquids at different electrodes.

2 Experimental

2.1 Chemicals

Bis(cyclopentadienyl) cobalt(III) hexafluorophosphate (cobaltocenium hexafluorophosphate, \( {\text{Cc}}^{ + } {\text{PF}}_{6}^{ - } \), Aldrich), silver perchlorate (AgClO4, Fluka), and cryptand 222 (4.7.13.16.21.24-hexaoxa-1.10-diazabicyclo [8.8.8] hexacosane, Merck) were used as purchased. Acetonitrile (AN, Merck) was distilled before use. Room temperature ionic liquids: N-methyl-N-propylpyrrolidinium bis(trifluoromethanesulfonyl)imide (MePrPyrrNTf2) (Iolitec), N-butyl-N-methylimidazolium triflate (BuMeImOTf) (Iolitec), N-butyl-N-methylpyrrolidinium bis(trifluoromethanesulfonyl)imide (BuMePyrrNTf2) (Iolitec), N-butyl-N-methylpyrrolidinium triflate (BuMePyrrOTf) (Merck), diethylmethylsulphonium bis(trifluoromethanesulfonyl)imide ([Et2MeS][NTf2] (Iolitec), and triethylsulphonium bis(trifluoromethanesulfonyl)imide ([Et3S][NTf2] (Iolitec) were used as purchased. N-ethyl-N-methylimidazolium dicyanoimide (EtMeImN(CN)2) was obtained according to a published procedure [35] by metathesis of EtMeImBr with AgN(CN)2 in an aqueous solution [36]. N-methyl-N-propylpiperidinium bis(trifluoromethanesulfonyl)imide (MePrPipNTf2) was obtained according to a method described in the literature [37].

2.1.1 Water Content and Purity

The water content in aprotic ionic liquids, analyzed with a standard Karl–Fisher titrant (HYDRANAL® Composite 1, 1 mL/10 mg H2O), was below the detection limit. All of the RTILs were colorless. The purities of ionic liquids for electrochemical purpose were analyzed with cyclic voltammetry on platinum, gold and glassy carbon working electrodes. No reduction or oxidation peaks were detected between the anodic and cathodic decomposition potentials.

2.2 Apparatus and Procedures

Voltammetric measurements were performed in a three-electrode arrangement. Working electrodes: Au (1.50 mm diameter, 1.77 mm2), Pt (1.50 mm, 1.77 mm2) and glassy carbon (3.00 mm diameter, 7.07 mm2) were disc shaped and sealed in poly(tetrafluoroethylene). Before measurements, the electrodes were polished with aluminum oxide paste in water (Al2O3, 150 mesh, Merck) and then washed with acetone. The counter electrode was a platinum sheet (0.5 × 1.0 cm). The reference electrode consisted of a silver wire immersed in a solution of AgClO4 (0.01 mol·dm−3) and cryptand 222 (0.1 mol·dm−3) in acetonitrile [17]. The reference electrode compartment was separated by a glass frit from the cell containing the ionic liquid. Preparation of the solutions, weighing of the samples, and cell assembly were performed in a glove-box under a dry argon atmosphere. Tested electrolytes were deaerated with argon for 30 min prior to measurements. Voltammetric curves were obtained with the μAutoLab Electrochemical System (Eco Chemie, The Netherlands) at (25 ± 0.1)  °C. The initial scan was carried out to more negative potentials (reduction of cobaltocene: Cc+ + e− → Cc0) followed by the reverse anodic scan (oxidation of Cc0). Two reduction/oxidation scans were recorded in each case. The baseline of each neat ionic liquids was measured before experiments with Cc/Cc+ solutions. The ohmic resistance R between electrodes was determined from impedance spectra (using an ac impedance analyzer Atlas-Sollich, Poland), in the frequency range of 100 kHz to 1 Hz with 10 mV amplitude.

3 Results and Discussion

3.1 Electrolyte Conductivity and IR Ohmic Drop

The conductivity of the electrolyte influences the resistance between electrodes of the cell. In the case of ionic liquid electrolytes, the specific conductivity is typically between ca. 10 and 0.01 mS·cm−1, according to Ref. [8], which may lead to resistances differing by three orders of magnitude. The resistance was obtained by deconvolution of impedance spectra according to an equivalent circuit, consisting of resistance R in series with the Warburg impedance and charge transfer resistance, in parallel to the double layer capacity. The resistance R, determined from impedance spectra, was between ca. 200 Ω (Et3SNTf2) and 2,400 Ω (MePrPipNTf2). At a typical current level of 10 μA, the IR distortion of the potential was in the range of 0.2–24 mV.

3.2 CV Curves

Figure 1 presents typical cv curves, after background current and IR drop corrections, for cobaltocenium/cobaltocene (Cc+/Cc) reduction/oxidation in BuMeImOTf ([\( {\text{CcPF}}_{6}^{ - } \)] = 9.18 mmol·L−1) at potential sweep rates from (2 to 200) mV·s−1, versus the Ag|Ag+222, AN reference electrode. Potentials of the peak maximum, E pa and E pc, for Cc+ cathodic reduction and Cc anodic oxidation were −854 and −762 mV, respectively. Similar CV curves were obtained for a number of ionic liquids as solvents recorded at the three different electrodes (Pt, Au, GC). Measurements of ΔE 1/2 (Cc+/Cc0) over a period of 24 h gave stable values within ca. 1–2 mV, indicating no significant changes in the liquid junction potential between RTILs and the reference electrode electrolyte. Differences between the cathodic and anodic peak potentials, E pa−E pc, ½(E pa + E pc), and E p−E p/2 values (E p/2 is the half-peak potential), and peak current densities j pa and j pc, in the studied ionic liquids, are collected in Table 1. The difference between the cathodic and anodic peak potentials, E pa−E pc, is ≥81 mV, while the value predicted by the theory for reversible processes, according to Ref. [38], is (57–60) mV depending on the switching potential. The results indicate a quasi-reversible redox process for the cobaltocenium/cobaltocene couple. A similar behavior was found for the ferrocene|ferrocenium couple in a number of ionic liquids in our previous paper [34]. On the other hand, the E p−E p/2 values are close to 56 mV, typical of reversible processes [38]. In some protic RTILs, the E p−E p/2 values are close to the theoretical value of 56 mV, but in some cases they are higher (even as much as 70 mV). The formal potential ½(E pa + E pc) for the Cc+/Cc couple in aprotic ionic liquids may be approximated by the value−(831 ± 13) mV (versus the Ag|Ag+222 in AN reference). Formal potentials obtained in this study for cobaltocene may be referred to potentials for ferrocene measured in a number of protic and aprotic ionic liquids [34]. Table 2 presents differences between formal potentials of ferrocene/ferrocenium and cobaltocenium/cobaltocene couples in ionic liquids as well as molecular liquids; these results were calculated from the ½(E pa + E pc) values measured versus the cryptate electrode Ref. [34], or versus reference systems mentioned in references [15, 28–41]. Inspection of Table 2 shows that the E 1/2(Fc/Fc+)−E 1/2 (Cc+/Cc) difference may be approximated by the value (1.350 ± 0.020) V and is in agreement with values obtained by other authors [15, 32, 40, 41]. Such good agreement of the potential difference (±0.02 V) suggests that the solvation of both reference couples is nearly independent of the electrolyte.

Fig. 1
figure 1

Cyclic voltammetry for cobaltocene Cc+/Cc in BuMeImOTf on a Pt electrode at different sweep rates (from 2 to 200 mV s−1); reference electrode: Ag/Ag+222 in acetonitrile

Full size image
Table 1 Electrochemical parameters for cobaltocene cyclic voltammetry in aprotic ionic liquids at platinum, gold and glassy carbon electrodes, with reference electrode Ag|Ag+ (0.01 mol·L−1) cryptand 222 + (0.1 mol·L−1) AN
Full size table
Table 2 A comparison of formal potentials of cobaltocene and ferrocene in ionic liquids (this study and a previous paper or literature results)
Full size table

3.3 Diffusion

Diffusion coefficients of the cobaltocenium cation were calculated from the Randles–Sevcik equation for a one-electron reduction:

$$ \left| {j_{\text{pc}} } \right| = 0.4463\;\left( {\frac{{F^{3} }}{R\;T}} \right)^{1/2} \;\left[ {{\text{Cc}}^{ + } } \right]\;\;D^{1/2} \;\nu^{1/2} $$
(1)

where F is the Faraday constant, R is the gas constant, T is the temperature, [Cc+] is the cobaltocenium bulk concentration, D stands for the Cc+ diffusion coefficient, and v is the sweep rate. Figure 2 presents examples of the cathodic peak current density as a function of square root of the sweep rate. Such a linear j pc = f(\( \sqrt v \)) function was observed in all of the ionic liquids. Values of the cobaltocenium diffusion coefficients, collected in Table 3, show that in the case of viscous aprotic RTILs the diffusion coefficients are of the order of (10−6 or 10−7) cm2 ·s−1. Diffusion coefficients are between 0.52 × 10−7 cm2 ·s−1 and 5.2 × 10−7 cm2 ·s−1. Recently, three ionic liquids (EtMeImNTf2, BuMePyrrNTf2 and BuMeImPF6) were studied by cyclic voltammetry and rotating disc voltammetry [30]. The diffusion coefficient obtained here for BuMePyrrNTf2 is somewhat lower, ca. 0.72 × 10−7 cm2 ·s−1, in comparison with 1.29 × 10−7 cm2 ·s−1 reported in Ref. [30]. Values reported for other RTILs are D(EtMeImNTf2) = 3.27 × 10−7 cm2 ·s−1 and D(BuMeImPF6) = 3.73 × 10−8 cm2·s−1 [30]. Recently the diffusion coefficient of Cc+ in BuMeImBF4 and BuMeImPF6 was reported [32] to be slightly concentration sensitive. Values of D(Fc) in aprotic RTILs are usually of the order of (10−8–10−7) cm2 ·s−1, similar to D(Cc+). Diffusion coefficients of the ferrocene/ferrocenium and cobaltocenium/cobaltocene couples have also been reported for allyl substited pyrrolidinium, piperidinium and morpholinium-based ionic liquids [33]. Diffusion coefficient values are in the same range of (10−7–10−6) cm2 ·s−1, depending on IL’s viscosity.

Fig. 2
figure 2

Dependence of the cathodic current peak density on the square root of sweep rate for cobaltocene reduction in ionic liquids: black triangle, MePrPyrrNTf2; black circle, BuMePyrrNTf2; black diamond, Et3SNTf2

Full size image
Table 3 Cobaltocenium (Cc+) diffusion coefficients D (cm2 s−1) in ionic liquids, the ionic liquid viscosity η together with the Dη product
Full size table

According to the Stokes–Einstein equation, Eq. 2, the main factor influencing the diffusion coefficient is the medium’s viscosity, η,

$$ D\eta = \frac{{k_{\text{B}} T}}{{6{{\uppi}}r}}, $$
(2)

where k B is the Boltzmann constant and r is the cobaltocenium radius.

Table 3 shows the Dη values calculated on the basis of literature data on ionic liquid viscosities published in references [5, 42–48], which correspond to the Walden product. Inspection of Table 3 suggests that the Dη product may be approximated by the value (0.80 ± 0.4) × 10−7 cm·g·s−2. This result may also suggest that the cobaltocenium radius is constant and independent of the medium. An interesting aspect is a comparison (ratio) of diffusion coefficients of both popular metallocecenes, ferrocene and cobaltocenium, used as electrode potential references, D(Fc)/D(Cc+). Here, the solvation of both forms may be different due to the fact that ferrocene is a neutral molecule, while cobaltocenium is a cation. A comparison of D(Cc+) values (Table 3) with the corresponding D(Fc) literature values shows that D(Fc)/D(Cc+) > 1. For example, according to Ref. [31], the D(Fc)/D(Cc+) ratio is 3.0 in EMImBF4 {D(Fc) = 2.70 × 10−7 cm2 ·s−1}, while the corresponding value in EMImNTf2, reported in Ref. [30] is 1.39 {D(Fc) = 3.27 × 10−7 cm2 ·s−1}. For comparison, the D(Fc)/D(Cc+) ratio is also higher than unity in solutions with molecular solvents. For example D(Fc)/D(Cc+) = 1.77 in acetonitrile {D(Fc) = 2.3 × 10−5 cm2·s−1 [49] and D(Cc+) = 1.30 × 10−5 cm2·s−1 [39]}. All of this information suggests that the Fc molecule and Cc+ cation, although of similar shape and radius, are probably solvated differently, as suggested in Ref. [50]. The cobaltocenium cation may interact with anions present in the electrolyte and therefore have a higher effective radius and, hence, a somewhat lower diffusion coefficient. In general, diffusion coefficients of large organic compounds determined in RTILs are on the order of 10−7 cm2 ·s−1 which is two orders of magnitude lower than in conventional molecular solvents [51].

4 Conclusions

1.:

Well-defined, diffusion controlled, anodic and cathodic peaks were found for the Cc+/Cc redox couple in a number of RTILs

2.:

Values of E p−E p/2 are close to 56 mV, typical of reversible processes. On the other hand, the difference between cathodic and anodic peak potentials, E pa−E pc, is ≥81 mV, while the value predicted by theory for reversible processes is 57–60 mV, which indicates a quasi-reversible redox process

3.:

The E 1/2(Fc/Fc+)−E 1/2 (Cc+/Cc) difference may be approximated by the value (1.350 ± 0.020) V

4.:

Values of the cobaltocenium (Cc+) diffusion coefficients D are in the range of 0.5 × 10−7 cm2 ·s−1–5.2 × 10−7 cm2 ·s−1, depending on the medium’s viscosity η. The Dη value, corresponding to the Walden product, falls within the narrow range of 0.61 × 10−7 cm·g·s−2–1.23 × 10−7 cm·g·s−2