Anal. Chem. 1998, 70, 3114-3118
A Simple Method for Examining the Electrochemistry of Metalloporphyrins and Other Hydrophobic Reactants in Thin Layers of Organic Solvents Interposed between Graphite Electrodes and Aqueous Solutions Chunnian Shi and Fred C. Anson*
Arthur Amos Noyes Laboratories, Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California 91125
A method is described for the preparation of stable, adherent, thin layers of organic solvents interposed between the surfaces of graphite electrodes and aqueous supporting electrolytes in which they are immersed. The electrochemistry of reactants dissolved in the thin layers is examined and utilized to evaluate surface coverages and formal potentials for molecules, such as cobalt tetraphenylporphyrins, that do not exhibit useful electrochemical responses when adsorbed on graphite. The thin layers of organic solvent can be used to concentrate analytes by extracting them from dilute aqueous solutions to produce enhanced sensitivity in electroanalytical applications. Electron transfer across the liquid/liquid interface created by the presence of the thin layer of immiscible organic solvent is also demonstrated.
Metalloporphyrins adsorbed on the surfaces of graphite, glassy carbon, or other inert electrodes are widely used as electrocatalysts for the electrochemical reduction or oxidation of a variety of important molecules (O2, H2O2, NO, HNO2, CO2, etc.). Efforts to establish the mechanisms by which the adsorbed metalloporphyrins operate as catalysts typically require knowledge of the quantities of the porphyrin present on the electrode surface and of the electrode potentials where the metal centers in the adsorbed porphyrins undergo changes in oxidation state from catalytically inactive to active forms. For example, cobalt(III) porphyrins must be reduced to their cobalt(II) forms to serve as catalysts for the reduction of O2,1 and iron porphyrins must be reduced to the Fe(0) oxidation state to become catalytically active toward the reduction of CO2.2 In cases where the adsorbed catalyst gives rise to a clear, reversible cyclic voltammetric response in the absence of substrate, its formal potential can be readily evaluated and measurement of the area under the voltammetric peaks allows the quantity of electroactive catalyst present on the electrode
surface to be determined. Examples of catalysts that exhibit these properties include cobalt tetramethylporphyrin3 and cobalt porphine.4 By contrast, cobalt tetraphenylporphyrin (CoTPP) adsorbed on graphite yields very indistinct cyclic voltammetric responses which has led to a wide range of reported values for the formal potential of the Co(III/II) couple of this porphyrin adsorbed on graphite.1,5-7 We have found that CoTPP and other molecules that do not exhibit clear cyclic voltammetric responses when adsorbed on graphite can be made to do so by dissolving them from the electrode surface into a thin layer of an immiscible organic solvent that is interposed between the electrode surface and the aqueous supporting electrolyte. This report is devoted to a description of the simple procedure that has been developed to produce stable, thin layers of organic solvents at electrodes and to studies of some electrochemical processes within the thin layers. The thin layers are also useful in studies of the rates of charge transfer across liquid/liquid interfaces as described in a separate report.8 EXPERIMENTAL SECTION Materials. Cobalt porphyrins were obtained from Porphyrin Products, Inc. (Logan, UT). They were purified by column chromatography on neutral alumina. After chromatography the porphyrins exhibited spectra that matched those previously published.7,9 Co(salen) (salen ) N,N′-ethylenebis(salicylideneamine)) was available from previous studies. Its preparation followed a literature procedure.10 CHCl3 and CH2Cl2 (Aldrich) were passed through a column of basic alumina just before use to remove traces of an oxidizing impurity. Benzonitrile (Aldrich, HPLC grade), nitrobenzene (Fluka), and other chemicals were of high purity and were used as received. Laboratory distilled (3) (4) (5) (6) (7)
* Corresponding author: (e-mail)
[email protected]; (phone) 626 395 6000; (fax) 626 405 0454. (1) Ni, C.-L.; Anson, F. C. Inorg. Chem. 1985, 24, 4754. (2) Hammouche, M.; Lexa, D.; Momenteau, M.; Saveant, J.-M. J. Am. Chem. Soc. 1991, 113, 8455.
Shi, C.; Anson, F. C. Inorg. Chem. 1998, 37, 1037. Shi, C.; Steiger, B.; Yuasa, M.; Anson, F. C. Inorg. Chem. 1997, 36, 4294. Buttry, D. A.; Anson, F. C. J. Am. Chem. Soc. 1984, 106, 59. Fukuzumi, D.; Mochizuki, S.; Tanaka, T. Inorg. Chem. 1989, 28, 2459. Yuasa, M.; Steiger, B.; Anson, F. C. J. Porphyrins Phthalocyanines 1997, 1, 181. (8) Shi, C.; Anson, F. C. J. Phys. Chem., submitted. (9) Wayland, B. B.; Minkiewicz, J. V.; Abd-Elmageed, M. E. J. Am. Chem. Soc. 1974, 96, 2795. (10) Gilbert, W. C.; Taylor, L. T.; Dillard, J. G. J. Am. Chem. Soc. 1973, 95, 2477.
3114 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
S0003-2700(98)00426-0 CCC: $15.00
© 1998 American Chemical Society Published on Web 07/03/1998
water was further purified by passage through a purification train (MilliQ Pus). Cylindrical pyrolytic graphite electrodes with the edges of the graphitic planes exposed were mounted to stainless steel shafts with heat-shrinkable tubing to produce an exposed circular disk with an area of 0.32 cm2. Electrical connection was made to the backside of the disk using a drop of Hg and a copper wire. Apparatus and Procedures. Conventional electrochemical cells, instrumentation, and procedures were employed. The reference electrode was a sodium chloride saturated calomel electrode (SSCE) with a potential 5 mV more negative than a standard SCE. The following procedure was used to position a thin layer of organic solvent, benzonitrile or nitrobenzene, between the surface of the graphite electrodes and the aqueous solution in which they were immersed: The graphite electrode surface was prepared by polishing with wet 600-grit SiC paper followed by washing and sonication in pure water. The surface was dried gently with an adsorbent tissue, and then the air from a heat gun (set at 650 °C) was blown across its surface for several seconds to remove residual water. The resulting surface was moderately hydrophobic as indicated by the failure of a drop of water to spread across its surface.11 The cooled electrode was held upside down while 1 µL of the desired solvent was transferred to the dry surface with a syringe. In contrast with water, the organic solvent spread rapidly across the graphite surface and adhered to it. The electrode was immediately inverted and immersed in the aqueous supporting electrolyte. In cases where adsorbed reactants were desired on the electrode surface, the polished and sonicated surface was exposed to a solution of the adsorbate and the adsorption allowed to proceed or aliquots of solutions of the adsorbate in CHCl3 or CH2Cl2 were transferred to the electrode surface and the solvent was allowed to evaporate before the thin layer of a second organic solvent was applied to the surface. RESULTS AND DISCUSSION Formation and Stability of the Thin Layers of Organic Solvents. The simple procedure described in the Experimental Section for forming adherent thin layers of organic solvents on graphite electrodes worked well with nitrobenzene or benzonitrile as the solvents. It was less successful with more volatile solvents such as benzene or dichloromethane. Typically, the thickness of the thin layer of organic solvent employed was ∼30 µm. A schematic depiction of a graphite electrode with the thin layer of organic solvent in place is shown in Figure 1. The spontaneous spreading of the organic solvents across the graphite surface produced a thin coating of the organic phase that effectively prevented reactants in the aqueous phase that were not soluble in the organic phase from reaching the electrode surface. Shown in Figure 2 are cyclic voltammograms for a solution of Fe(CN)64- at the bare EPG electrode (curve A) and at the same electrode after a thin layer of nitrobenzene was placed on its surface (curve B). It is apparent that the nitrobenzene layer prevents the Fe(CN)64- anions from reaching the underlying graphite surface to be oxidized. When the experiment was repeated using nitrobenzene in which 1.8 mmol/L ferrocene had been dissolved, the voltammo(11) Kinoshita, K. Carbon Electrochemical and Physicochemical Properties; John Wiley and Sons: New York, 1988; p 149.
Figure 1. Schematic depiction of the electrochemical cell with the edge plane graphite (EPG) electrode covered with an adherent thin layer of an immiscible organic solvent. The thin layer is much thinner (∼30 µm) than indicated in the drawing.
Figure 2. Cyclic voltammetric responses at an EPG electrode in a 0.73 mM solution of Fe(CN)64- in 2 M HClO4. (A) Bare electrode. (B) Repeat of (A) after a thin layer of pure nitrobenzene was applied to the surface of the electrode as in Figure 1. (C) Repeat of (B) but with the thin layer consisting of 1.8 mM ferrocene in nitrobenzene. Scan rate, 5 mV s-1.
gram in Figure 2C was obtained. The ferrocene dissolved in the thin layer of nitrobenzene readily undergoes oxidation at the graphite electrode. This result demonstrates that the conductance of the thin layer of nitrobenzene is adequate to sustain the flow of significant electrochemical currents. Thus, the lack of current Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3115
Figure 3. Cyclic voltammetric responses at an EPG electrode covered with a thin layer of nitrobenzene. (A) The aqueous phase contained 0.70 mM Fe(CN)63- in 2 M HClO4. (B) The aqueous phase contained only 2 M HClO4. Ferrocenium (1.3 mM) (generated by the electro-oxidation of ferrocene) was present in the thin layer of nitrobenzene. (C) Repeat of (B) with 1.9 mM Fe(CN)63- present in the aqueous phase. Scan rate, 5 mV s-1 throughout.
in Figure 2B is not the result of the low ionic conductance of the organic phase. The supporting electrolyte in the nitrobenzene phase is HClO4 that partitions into the thin layer from the adjoining aqueous solution of 2 M HClO4. Nitrobenzene that was equilibrated with 2 M HClO4 by vigorous agitation in a separatory funnel and then analyzed by titration with a standard solution of NaOH was found to contain ∼2 mM HClO4. This concentration of supporting electrolyte was too small to allow undistorted cyclic voltammograms for ferrocene to be obtained in a conventional experiment in which the electrode was immersed in a 1.8 mM solution of ferrocene in nitrobenzene. However, in the configuration of Figure 1 the ohmic resistance of the thin nitrobenzene layer was small (∼31 Ω), and the undistorted voltammogram in Figure 2C resulted. The Fe(CN)64- in the aqueous solution used to record Figure 2C was without effect because the ferrocenium cation is too weakly oxidizing to accept electrons from Fe(CN)64-. However, when the reactant in the aqueous phase was changed to Fe(CN)63-, which is capable of oxidizing Fc to Fc+, electron transfer was observed. Shown in Figure 3A is the voltammogram for an electrode covered with a thin layer of nitrobenzene and placed in a solution containing 0.70 mM Fe(CN)63-. The nitrobenzene layer prevents the reduction of the Fe(CN)63- directly at the electrode surface. In Figure 3B is shown the response obtained when 1.3 mM ferrocenium cation was present in the thin layer of nitrobenzene with only 2 M HClO4 in the aqueous phase. Addition of Fe(CN)63- to the aqueous phase produced the response shown in Figure 3C. The enhanced cathodic current results from the cross-phase transfer of electrons from the Fc generated in the nitrobenzene to the Fe(CN)63- in the aqueous phase. The magnitude of such current enhancements can be used to determine the rates of electron transfer across the phase boundary with 3116 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
suitable combinations of reactants in the two phases. The results of a series of experiments of this type are described in a separate study.8 The voltammetric response in Figure 2B is persistent. No evidence of the presence of Fe(CN)64- in the aqueous phase appeared after 30 min of continuous cycling of the potential. Thus, the thin nitrobenzene film is not removed, e.g., by dissolution of nitrobenzene in the aqueous solution (the solubility of nitrobenzene in 2 M HClO4 was estimated spectrophotometrically to be ∼3 mM). The response from the Fc+/Fc (Fc ) ferrocene) couple in Figure 2C diminishes with continuous cycling, probably because the Fc+ cations partition into the aqueous phase. No response from the Fe(CN)64- anions in the aqueous phase developed as the response from the Fc+/Fc couple decreased, which showed that the decrease did not result from the loss of the nitrobenzene layer from the electrode surface. Estimation of the Thickness of the Film of Organic Solvent. Voltammetric responses such as that in Figure 2C, which are produced by reactants that are dissolved in the thin layer of organic solvent on the electrode surface, do not depend on the thickness of the layer. At potential scan rates low enough to ensure that all of the reactant contained in the thin layer is oxidized or reduced during the scan, the peak current is given by the equation for thin layer cyclic voltammetry12
ip ) n2F2νΓ/4RT
(1)
where ν is the scan rate, Γ is the total quantity of reactant in the thin layer, and the other symbols have their usual significance. Thus, any changes in the thickness of the thin film of organic solvent would not be discernible from the thin-layer voltammetric response. However, if the scan rate is increased to values high enough so that the thickness of the diffusion layer at the electrode surface is considerably smaller than the film thickness, the peak current becomes diffusion-limited and its magnitude is given by an equation that contains the concentration of the reactant.13 Under these conditions, the response does depend on the thickness of the film because the reactant concentration increases as the film thickness decreases due to evaporation or dissolution. The ratio of the peak currents measured under thin-layer and semiinfinite diffusion-controlled conditions can therefore be used to obtain estimates of the thickness of the film. As long as the reactant redox couple adheres to the Nernst equation at the highest scan rate employed, the film thickness, δ, can be calculated from eq 2, where D is the diffusion coefficient of the
δ ) 0.286D1/2Riνt-1νd1/2
(2)
reactant in the organic solvent and Ri is the ratio of the peak currents obtained in the thin-layer and diffusion-limited responses recorded at scan rates of νt and νd, respectively. Equation 2 was used to evaluate the thickness of a film of nitrobenzene in which ferrocene had been dissolved. The peak current ratio was Ri ) 0.036 at scan rates of 5 and 500 mV s-1 (12) Bard, A. J.; Faulkner, L. R. Electrochemical Methods; John Wiley and Sons: New York, 1980; p 410. (13) Reference 11, p 218.
Figure 4. (A) Cyclic voltammetry of 6 × 10-9 mol of CoTPP adsorbed on the EPG electrode. Supporting electrolyte, 2 M HClO4. Scan rate, 5 mV s-1. Current scale, S ) 2 µA. The dashed curve was recorded with the bare EPG electrode. (B) Voltammograms obtained after a thin layer of nitrobenzene (s) or benzonitrile (- - -) was applied to the surface of the electrode from (A) and the scan repeated. Current scale, S ) 5 µA. (C) Quantities of CoTPP dissolved in thin layers of nitrobenzene on the surface of EPG electrodes vs the quantities deposited on the surface from aliquots of a chloroform solution of CoTPP. The areas under voltammetric peaks such as the solid curve in (B) were used to calculate the values of ΓCoTPP.
and D was measured as 5.9 × 10-6 cm2 s-1 in a separate experiment using a bare electrode in a solution of ferrocene in nitrobenzene. The value of δ calculated from eq 2 was 3.5 × 10-3 cm compared with the value, 3.2 × 10-3 cm, calculated from the volume of nitrobenzene that was applied to the surface of the electrode. The reasonable agreement between these two thicknesses showed that very little of the nitrobenzene in the thin layer was lost by evaporation or dissolution into the aqueous electrolyte. Comparable agreement was obtained when the experiment was repeated using benzonitrile to form the thin layer of organic solvent. Use of Thin Layers of Organic Solvents To Examine Adsorbed Cobalt Porphyrins. Shown in Figure 4A is the cyclic voltammetric response obtained when 6 × 10-9 mol of CoTPP dissolved in 2 µL of CHCl3 was transferred to an EPG electrode and the solvent allowed to evaporate. The broad, poorly defined shape of the response makes it difficult to obtain a reliable estimate of the quantity of CoTPP present on the surface. However, when a thin layer of nitrobenzene or benzonitrile was applied to the
electrode’s surface, the cyclic voltammograms in Figure 4B were obtained. The initially adsorbed CoTPP dissolved in the thin layer of the organic solvent where it exhibited a clear, reversible cyclic voltammogram. The quantity of CoTPP dissolved in the thin layer of nitrobenzene as determined from the area under the cathodic peak was 5.8 × 10-9 mol, quite close to the quantity of CoTPP initially transferred to the electrode surface. Thus, very little of the CoTPP initially deposited on the electrode was lost from the surface during the operations involved in coating the electrode with a thin layer of nitrobenzene. Shown in Figure 4C are the results of a set of similar experiments in which increasing quantities of CoTPP (dissolved in CHCl3) were transferred to the electrode surface and the solvent was allowed to evaporate before a thin layer of nitrobenzene was applied to the electrode surface. The areas encompassed by the cathodic peaks of subsequently recorded cyclic voltammograms were used to evaluate the quantities of CoTPP dissolved in the thin layer of nitrobenzene. These values are plotted against the quantities of CoTPP initially deposited on the electrode. The data points all lie close to the line of unit slope, which confirms the reliability of this thin-layer method for determining the quantity of CoTPP present on the electrode surface. Similar results were obtained with several other cobalt porphyrins. Evaluation of the Formal Potential of Adsorbed CoTPP. The poor definition of cyclic voltammetric responses such as the one in Figure 4A makes the evaluation of the formal potential for the Co(III/II) couple of the adsorbed CoTPP dubious. However, by using a thin layer of benzonitrile on the electrode surface, a value for the desired formal potential could be estimated. (Benzonitrile was used instead of nitrobenzene because it appeared to be freer of traces of oxidants.) The electrode on which the CoTPP was adsorbed was held at a potential believed to be in the vicinity of the formal potential of the Co(III/II) couple for several minutes in deaerated 2 M HClO4. The electrode was then disconnected from the potentiostat and its equilibrium potential at open circuit, oc EH , was measured. A thin layer of benzonitrile was then 2O applied to the electrode surface in the absence of O2 and the electrode was placed again in the aqueous solution where its open circuit potential, Eoc org, was measured. The formal potential of the Co(III/II) couple of CoTPP in the benzonitrile thin layer, Eforg, was then determined by recording a cyclic voltammogram similar to the dashed curve in Figure 4B. The formal potential of the Co(III/II) couple for adsorbed CoTPP in the aqueous phase, f EH , was then calculated from eq 3. Repetition of this proce2O f oc EH ) Eforg - (Eoc org - EH2O) 2O
(3)
dure at several initial potentials led to values of the formal potential that were reproducible to (20 mV. For CoTPP, the value obtained f for EH was 0.44 V vs SCE, which agrees with some, but not all, 2O previously estimated values.1,5-7 The possibility, suggested by a reviewer, that the shape of the response in Figure 4A might be an indication that a portion of the adsorbed CoTPP was not electroactive was ruled out by the following experiment: The initial potential of a CoTPP-coated electrode was adjusted to 0.6 V (where all electroactive CoIITPP would be oxidized to CoIIITPP). The subsequently measured value of Eoc org with a benzonitrile thin layer was typically ∼0.5 V. Essentially no CoIITPP can be Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3117
on electrode surfaces can be used to extract suitable reactants from aqueous solutions into the organic phase. When the extractable reactant is present at low concentrations, a significant improvement in the sensitivity of electroanalytical determinations can result. For example, shown in Figure 5A is a cyclic voltammogram at a bare EPG electrode in an aqueous solution saturated with ferrocene. The solubility of ferrocene is so small that a very weak voltammetric response is obtained. However, when a thin layer of nitrobenzene was applied to the electrode the cyclic voltammetric response increased continuously as the electrode potential was scanned over the ferrocenium/ferrocene couple because the ferrocene was extracted into the thin layer of nitrobenzene (Figure 5B). Similar behavior was obtained with a solution of the slightly soluble Co(salen) complex. The response obtained at a bare electrode (Figure 5C) showed only a small response from the 67 µM solution of Co(salen), but the response was significantly enhanced after a thin layer of benzonitrile was introduced on the electrode surface (Figure 5D). The potential applications of this approach in the detection and determination of low concentrations of molecules that can be extracted from aqueous into organic phases is apparent.
Figure 5. (A) Cyclic voltammogram recorded at a bare EPG electrode in 2 M HClO4 saturated with ferrocene. (B) Continuously recorded cyclic voltammograms in the same solution after a thin layer of nitrobenzene was applied to the electrode surface. (C) Voltammogram at a bare EPG electrode in 0.3 M LiClO4 containing 67 µM Co(salen). (D) Continuously recorded cyclic voltammograms in the same solution after a thin layer of benzonitrile was applied to the electrode surface. Scan rate, 5 mV s-1 throughout.
present in the benzonitrile thin layer at this potential (dashed curve in Figure 4B) so that all of the CoIITPP initially present on the electrode surface must have been electro-oxidized at 0.6 V. Use of Thin Layers of Organic Solvent To Enhance Electroanalytical Sensitivity. Thin layers of organic solvents (14) (15) (16) (17)
Samec, Z.; Marecek, V. J. Electroanal. Chem. 1986, 200, 17. Wei, C.; Bard, A. J.; Mirkin, M. V. J. Phys. Chem. 1995, 99, 16033. Tsionsky, M.; Bard, A. J.; Mirkin, M. V. J. Phys. Chem. 1996, 100, 17881. Tsionsky, M.; Bard, A. J.; Mirkin, M. V. J. Am. Chem. Soc. 1997, 119, 10785. (18) Delville, M.-H.; Tsionsky, M.; Bard, A. J. Langmuir 1998, 14, 2274.
3118 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
CONCLUSIONS The primary result described in this report is the remarkable ease with which thin layers of suitable organic solvents can be applied to, and are retained by, graphite electrode surfaces. The resulting thin layers of organic liquids interposed between the surface of the electrode and aqueous supporting electrolytes were utilized in this study to examine the electrochemistry of molecules that were introduced into the thin organic phase by dissolution from the electrode surface where they were adsorbed. However, the same thin layers of organic solvents can also be exploited more generally in studies of charge transfer and chemical reactions at the liquid/liquid interface.8 There has been considerable recent interest in electrochemical studies at liquid/liquid interfaces,14-18 and the simple technique described in this study for creating such interfaces may prove to have considerable utility in such studies. ACKNOWLEDGMENT This work was supported by the U.S. National Science Foundation. Received for review April 22, 1998. Accepted June 10, 1998. AC980426K
