Anal. Chem. 1997, 69, 4819-4827
Spectroelectrochemical Sensing Based on Multimode Selectivity Simultaneously Achievable in a Single Device. 2. Demonstration of Selectivity in the Presence of Direct Interferences Yining Shi, Carl J. Seliskar,* and William R. Heineman*
Department of Chemistry, University of Cincinnati, P.O. Box 210172, Cincinnati, Ohio 45221-0172
Three modes of selectivity based on charge-selective partitioning, electrolysis potential, and spectral absorption wavelength were demonstrated simultaneously in a new type of spectroelectrochemical sensor. Operation and performance of the three modes of selectivity for detection of analytes in the presence of direct interferences were investigated using binary mixture systems. These binary mixtures consisted of Fe(CN)63- and Ru(bpy)32+ and of Fe(CN)64- and Ru(CN)64- in aqueous solutions. Results on the Fe(CN)63-/Ru(bpy)32+ binary mixture showed that an anion-exchange coating consisting of PDMDAAC-SiO2 [where PDMDAAC is poly(dimethyldiallylammonium chloride)] and a cation-exchange coating consisting of NafionSiO2 can trap and preconcentrate analytes with charge selection. At the same time, such coatings exclude interferences carrying the same type of charge as that of the exchange sites in the sensor coating. Using the Fe(CN)64-/Ru(CN)64- binary mixture, the Fe(CN)64- component can be selectively detected by restricting the modulation potential cycled to a range specific to the redox-active Fe(CN)64- component and simultaneously monitoring the optical response at the overlapping wavelength of 420 nm. It was also shown that, when the wavelength for optical monitoring was chosen as 500 nm, which is specific to the Ru(CN)64- component, interference from the Fe(CN)64- component for spectroelectrochemical detection of Ru(CN)64- was significantly suppressed, even though the cyclic modulation potential encompassed the redox range for the Fe(CN)64- component. The issue of selectivity remains central to the operation of any chemical sensor. Indeed, one of the significant problems in chemical sensor development is achieving adequate selectivity for determination of the analyte in a real sample, where interferences often confuse the sensor measurement. Selectivity in presently available chemical sensors is typically achieved in one of several ways. For example, potentiometric ion-selective electrodes operate with a single mode of selectivity: selective interaction of the analyte with a membrane to generate a potential. This approach works extremely well in a few cases, such as the pH electrode, for which highly selective membrane reactions are available. Electrochemical sensors of the voltammetric variety and optical sensors (opt(r)odes) typically have two modes of selectivity. The first is potential selection for the voltammetric sensor and S0003-2700(97)00520-9 CCC: $14.00
© 1997 American Chemical Society
wavelength selection for the optical sensor. However, this first level of specificity, supplied by choosing a potential or wavelength specific to the analyte, is rarely adequate to meet all the challenges which these sensors experience in real samples, where interferences of similar electrochemical or optical properties might be present. The second mode of selectivity is provided by modification of the electrode1,2 or optical surface3,4 with a selective coating or membrane which partially screens the analyte out from a mixture containing interferences. However, the second level of selectivity offered by such coatings or membranes is, itself, limited in many real sample applications, since it is usually based on some general screening effects such as size exclusion, hydrophobic barriers, and charge exclusion, rather than analyte-specific discrimination. In this context, the addition of a third level of selectivity to the presently available sensors can be an effective approach for extending the applicability of chemical sensors in the real world. Some biosensors achieve three levels of selectivity by including a highly selective biological reaction, such as enzyme catalysis or antibody-antigen binding.5 For example, the ubiquitous glucose sensor achieves adequate selectivity for the measurement of blood glucose by means of a selective enzyme-catalyzed reaction, potential control of the detection electrode, and sizeselective screening by a membrane.6 Although the selectivity of biomaterials can be excellent, in practice biosensor devices often lack ruggedness due to sensitivity of the biocomponent to heat and other environmental factors. In the first part of this research,7 we have demonstrated a new type of spectroelectrochemical sensor that can be used to quantify an analyte, e.g., Fe(CN)64-, down to at least 8 × 10-6 M. The functioning of the sensor, illustrated in Scheme 1, consists of measuring the optical attenuation in an optically transparent electrode (OTE) under attenuated total reflection (ATR) as an analyte partitions into a selective coating and undergoes an oxidation-reduction reaction. Sensing is based on the change in the attenuation of light passing through the OTE that ac(1) Janata, J.; Josowicz, M.; DeVaney, D. M. Anal. Chem. 1994, 66, 207R. (2) Molecular Design of Electrode Surfaces; Murray, R. W., Ed.; Wiley: New York, 1992. (3) Arnold, M. A. Anal. Chem. 1992, 64, 1015A. (4) Graber, N.; Ludi, H.; Widmer, H. M. Sens. and Actuators 1990, B1, 239. (5) Biosensors: Fundamentals and Applications; Turner, A. P. F., Karube, I., Wilson, G. S., Eds.; Oxford University Press: New York, 1987. (6) Strobel, H. A.; Heineman, W. R. Chemical Instrumentation: A Systematic Approach; Wiley: New York, 1989; pp 1097-1098. (7) Shi, Y.; Slaterbeck, A. F.; Seliskar, C. J.; Heineman, W. R. Anal. Chem. 1997, 69, 3679.
Analytical Chemistry, Vol. 69, No. 23, December 1, 1997 4819
Scheme 1
companies an electrochemical reaction of the analyte at the electrode surface. Thus, for an analyte to be detected, it must partition into the selective coating and be electrolyzed at the potential applied to the electrode, and either the analyte or its electrolysis product must absorb light at the wavelength chosen. Selectivity for the analyte relative to other solution components is obtained by choice of coating material, electrolysis potential, and wavelength for optical monitoring. In this paper, we demonstrate the role of the three modes of selectivity, i.e., electrochemistry, spectroscopy, and selective partitioning, in the detection of analyte in the presence of direct optical and electrochemical interferences. EXPERIMENTAL SECTION Materials. The following chemicals were used: tetraethoxysilane (TEOS, Aldrich), poly(dimethyldiallylammonium chloride) (PDMDAAC, 20 wt % aqueous solution, Polysciences), potassium ferricyanide and potassium ferrocyanide (Aldrich), potassium nitrate (Fisher Scientific), Nafion perfluorinated ionexchange material (5 wt % solution in a mixture of lower aliphatic alcohols and 10% water, Aldrich), tris(2,2′-bipyridyl)ruthenium(II) chloride hexahydrate (Aldrich), and potassium hexacyanoruthenate(II) hydrate (Aldrich). All reagents were used without further purification. Reagent solutions were all made by dissolving the appropriate amounts of chemicals into 0.1 M potassium nitrate solution (prepared with deionized water from a Barnstead water purification system). ITO tin float glass (11-50 Ω/sq, 150 nm thick indium tin oxide layer over tin float glass) was purchased from Thin Film Devices. This ITO tin float glass was cut into 1 in. × 3 in. slides, scrubbed with Alconox, and rinsed thoroughly with deionized water and then 1-propanol prior to use. Preparation of Silica Sol-Gel Stock Solutions. The silica sol-gel stock solutions for the PDMDDAC-SiO2 coating solutions were prepared according to a previously reported protocol.8 For the Nafion-SiO2 coating solutions, a different recipe for the preparation of the silica sol-gel stock solutions was used: 2.0 mL of TEOS, 4.0 mL of deionized water, and 0.1 mL of 0.1 M HCl were combined in a sealed 30 mL vial and stirred at room temperature for 24 h. The resulting transparent sol-gel solution was then used as the stock solution for blending with the Nafion solutions. In this recipe, a lower Si:H2O ratio was employed in order to compensate for the pH effect of the Nafion component (a superacid) on the pore structures of the resulting sol-gel thin films. Formation of Polyelectrolyte-SiO2 Composite Thin Films on ITO Glass Substrates. The preparation of PDMDAAC-SiO2 and of Nafion-SiO2 composite thin films on the ITO glass substrates were conducted as previously described.8 While the (8) Shi, Y.; Seliskar, C. J. Chem. Mater. 1997, 9, 821.
4820
Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
PDMDAAC-SiO2 films on ITO glass were baked overnight in an oven at ∼50 °C, the Nafion-SiO2 films were dried at ambient conditions overnight. Thicknesses of the PDMDAAC-SiO2 films were 0.75 µm, and those of the Nafion-SiO2 films were 0.34 µm, as determined with a Hewlett-Parkard 8453 diode array spectrophotometer using an interference fringe method.8 Instrumentation. The instrumental setup for performing ATR spectroelectrochemistry is standard and consists of the following. The light source was a xenon arc lamp (ILC Technology, Model 302UV). Light from the source was focused into a 1 mm optical fiber that was butt-coupled to a Spex Minimate or Bausch and Lomb 0.25 m monochromator. Light of the monochromatorselected wavelength was coupled to a 600 µm silica step-index optical fiber (Fiberguide Superguide G, NA ) 0.22) with a microscope objective positioned directly against the exit slit of the monochromator. The light from this fiber was coupled into the ITO glass slide of the spectroelectrochemical cell with a Schott SF6 coupling prism (Karl Lambrecht, Chicago, IL). A highviscosity refractive index standard fluid (Cargille, n ) 1.517) was used to span the prism/ITO glass gap. The angle of the incident light into the prism and the alignment of the cell were adjusted to maximize the ATR throughput, as determined by measuring the intensity of the decoupled light. Light propagating through the ITO glass slide was decoupled by using another SF6 coupling prism, and the emergent beam was focused into a silica fiber bundle (C Technology), which transported the light to a Spex 1870 0.5 m monochromator fitted with a phototube (Hamamatsu R955). The current from the phototube was detected by a gated integrator (Stanford Research Systems SR250) run continuously, digitized, and sent to a PC (IBM 80286). PC-logged data files were imported into commercial spreadsheet and graphics algorithms for data manipulation and plotting. Spectroelectrochemical Cells. ATR spectroelectrochemical measurements were performed on an assembled plastic cell (1.0 cm × 4.5 cm × 2.5 cm). Electrical contacts with the ITO glass slide were made with four copper clips attached to each exposed corner of the ITO glass. The reference electrode for the spectroelectrochemical cell was BAS Ag/AgCl (3 M NaCl), and platinum wires were used as the auxiliary electrodes. Procedures for Spectroelectrochemistry. Aqueous 0.1 M KNO3 solution was employed as the supporting electrolyte in all experiments. All films coated on ITO glass were soaked in deionized water overnight prior to any experiment to ensure sufficient equilibration with water. These water-equilibrated films were then stored in a capped, moisturized bottle containing a small amount of water for immediate or later use. For sensing measurements of mixture solutions containing Ru(bpy)32+ and Fe(CN)63-, the ATR cell assembled with a filmcoated ITO glass slide was filled with ∼10 mL of pure 0.1 M KNO3 and cycled for one or two times to ensure good electrical connections and electrochemical communication between the ITO glass slide and the electrolyte solution. This blank solution stayed in the cell for about 3 min and then was replaced with the mixture solution containing equal molar Ru(bpy)32+ and Fe(CN)63- of about the same volume under opened-circuit condition. The resulting optical attenuation signal was recorded at a preset wavelength for about 14 min. The sensing system was then switched to closed-circuit condition and subjected to cyclic potential modulation within a preset potential range, with optical recording continuing.
Sensing measurements of mixture solutions containing Fe(CN)64- and Ru(CN)64- were performed by the same procedure as that for mixture solutions containing Ru(bpy)32+ and Fe(CN)63-, except that only the optical signals induced by cyclic potential modulation were recorded. Sensing measurements based on potential and wavelength selections were carried out on the same Fe(CN)64- and Ru(CN)64--containing spectroelectrochemical system. RESULTS AND DISCUSSION The major goal of this research was to demonstrate the multimode selectivity achievable in the new sensor. This demonstration consisted of applications of a specific sensor fabricated with a sol-gel ion-exchange film on an ITO transparent electrode which could be operated in an ATR mode. Example applications were chosen to illustrate the independent modes of selectivity provided (i) by an ion-exchange coating, (ii) by the choice of the electrolysis potential, and (iii) by the selection of the wavelength at which the sensor was operated. These three modes of selectivity were combined synergistically to achieve selective detection of an analyte in the presence of various direct interferences. The interactions of the three modes of selectivity are dependent on ionic, optical, and electrochemical properties of the components present in the solution mixtures to which the sensors are exposed. In this regard, there are three general types of test sample mixtures which one might envision against which one of the three modes of the sensor’s selectivity could be operative, i.e., capable of detecting analyte in the presence of direct interferences. The first type of sample mixture is the one against which selectivity is based on the sol-gel ion-exchange coating. In this type of sample, the analyte has a formal charge different from that of the interference, but the analyte has an optical absorption wavelength and an electrolysis potential which coincide with those of the interference. The second type of sample mixture is the one in which the electrolysis potential of the sensor selectively discriminates against the interference. In this type of sample, the analyte has a formal charge of the same sign and the same optical absorption wavelength as the interference. However, the analyte has a well-separated electrolysis potential. The third type of sample mixture is the one against which the sensor’s wavelengthbased selectivity is effective. In this type of sample, the analyte has the same type of formal charge and an overlapping electrolysis potential as the interference but the analyte has a well-separated optical absorption wavelength. To demonstrate the capability of the sensor for selective sensing in each of these three types of sample mixtures, we have chosen the metal complexes Fe(CN)63-/4-, Ru(bpy)33+/2+, and Ru(CN)63-/4- as the model solution components. Table 1 and Figure 1 show some of the ionic, electrochemical, and optical properties of these metal complexes which specifically relate to the results presented in this paper. The binary solute system consisting of Fe(CN)63- and Ru(bpy)32+ was taken as the first type of sample mixture in which the components carry a different type (9) Standard Potentials in Aqueous Solution; Bard, A. J., Parsons, R., Jordan, J., Eds.; Dekker: New York, 1985. (10) Waltz, W. L.; Akhtar, S. S.; Eager, R. L. Can. J. Chem. 1973, 51, 2525. (11) Durham, B.; Walsh, J. L.; Carter, C. L.; Meyer, T. J. Inorg. Chem. 1980, 19, 860. (12) Fabian, R. H.; Klassen, D. M.; Sonntag, R. W. Inorg. Chem. 1980, 19, 1977. (13) DeFord, D. D.; Davidson, A. W. Anal. Chem. 1951, 73, 1469. (14) Hicks, K. W.; Chappelle, G. A. Inorg. Chem. 1980, 19, 1623.
Table 1. Electrochemical and Optical Properties of Metal Complexes formal potential metal species Fe(CN)63-/4Ru(bpy)33+/2+ Ru(CN)63-/4-
bare ITO PDMDAACλmax standard (V vs Ag/ SiO2/ITO mol (V vs SHE) AgCl) (V vs Ag/AgCl) (nm) (M-1 cm-1) +0.361a +1.53c +0.86e
+0.22 +1.18 +0.75
+0.11 +1.07 +0.64
420b 454d 462f
1040b 14 600d 1005f
a Reference 9. b Reference 10. 3- c mol refers to Fe(CN)6 . Reference 11. d Reference 12. mol refers to Ru(bpy)32+. e Reference 13. f Reference 14. mol refers to Ru(CN)63-.
of formal charge and show strongly overlapping absorption spectra. Choice of the electrolysis potential range can be made to encompass both redox couples, thus not discriminating between these complexes on this basis. The binary system consisting of Fe(CN)64- and Ru(CN)64- was chosen as the second type of sample mixture. In this system, Fe(CN)64- functioned as the analyte and Ru(CN)64- as a direct optical interference, since both have the same type of formal charge and their oxidized forms, Fe(CN)63- and Ru(CN)63-, have overlapping optical absorptions at 420 nm. However, these metal complexes have well-separated redox potentials. Interestingly, the Fe(CN)64- and Ru(CN)64binary system can also be used as the third type of sample mixture. In this case, Ru(CN)64- functioned as the analyte, leaving Fe(CN)64- as the electrochemical interference of the same formal charge. The electrolysis potential range can again be chosen to cover both redox couples, leaving neither sensor selectivity based on potential nor ion-exchange modes effective. However, Fe(CN)63has a negligible molar absorptivity at 500 nm, while Ru(CN)63has a molar absorptivity at this wavelength which is about 40% of its maximum value (see Figure 1b2) at its long-wavelength maximum, which is sufficient to provide optical selectivity. Selectivity Provided by the Sol-Gel Ion-Exchange Coating. According to the sensor concept described earlier,7 selectivity for the detection of Fe(CN)63- in the presence of Ru(bpy)32+ can be provided by an anion-selective coating such as a sol-gel PDMDAAC-SiO2 film.8 This sol-gel ion-exchange film is capable of concentrating Fe(CN)63- by anion exchange and excluding cationic Ru(bpy)32+. Exclusion of Ru(bpy)32+ from the PDMDAAC-SiO2 film prevents both access of this solution component to the electrode for electrochemical reduction/oxidation and its interaction with the evanescent field that exists within the sensing film thickness. Figure 2 illustrates the selective spectroelectrochemical sensing of Fe(CN)63- in the presence of Ru(bpy)32+ (both 5.0 × 10-4 M). As shown in Figure 2a, when the solution mixture was added to the PDMDDAC-SiO2 film-coated sensor, a sharp decrease in transmittance at 450 nm was detected. This optical response, however, cannot be simply taken as the sensing signal for either Fe(CN)63- or Ru(bpy)32+ without insertion of an additional level of selectivity, since both species absorb at 450 nm (see Figure 1b1). However, when this system was subjected to electrolysis potential cycled between +1.25 and -0.30 V, which encompasses both redox couples, the optical response, as shown in Figure 2b, was modulated only within the Fe(CN)63- redox window. Figure 2c shows the corresponding cyclic voltammogram, which clearly demonstrates only the redox peaks for the Fe(CN)63-/4- couple. Furthermore, when the cycled potential was restricted to the range between +0.4 and -0.30 V, which covers Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
4821
Figure 1. Cyclic voltammograms (panel a) and absorbance spectra (panel b) of analytes. The upper left figure (a1) shows the voltammogram for an equimolar mixture of 5.0 × 10-4 M Fe(CN)63- and Ru(bipy)32+ at a bare ITO glass electrode. The upper right figure (a2) shows the voltammogram for an equimolar mixture of 5.0 × 10-4 M Fe(CN)64- and Ru(CN)64- at a bare ITO glass electrode. Cyclic voltammetry parameters were: ν ) 10 mV/s, E/V vs Ag/AgCl. The lower left figure (b1) shows the spectra of Ru(bipy)32+ (1.0 × 10-4 M) and Fe(CN)63- (1.0 × 10-4 M) in aqueous solution. The encircled numerals indicate measurement wavelengths 420 (1) and 450 nm (2). The lower right figure (b2) shows the spectra of Fe(CN)63- (1.0 × 10-3 M) as in b1 and Ru(CN)63- (1.0 × 10-3 M) in aqueous solution. The spectrum for Ru(CN)63- has been scaled with respect to that for Fe(CN)63- by the ratio of the literature values (Table 1) of the molar extinction coefficients. The encircled numerals correspond to sensor measurement wavelengths 420 (1) and 500 nm (2).
only the Fe(CN)63-/4- couple, an electrochemically modulated optical response similar to the one shown in Figure 2b was observed; i.e., the transmittance at 450 nm changed with the same amplitude as that shown in Figure 2b but with a shorter time period, as expected for scanning at a fixed scan rate over a smaller potential range. In contrast, when the potential was cycled between +1.25 and +0.80 V, which corresponds to the Ru(bpy)32+ redox couple, no modulation of the transmittance at 450 nm was observed. These results clearly demonstrate that the optical response shown in Figure 2a was exclusively due to the partitioning of Fe(CN)63- into and to the exclusion of Ru(bpy)32+ by the sol-gel PDMDAAC-SiO2 film of the sensor. As a result, Ru(bpy)32+ was not detectable by the optical measurement, even though this interference has a molar absorptivity at 450 nm, more than 13 times greater than that of the Fe(CN)63- analyte. The sensor selectivity for the components in this binary mixture can be totally reversed by simply changing the sol-gel ion-exchange film coating from an anion exchanger to a cation exchanger. Thus, the device becomes a sensor for Ru(bpy)32+ with selectivity against Fe(CN)63- when a cation-selective solgel ion-exchange film consisting of Nafion-SiO2 is spin-coated onto the sensor. As shown in Figure 3a,b, the sensor responded optically at 420 nm to an equal molar mixture of Fe(CN)63- and Ru(bpy)32+. This response was electrochemically modulated only 4822 Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
within the potential window specific to the Ru(bpy)33+/2+ couple. This result was confirmed by the cyclic voltammogram shown in Figure 3c, which also showed only the Ru(bpy)33+/2+ couple, despite the fact that the potential range cycled encompassed a range covering both couples. Separate experiments with the electrolysis potentials between +1.25 and +0.80 V and between +0.40 and -0.30 V showed that the change in transmittance at 420 nm was modulated only in the former range, that is, only within the redox window for the Ru(bpy)33+/2+ couple. Thus, the optical response at 420 nm can be uniquely ascribed to cationic Ru(bpy)32+ that has partitioned into the Nafion-SiO2 coating without interference from Fe(CN)63-. The above experiments show that sol-gel-based chargeselective coatings can impart selectivity to a sensor by electrostatic exclusion of interferences carrying the same type of formal charge as that of the exchange sites within the coating. Thus, charged analytes should be selectively detectable in the presence of oppositely charged interferences, even though they might possess the same redox potential and the same optical absorption wavelength. Selectivity Provided by Potential Selection. The second example of selectivity involves a sensor with a sol-gel ionexchange coating which provides no selectivity against an interference and a sensor which does not discriminate on the basis of wavelength. Rather, selectivity is achieved by means of electroly-
Figure 2. Percent transmittance changes (a and b) at 450 nm and the associated cyclic voltammogram (c) of a PDMDAAC-SiO2-coated ITO sensor exposed to an equimolar mixture of 5.0 × 10-4 M Ru(bipy)32+ and Fe(CN)63-. Panel a corresponds to uptake of Fe(CN)63- by the anion-selective film. Panel b corresponds to the transmittance changes at 450 nm induced by cyclic voltammetric scans between the limits +1.25 T -0.30 V, as indicated in the voltammogram in panel c. Panel b was recorded chronologically after panel a. Cyclic voltammetry parameters were ν ) 10 mV/s, E/V vs Ag/AgCl.
Figure 3. Percent transmittance changes (a and b) at 420 nm and the associated cyclic voltammogram (c) of a Nafion-SiO2-coated ITO sensor exposed to an equimolar mixture of 5.0 × 10-4 M Ru(bipy)32+ and Fe(CN)63-. Panel a corresponds to uptake of Ru(bipy)32+ by the cation-selective film. Panel b corresponds to the transmittance changes at 420 nm induced by cyclic voltammetric scans between the limits +1.25 T -0.30 V, as indicated in the voltammogram in panel c. Panel b was recorded chronologically after panel a. Cyclic voltammetry parameters were ν ) 10 mV/s, E/V vs Ag/AgCl.
sis potential control. The analytes Fe(CN)64- and Ru(CN)64constitute a test system in which both species partition into a solgel PDMDAAC-SiO2 film on the sensor and both also strongly absorb at 420 nm (Figure 1b2). However, these two solution components constitute redox couples which are well separated (Figure 1a2), and selectivity can be achieved by electrochemically modulating the optical attenuation within the sensor over a redox potential range specific to either couple. Figure 4 shows the uptake kinetics of a sensor exposed to these two solution species
(both 5.0 × 10-5 M) as they copartition into the PDMDAAC-SiO2 sensor coating without significant discrimination in terms of peak current growth. Since both of these solution components show a strong absorption at 420 nm, the sensor optical response at this wavelength, induced by potential cycling between +0.90 and -0.10 V, which covers the redox windows of both Fe(CN)64- and Ru(CN)64-, should reveal contributions from both species. Indeed, they have copartitioned into the PDMDAAC-SiO2 coating without Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
4823
Figure 4. Cyclic voltammograms indicating the copartitioning of Fe(CN)64- and Ru(CN)64- from an equimolar solution (5.0 × 10-5 M) into a PDMDAAC-SiO2 coating on a sensor. The cyclic voltammetry parameters were ν ) 10 mV/s, E/V vs Ag/AgCl; scan limits, +0.90 T -0.10 V.
discrimination. As the sensor was cycled over this potential range, the transmittance at 420 nm was modulated, producing a complex waveform. Figure 5a shows the repeating pattern of this waveform, and Figure 5b shows the corresponding cyclic voltammogram when the electrolysis potential was repetitively cycled over both Fe(CN)64- and Ru(CN)64- couples beyond the 15th cycle of the system described in Figure 4. In Figure 5a, the maximum transmittance at the beginning of the data record (arrow 1) corresponded to the potential value of -0.10 V, where both species were in their reduced (colorless) forms. This same point is illustrated by a similar arrow in the cyclic voltammogram in Figure 5b. As the cycled potential proceeded in the positive direction, the partitioned Fe(II) complex began to be electrooxidized to the Fe(III) complex, which, in turn, produced a decrease in the transmittance at 420 nm. After the potential was scanned over the anodic wave for the Fe(II)/(III) complex couple, most of the accessible Fe(II) complex was electrooxidized to the Fe(III) complex (indicated by arrows 2). This portion of the curve is followed by a transmittance change which is a positive-going “ramp” with a slope much smaller than the one which preceded (indicated by the section from arrow 2 to arrow 3). This transmittance change continued until the cycled potential approached a value where electrooxidization of the partitioned Ru(II) complex began to occur. This point was indicated by the occurrence of a steeper change in the transmittance curve with an origin at arrow 3. When most of the accessible Ru(II) complex was electrooxidized, that is, as the potential scanned just over the value of +0.90 V (arrow 4), the transmittance curve leveled off (indicated by the curve section from arrows 4 to 5). This plateau in the transmittance is due to the fact that almost all the accessible Fe(II) complex and Ru(II) complex species have been electrooxidized to the Fe(III) and Ru(III) complex forms, respectively. When the cycled potential proceeded negatively from +0.90 V, the Ru(III) complex began to be electroreduced (indicated by arrow 5), leading to an increase in the transmittance until the 4824 Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
potential scanned over the cathodic wave for the Ru(II)/(III) complex couple (indicated by arrow 6). Further negative scanning of the potential to -0.10 V increased the transmittance, returning it to its original maximum value. At this point in the waveform, the electrochemical reformation of the Fe(II) complex (indicated by arrow 8) is complete, and the chemical system has been fully cycled. From consideration of Figure 5, it can be seen that the minimum transmittance at 420 nm (corresponding to a potential of +0.90 V) represents two optical attenuation contributions to the waveform: an anodic “shoulder” section (arrows 1-3), which resulted from the electrogenerated Fe(III) species, and a “head” section, which resulted from the electrogenerated Ru(III) species within the film. A detailed consideration of the waveform shape shows that the Ru(III) complex contribution (i.e., the head section) is smaller in magnitude than the Fe(III) complex contribution (i.e., the shoulder section). We attribute this to the large difference in molar absorptivity at 420 nm between the two species (see Figure 1b2 and Table 1). Furthermore, the waveform is unsymmetrical, that is, the contribution of the anodic (left) shoulder is smaller than its cathodic (right) counterpart. A series of experiments aimed at understanding this trait in the waveform has shown that it is due to a kinetically incomplete electrooxidation of the partitioned Fe(II) complex species during the anodic portion of the potential scan (to be published). To detect the analyte, Fe(CN)64-, spectroelectrochemically at the wavelength of 420 nm in the presence of its optical interference, Ru(CN)64-, selective potential modulation should be an effective approach, since Fe(CN)64- has a redox window wellseparated from that of the Ru(CN)64- component. Figure 6 shows the selective spectroelectrochemical sensing of Fe(CN)64- in the presence of Ru(CN)64- (both 5.0 × 10-5 M) by potential cycling within selected redox windows. On the top portion of Figure 6 are shown the linear cyclic potential excitations by which the corresponding spectroelectrochemical modulations were achieved. Correspondence between them is indicated by the vertical dashed lines. Figure 6a shows the repetitive waveform of the optical response of the sensor induced by three consecutive cycles over the potential range +0.90 T -0.10 V. Figure 6b shows selective sensing of the Fe complex component obtained when the cycled potential was restricted to the range between +0.40 and -0.10 V, which just covers the Fe(CN)63-/4- redox couple. Here, the minimum transmittance plateau, corresponding to +0.40 V (indicated by the vertical dashed lines in Figure 6b), can be assigned to the attenuation contribution from the Fe(III) complex component, without significant interference from the Ru(III) complex component, since all the accessible Ru complex is in the reduced Ru(II) complex colorless form. In a similar manner, the attenuation contribution from the Ru complex component can also be isolated electrochemically by restricting the cycled potential to the Ru complex couple, e.g., between +0.90 and +0.40 V. As shown in Figure 6c, when the potential was cycled between +0.90 and +0.40 V, only the Ru complex was subjected to electrochemical modulation. The detailed shape of this waveform shows that it is nearly symmetrical (sinusoidal) in shape. This attenuation corresponds to a transmittance change of about 7%, which can be uniquely attributed to the partitioned Ru complex component. It is noted that this waveform was modulated electrochemically on a constant transmittance background of about 84%, corresponding to a fully oxidized Fe complex species expected after a prolonged
Figure 5. Transmittance changes (a) induced by a single cycle of voltammetric scanning of the sensor with copartitioned Fe(CN)64- and Ru(CN)64- from an equimolar solution (5.0 × 10-5 M) into the sensor’s PDMDAAC-SiO2 coating. (b) The 15th cyclic voltammogram in the uptake experiment of Figure 4. The underlying redox chemistry at various points (indicated by encircled numbers with arrows) in the transmittance and excitation potential curves are discussed in the text. The sensor was operated at 420 nm, and all voltammetry parameters were the same as in Figure 4.
Figure 6. Transmittance changes at 420 nm of a sensor with copartitioned Fe(CN)64- and Ru(CN)64- from equimolar solution (5.0 × 10-5 M) with selected potential scan ranges. (a) Three full cyclic voltammetric scans over the potential range +0.90 T -0.10 V, as in Figure 5b. (b) The transmittance changes over the potential range +0.40 T -0.10 V. (c) The changes over the potential range +0.9 T +0.40 V. Chronologically, panel a was recorded first, followed by panel b and finally panel c.
potential modulation between +0.90 and +0.40 V, a range which is positive with respect to the oxidation potential of Fe(CN)64(see Figure 5b and Table 1). Selectivity Provided by Wavelength Selection. The third example of selectivity provided by the present sensor design pertains to wavelength selection. In this case, the copartitioned interference in the (nonselective) film possesses an overlapping
redox potential but a distinguishable absorption wavelength as compared with that of the analyte of interest. To demonstrate this concept, we again chose the binary mixture containing equal molar Fe(CN)64- and Ru(CN)64- components as the model system. However, we identify the Fe(CN)64- component as the interference, while the Ru(CN)64- component functions as the analyte of interest. Thus, this test system is the same as that described in Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
4825
Figure 7. Transmittance changes at 500 nm of a sensor with copartitioned Fe(CN)64- and Ru(CN)64- from equimolar solution (5.0 × 10-5 M) with selected potential scan ranges. (a) Three full cyclic voltammetric scans over the potential range +0.90 T -0.10 V, as in Figure 6. (b) The transmittance changes over the potential range +0.4 T -0.10 V. (c) The changes over the potential range +0.9 T +0.40 V. Chronologically, panel a was recorded first, followed by panel b and finally panel c.
Figure 6, except that the monitoring wavelength is now set at 500 nm and the potential selectivity mode employed in the previous section was not used. As shown in Figure 1b2, the Fe(CN)63- species has a negligible molar absorptivity at 500 nm compared with Ru(CN)63-, which has a molar absorptivity at 500 nm about 40% of its near-maximum value at 455 nm. Therefore, if the sensor wavelength is selected as 500 nm, one can expect that the attenuation contribution to the modulation waveform from the Fe complex component will be suppressed. On the other hand, the Ru complex component will be detectable optically when the mixture system is subjected to redox potential cycling over a wide potential range covering both redox couples. Figure 7 shows the selective sensing of Ru(CN)64- at 500 nm in the presence of Fe(CN)64-. As shown by comparison of parts a and b of Figure 7 with those of Figure 6, the attenuation corresponding to the Fe complex component was, indeed, suppressed to an almost negligible level, whereas the attenuation corresponding to the Ru complex component was enhanced to a level almost 3 times that shown in Figure 6a. Considering that the molar absorptivity of Ru(CN)64- at 500 nm is less than that at 420 nm, the transmittance change for the “head” section shown in Figure 7a cannot be totally attributed to the modulated Ru complex component. Other possible attenuation contributions may include a competition in the ion-exchange process of the two analyte anions which is changing the partition coefficients of one or both of the analytes into the film, a redox cross-reaction between one constituent of the sample and a different oxidation state being electrogenerated within the film, and changes in the electrochromic properties of the analytes, as well as differences in the optical constants of the ITO-SnO2 layers and/or ITO-SnO2/PDMDAAC-SiO2 interfaces at the two wavelengths. Despite this unusual observation which is under current investigation, this experiment has shown that wavelength can also function as a means for selective analyte sensing, even though the analyte has the same redox potential 4826 Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
as that of the interference. In this case, the cycled potential itself merely serves as a modulation technique for optically selective detection of the analyte. This wavelength selectivity is more pronounced when the cycled potential is restricted to a redox window specific either to the analyte or to the interference, as shown in parts b and c of Figure 7. Compared with both parts b and c of Figure 6, the transmittance waveform for the Fe complex component shown in Figure 7b responded insignificantly to the potential modulation cycled, even in the redox window specific to the Fe complex. In contrast, the transmittance waveform for the Ru complex component shown in Figure 7c exhibited a strong, potential-dependent response to the potential modulation cycled in the redox window specific to the Ru complex component. Again, this response cannot be totally attributed to the modulated Ru complex component as mentioned above. These results provide evidence that the wavelength selectivity provided by the present sensor design can also be employed as a powerful tool for selective sensing of analyte in the presence of an interference that carries the same type of charge and possesses even the same redox potential as that of the analyte, providing the analyte has an absorption wavelength well-separated from that of the interference. The coordination of both selectivity modes due to wavelength and potential selection is also shown. CONCLUSIONS We have demonstrated a new sensor with three modes of selectivity simultaneously achieved. Specifically, we have demonstrated a sensor for selective detection of charged analytes in the presence of charged, optical, and electrochemical interferences. These three modes of selectivity were realized through incorporation of a sol-gel ion-exchange coating, the selection of electrolysis potential, and the selection of optical absorption wavelength provided by the sensor design. The interplay of the three modes of selectivity are dependent upon the specific ionic,
optical, and electrochemical properties of the components present in the solution mixture to which the sensor is exposed. Although the selective spectroelectrochemical sensing in this work was accomplished with the cyclic potential modulation, other forms of modulation excitation (such as step potential modulation) could be feasible and possibly more selective and sensitive. In the specific sensor design discussed in this paper, one of the modes of selectivity is provided by an ion-exchange coating. Thus, one might expect that the effect of ionic strength, as well as other effects, particularly the one resulting from competition for ion-exchange sites of the sensing films between the charged analytes and interferences carrying the same type of charge, could limit the performance and applicability of this type of sensor. In
a subsequent paper, we will present further results dealing with these and other effects, as well as quantitative sensing in the presence of direct interferences with varying component ratios. ACKNOWLEDGMENT Support provided by the Department of Energy (Grant DEFG07-96ER62311) is gratefully acknowledged. Received for review May 20, 1997. Accepted September 15, 1997.X AC970520L X
Abstract published in Advance ACS Abstracts, November 1, 1997.
Analytical Chemistry, Vol. 69, No. 23, December 1, 1997
4827
