Anal. Chem. 2007, 79, 6515-6526
Detailed Analysis of the Electron-Transfer Properties of Azurin Adsorbed on Graphite Electrodes Using dc and Large-Amplitude Fourier Transformed ac Voltammetry Barry D. Fleming,† Jie Zhang,‡ Darrell Elton,§ and Alan M. Bond*,†
School of Chemistry, Monash University, Victoria 3800, Australia, Institute of Bioengineering and Nanotechnology, 31 Biopolis Way, The Nanos, #04-01, Singapore 138669, and Department of Electronic Engineering, La Trobe University, Bundoora, Victoria 3086, Australia
The analysis of dc cyclic voltammograms of surfaceconfined metalloproteins is complicated by large background currents, significant ohmic iRu drop, and frequency dispersion related to protein and electrode surface inhomogeneity. The use of large-amplitude Fourier transform ac voltammetry for the quantification of the electrontransfer properties of a thin film of redox-active protein azurin adsorbed onto edge-plane, basal-plane, and highly oriented pyrolytic graphite electrode surfaces has been evaluated and compared to results obtained by dc cyclic voltammetry. In principle, it has been established that fourth and higher harmonic sine-wave data are ideally suited for analysis of electron-transfer processes as they are almost completely devoid of background capacitance current contributions. However, uncompensated resistance has a higher impact on these components, as is the case with fast scan rate dc techniques, so strategies to include this term in the simulations have been investigated. Application of recommended strategies for the evaluation of the electron-transfer properties of azurin adsorbed onto three forms of graphite, each having different background or uncompensated resistance values, is described and compared to results obtained by traditionally used forms of cyclic voltammetry. The electrontransfer rate constant, k0′, of azurin at a highly oriented pyrolytic graphite electrode surface was ∼250 s-1, compared with g1000 s-1 at edge-plane and basal-plane graphite electrodes. The significantly lower k0′ value found at the highly oriented pyrolytic graphite electrode was related to the relatively low level of edge-plane defect sites present at the surface of this electrode. However, analysis of high ac harmonics suggests that frequency dispersion is substantial at all electrode surfaces. Such effects in these diffusionless situations are significantly enhanced * Corresponding author. E-mail:
[email protected]. Fax: +61 3 9905 4597. † Monash University. ‡ Institute of Bioengineering and Nanotechnology. § La Trobe University. 10.1021/ac070448j CCC: $37.00 Published on Web 08/02/2007
© 2007 American Chemical Society
relative to solution-phase voltammetry, where overlay of diffusion layers minimizes the impact of heterogeneity. The voltammetric study of redox-active proteins adsorbed as a thin film onto an electrode surface has provided significant advances in our understanding of protein electron transfer and coupled chemical reactions.1-5 The removal of diffusion present in conventional solution-phase studies simplifies the theoretical treatment applicable in diffusionless thin-film voltammetry. However, substantial problems still exist to ensure analysis of the faradaic current is devoid of artifacts arising from corrections for the large background present or uncompensated resistance. Possible influences of a heterogeneous electrode surface or heterogeneous layer of adsorbed material, which may give rise to kinetic or thermodynamic dispersion, also need to be considered. A problem with dc voltammetric background correction arises because only a minor portion of the data acquired during a dc voltammetric experiment may be derived from the electrontransfer process, and in practice, dc faradaic-to-background current ratios are often BPG > HOPG over the whole scan rate range. This can be understood in terms of the different densities of surface oxides that can influence the surface charge, with the EPG electrode having significantly more surface oxide groups at the exposed edge planes of the graphite. The slight scan rate
dependence of the apparent capacitance has been attributed to defects in the seal between the graphite and insulating epoxy resin,26 although again this could be associated with electrode heterogeneity. The rate of electron transfer between adsorbed azurin and each of the graphite electrodes may be determined from dc cyclic voltammetric measurements by comparing the separation of the oxidation and reduction peaks (Epox - Epred) as a function of scan rate with that expected theoretically from simulations that were based on Butler-Volmer kinetic theory and other assumptions such as a Langmuir isotherm.27 The values of Γtot and Cdl used in the simulations were estimated directly from the experimental voltammetric data, while the value used for Ru was measured independently (see Experimental Section) and the charge-transfer coefficient R was assumed to be 0.5. Experimental data and simulated data are provided in Figure 5. The value of k0′ estimated via this procedure was in the range of 250 (slow scan rate) to 1000 s-1 (fast scan rate) for azurin on HOPG, and g1000 s-1 on BPG and EPG. It may seem tempting to assign a faster k0′ value to the azurin process confined on these latter electrode surfaces, but as shown in Figure 5, differences in simulated curves with k0′ ) 1000 s-1 and k0′ ∼ ∞ are very small, and so uncertainties in assigning rate constants of g1000 s-1 are concomitantly large. On the basis of ∆Ep data, the process can be considered reversible (23) Jeuken, L. J. C.; Armstrong, F. A. J. Phys. Chem. B 2001, 105, 5271-5282. (24) Fristrup, P.; Grubb, M.; Zhang, J.; Christensen, H. E. M.; Hansen, A. M.; Ulstrup, J. J. Electroanal. Chem. 2001, 511, 128-133. (25) Fujita, K.; Nakamura, N.; Ohno, H.; Leigh, B. S.; Niki, K.; Gray, H. B.; Richards, J. H. J. Am. Chem. Soc. 2004, 126, 13954-13961. (26) Wehmeyer, K. R.; Wightman, R. M. J. Electroanal. Chem. 1985, 196, 417421. (27) Laviron, E. J. Electroanal. Chem. 1979, 101, 19-28.
Figure 5. Scan rate dependence of peak separation (Epox - Epox) for azurin adsorbed onto (a) HOPG, (b) BPG, and (c) EPG electrodes measured in 0.1 M NaCl at pH 4. Experimental data are shown as points (b) while simulations at different values of k0′ are shown as either dashed (- - -) or full lines (s). Simulation parameters: Ei ) 0.650 V and Es ) 0 V, plus (a) Ru ) 650 Ω, Cdl ) 7.5 µF‚cm-2, Γtot ) 11 pmol‚cm-2, E0′ ) 0.307 V, R ) 0.5; (b) Ru ) 170 Ω, Cdl ) 36 µF‚cm-2, Γtot ) 24 pmol‚cm-2, E0′ ) 0.314 V, R ) 0.5; (c) Ru ) 170 Ω, Cdl ) 74 µF‚cm-2, Γtot ) 27 pmol‚cm-2, E0′ ) 0.303 V, R ) 0.5. 6520 Analytical Chemistry, Vol. 79, No. 17, September 1, 2007
Figure 6. Comparison between experimental (black, full line) and simulated (red, dashed line) cyclic voltammograms obtained at a scan rate of 20 V‚s-1 for azurin adsorbed onto (a) HOPG, (b) BPG, and (c) EPG electrodes measured in 0.1 M NaCl at pH 4. Simulation parameters as per Figure 5, except: (a) k0′ ) 250 s-1; (b) k0′ ) 1000 s-1; (c) k0′ ) 1000 s-1.
within the scan rate range examined at BPG and EPG electrodes, which translates to k0′ g 1000 s-1. The slower kinetics at the HOPG electrode can be attributed to the lower density of edgeplane defects and, hence, poor protein-electrode coupling.1 The much slower rate of electron transfer for surface-confined azurin at the HOPG (basal-plane) electrode, relative to the edge-plane case, also is noted for the solution-phase [Fe(CN)6]3-/4- couple in the Electrode Characterization section. In the case of azurin, weaker adsorption at HOPG is indicated via a much lower surface coverage. This could be a result of different orientations and electron-transfer distances on the two forms of graphite, with the situation on HOPG being less favorable for rapid electron transfer. However, electronic and other physicochemical differences of HOPG and edge plane graphite that are believed to be of fundamental significance in the case of the [Fe(CN)6]3-/4process21 also should be significant in surface-confined azurin electrode reactions. This implies the presence of a range of rates (kinetic dispersion) and hence explains why apparent k0′ values appear to be scan rate dependent. Of course kinetic dispersion also could be present in the EPG and BPG values, but as all molecules have k0′ g 1000 s-1, this could not be detected within the scan rate range employed in this study. Noninclusion of Ru in the simulations resulted in apparently significantly lower estimates of k0′ (e.g., 500 s-1 for BPG) and poorer agreement between experiment and theory. Notably, when Ru is included in the simulations, then the largest current
contribution to the iRu drop effect is from the background current. For example, ∼80% of the current associated with the azurinmodified EPG electrode was capacitative (background) in nature. This is a usual observation in surface-confined protein electrochemistry. Typical forms of dc background subtraction help identify peak positions, but do not achieve compensation for the effects of iRu drop on the positions of the peaks. While representing the most common dc cyclic voltammetric method used to calculate k0′ values, the peak separation approach is statistically limited as it relies on the analysis of only two data points. This is a highly inefficient use of a data set containing thousands of data points and also may potentially lead to inaccurate reporting of kinetic values. A more data efficient approach is to compare the complete voltammogram obtained at a given scan rate with that expected theoretically. This allows the peak current magnitude, peak position, and peak shape to be considered in the assessment of the kinetics. Figure 6 shows experimental cyclic voltammograms obtained at a scan rate of 20 V‚s-1 at each form of electrode compared with simulated dc cyclic voltammograms. Reasonable agreement could be obtained, but generally the shapes of the faradaic processes observed experimentally were broader than those predicted by theory. This again is consistent with kinetic or thermodynamic dispersion arising from nonequivalence or interaction of azurin molecules within the adsorbed protein layer. However, best estimates of the k0′ values determined by these simulations (250, g 1000, and g1000 s-1 for HOPG, BPG, Analytical Chemistry, Vol. 79, No. 17, September 1, 2007
6521
Figure 7. Fundamental (first) harmonic voltammograms before (a) and after (b) background subtraction for azurin adsorbed onto HOPG (black), BPG (red), and EPG (blue) electrodes. Experimental conditions: f ) 36 Hz, ∆E ) 80 mV, v ) 0.09686 V‚s-1, Ei ) 0.650 V, and Es ) 0 V.
and EPG, respectively) are consistent with those indicated by the peak separation approach. Large-Amplitude Sine Wave ac Voltammetry. The dc data indicated that the kinetics of electron transfer of azurin exhibit a dependence on the nature of the graphite surface and that some nonideality is present when analysis is undertaken using a theoretical model based on the assumption of a fully homogeneous system with all molecules having the same E0′, k0′ and obeying the Butler-Volmer relationship. To probe the system in much greater detail, large-amplitude FT ac voltammetry was employed. The potential applied in the ac experiments consisted of a single sinusoidal waveform of frequency f, and amplitude, ∆E, which was superimposed onto the linear dc potential. The dc and ac harmonic components generated by application of this waveform were separated using FT-inverse FT techniques as described in the Experimental Section. The frequency and amplitude of the sine wave used was varied from 10 to 191 Hz and 50 to 80 mV, respectively. The scan rate was maintained at ∼0.1 V‚s-1 for all these ac experiments. (1) Fundamental Harmonic. An example of the fundamental harmonic component, displayed as a current-time response, for each of the graphite azurin-modified electrodes obtained by employing a sinusoidal waveform (f ) 36 Hz, ∆E ) 80 mV) is shown in Figure 7. The fundamental harmonic component includes contributions from ac faradaic and ac capacitance currents (analogous to the dc case). As observed with dc voltammetry and by casual inspection of Figure 7a, the relative order of the capacitance (background) current magnitude was EPG > BPG > HOPG, and it exhibited a significant dependence on potential. The capacitance current decreased marginally with increasing applied frequency (not shown), which is an outcome analogous to that found with the dc data as the scan rate was increased. To assist with the evaluation of the faradaic contribution to the first harmonic current for each of the graphite electrodes, a background subtraction procedure was employed to remove the capacitance current contributions. A fourth-order polynomial function was fitted to the nonfaradaic potential regions before and after the potential region where significant faradaic current is present. The background current present at potentials where faradaic current is present was then predicted from the polynomial. Figure 7b shows the resultant first harmonic voltammograms of the reductive process (in simpler envelope form of presentation) 6522 Analytical Chemistry, Vol. 79, No. 17, September 1, 2007
after the background subtraction procedure has been implemented. Notably, the faradaic current density obtained with the HOPG electrode is ∼25% of that found at the BPG and EPG electrodes. Based on the Γtot values calculated from dc data, the faradaic current density of the azurin-HOPG electrode should be ∼40% of that observed at the BPG and EPG electrodes. This difference may be the combined result of protein loss, errors associated with background subtraction, losses of azurin by desorption, kinetic selectivity at this frequency, and iRu drop. The values of E0′ estimated from the first harmonic voltammograms over a range of frequencies for the azurin-modified HOPG, BPG, and EPG electrodes are shown in Table S-1 (Supporting Information) and are similar for measurements made at each surface. Figure 8a displays the fundamental harmonic voltammogram measured experimentally (f ) 36 Hz, ∆E ) 80 mV) for an azurinmodified BPG electrode compared with that generated by simulation (envelope representation). Reasonable agreement between experiment and theory could be attained using parameters E0′, Cdl, and Γfirst closely matching those determined from dc cyclic voltammetry and Ru measured experimentally, if a k0′ value of 1000 s-1 is assumed to apply. However, if a k0′ value of 108 s-1 (reversible process) was used in the simulation, then an increase of only 4 × 10-13 pmol‚cm-2 (or 2%) in Γfirst was required to achieve agreement between predicted and experimental peak currents. Thus, a measured k0′ value of g1000 s-1 is achieved from analysis of the fundamental harmonic, in agreement with dc cyclic voltammetry. However, it needs to be noted that the first harmonic, particularly when measured at low frequencies, is not sensitive to rapid kinetics, and for this purpose, higher harmonics are preferable for kinetic evaluation. A limitation of the simulation procedure presently developed is that it is only able to include a potential independent value of Cdl, and hence, nonlinear backgrounds (i.e., potential-dependent capacitance) such as that observed with graphite electrodes, may introduce some error into the analysis. An alternative is to first apply the background subtraction procedure to the experimental data (as previously described) and then to simulate the voltammogram. In this case, the simulation does not now need to include any contribution from Cdl but correction for iRu becomes less accurate. Figure 8b shows the fundamental harmonic voltammogram (as in Figure 8a) after the background subtraction procedure had been performed. Arguably, this permits a superior experi-
Figure 8. Experiment (black line) vs simulation (red line) comparison of the first harmonic voltammograms before (a) and after (b) background subtraction for azurin adsorbed onto a BPG electrode. Experimental conditions: f ) 36 Hz, ∆E ) 80 mV, v ) 0.09686 V.s-1, Ei ) 0.650 V, and Es ) 0 V. Simulation parameters: (a) Ru ) 170 Ω, Cdl ) 36 µF‚cm-2, Γfirst ) 21 pmol‚cm-2, and E0′ ) 0.321 V; k0′ ) 1000 s-1, R ) 0.5; (b) Ru ) 170 Ω, Cdl ) 0 µF‚cm-2, Γfirst ) 19.7 pmol‚cm-2, E0′ ) 0.318 V, k0′ ) 1000 s-1, and R ) 0.5.
Figure 9. Fourth harmonic voltammograms for azurin adsorbed onto HOPG (black), BPG (red), and EPG (blue) electrodes. Experimental conditions as per Figure 7, except: f ) 22 Hz, ∆E ) 80 mV, v ) 0.09686 V‚s-1, Ei ) 0.650 V, and Es ) 0 V.
ment-theory comparison to be undertaken given the zero baselines in each case. Again though, the background-subtracted first harmonic is relatively insensitive to changes in kinetics for k0' g 1000 s-1 under the conditions used to obtain the experimental data in Figure 7. Nevertheless, the shape of experimental and theoretical voltammograms now may be seen to match very closely when data are analyzed in the fundamental harmonic format. Indeed, the simulated and experimental shapes are almost indistinguishable, in contrast to dc data. As expected, kinetic dispersion is not detected under these conditions at the BPG electrode where ideal reversible behavior is indicated. (2) Second and Third Harmonics. An example of the second and third harmonic ac components displayed as a current-time response, when employing a sinusoidal waveform (f ) 22 Hz, ∆E ) 80 mV) for each of the azurin-modified HOPG, BPG, and EPG, is shown in Figure S-1. The second harmonics, and to a lesser degree the third harmonics, have a distinct level of background current present, which induces asymmetry in the current lobes lying either side of E0′. However, the values of E0′ estimated from the second and third harmonic voltammograms over a range of frequencies for the azurin-modified HOPG, BPG, and EPG electrodes are shown in Table S-1 and agree well with first harmonic estimates.
Figure 10. Theoretical fourth harmonic peak current data as a function of frequency for an electron-transfer process having a k0′ value of 1000 (b), 3000 (9), and 1 × 108 s-1 (2) in the absence (open symbols) and presence (closed symbols) of iRu drop effects (Ru ) 170 Ω). Simulation parameters as per experiment in Figure 9, plus: (a) Cdl ) 38 µF‚cm-2, Γfourth ) 11 pmol‚cm-2, E0′ ) 0.317 V, ∆E ) 0.08 V, and R ) 0.5.
(3) Fourth Harmonic. An example of the fourth harmonic ac component displayed as a current-time response, when employing a sinusoidal waveform (f ) 22 Hz, ∆E ) 80 mV) for each of the azurin-modified HOPG, BPG, and EPG electrodes, is shown in Figure 9. While these purely faradaic currents are small, they are still readily measurable above the noise. A key advantage of using the fourth and higher harmonics is that they do really contain negligible background capacitance current.12,13 This allows the direct analysis of the faradaic current to be undertaken, as no empirical background subtraction procedure is required. The values of E0′ estimated from the fourth harmonic voltammograms (f ) 22 Hz, ∆E ) 80 mV) were 0.316, 0.324, and 0.317 V for the azurin-modified HOPG, BPG, and EPG electrodes, respectively, which compared well with the values obtained from the dc data. Minimal variation in the value of E0′ calculated from the position of zero current was observed at each applied frequency (Table S-1). The magnitude of the separation between the cathodic and anodic peak potential values obtained from the fourth harmonic voltammograms for the azurin-modified BPG and EPG electrodes was typically
