Spectroelectrochemistry at Screen-Printed Electrodes: Determination

Article pubs.acs.org/ac

Spectroelectrochemistry at Screen-Printed Electrodes: Determination of Dopamine Noelia González-Diéguez, Alvaro Colina, Jesús López-Palacios, and Aránzazu Heras* Department of Chemistry, Universidad de Burgos, Pza. Misael Bañuelos s/n, E-09001 Burgos, Spain S Supporting Information *

ABSTRACT: A new device to perform spectroelectrochemical measurements in the UV/visible spectral region using screen-printed electrodes has been developed. Neurotransmitter dopamine has been selected as a proof of concept of the capabilities of the new device. The results obtained have allowed us both to study the oxidation mechanism of dopamine and to carry out the spectroelectrochemical detection of this neurotransmitter. Differences in dopamine oxidation mechanism have been observed depending on the initial concentration. Thus, dopamine concentrations lower than 10−3 M led to a higher generation of dopaminochrome and its derivatives with a band centered at 305 nm, which was the best wavelength to determine dopamine spectrophotometrically at these concentrations. However, if dopamine concentration is higher than 10−3 M, dopaminoquinone is stable enough to use its maximum of absorbance, 395 nm, to detect this neurotransmitter. Dopamine concentration can also be calculated from the electrochemical data in spectroelectrochemistry, the results being comparable to that obtained from spectroscopic data. Comparison between spectrophotometric and electrochemical determinations demonstrates that the two methods measure this analyte indistinctively, proving that spectroelectrochemistry represents an autovalidated technique. Partial least-squares regression has also been used, obtaining good results in the full dopamine concentration range. Finally, as spectroelectrochemistry is an intrinsically trilinear technique, PARAFAC has been used to study the effect of probable interfering species.

S

spectroelectrochemical measurements with different purposes,7,11,32,33 they are not very common in spectroelectrochemistry. This type of spectroelectrochemical experiments simplifies enormously the experimental setup, enabling the use of this analytical technique for routine analysis. The main advantage of spectroelectrochemistry at SPEs is to obtain two independent signals in a single experiment, both related to the same chemical system at exactly the same time, working with disposable electrodes. Second, but not less important, our experimental setup allows us to reduce substantially the sample volume. One-drop spectroelectrochemical measurements34,35 are possible thanks to the use of an optical fiber system and to the small size of the three-electrode system in SPEs. Dopamine (DA) is a chemical messenger in the mammalian central nervous system, a neurotransmitter that controls many biological functions such as cognition, emotion, endocrine regulation, motivation, locomotion, and so on. Furthermore, some pathologies such as Parkinson disease or schizophrenia and the use of some type of drugs are linked to a dysregulation of dopaminergic transmission.36 The study of DA reaction mechanism and the knowledge of DA level is of great importance because, for example, low levels of DA are related to Parkinsonism and high levels of this neurotransmitter are linked to schizophrenia.36 DA has been studied and determined

pectroelectrochemistry (SEC) is a hybrid technique that provides simultaneously electrochemical and spectroscopic information about a system susceptible to be oxidized and reduced. In theory, any electrochemical technique can be combined with any spectroscopic one, providing different information depending on the kind of analysis and system to be studied. Since 1964, when this new technique was presented1 linking chronopotentiometry with UV/visible absorptometric measurements, significant improvements and changes have been done. For example, electrodes are no longer limited to optically transparent ones and opaque surfaces can also been used. With this type of working electrode, absorptometric measurements can be performed using a long pathway spectroelectrochemical cell2,3 or a reflection setup.4,5 Actually, not only UV/visible absorption, but also Raman,6,7 infrared,8−11 electron paramagnetic resonance (EPR),12 X-ray,13,14 nuclear magnetic resonance (NMR),15,16 or fluorescence17,18 experiments have been successfully coupled with electrochemistry. Also, mass spectrometry9,10,19 and chromatographic separations19 have been combined with spectroelectrochemistry. Spectroelectrochemistry has been proven to be very useful in the study of reaction mechanism20−22 or characterization of materials,23,24 but it is not widely used for quantitative analysis.25−28 On the contrary, screen-printed electrodes (SPEs) have been extensively used for quantitative purposes due to its simple modification to perform selective, sensitive, and rapid analysis.29−31 Moreover, SPEs are disposable, guaranteeing fresh and reproducible surfaces for routine analysis. Although some recent works have used SPEs in © 2012 American Chemical Society

Received: July 2, 2012 Accepted: October 15, 2012 Published: October 15, 2012 9146

dx.doi.org/10.1021/ac3018444 | Anal. Chem. 2012, 84, 9146−9153

Analytical Chemistry

Article

electrochemically because of its intrinsic redox nature.37,38 Problems with this type of analysis arise from the polymerization of the quinoid form of DA, leading to a nonconductive polymer that renders the electrode surface inactive.39,40 Disposable electrodes such as SPEs avoid this problem because each analysis is carried out on a new surface very similar to the previous one, without the need of cleaning the working electrode surface. UV/visible absorption spectroscopy has also been selected to analyze DA solutions41,42 because both the neurotransmitter and its oxidation products absorb radiation in this spectral range. For this reason, spectroelectrochemistry has proven to be a very useful technique to study this neurotransmitter not only in the UV/visible spectrum region,42−44 but also in the infrared one.45 In this work we demonstrate that DA can be determined by using UV/visible absorption spectroelectrochemistry at screenprinted electrodes. Consequently, the main objective of this work is to show the experimental setup to perform spectroelectrochemical experiments at screen-printed electrodes. Our second objective is to demonstrate the capabilities of UV/visible SEC at SPEs in the study of both the oxidation mechanism of dopamine and its quantitative analysis in aqueous media in presence and absence of interfering compounds.

Figure 1. Scheme of the spectroelectrochemical cell using a screenprinted electrode (SPE; RE = reference electrode, WE = working electrode, CE = counter electrode). The spectroelectrochemical cell has been drawn in proportion to its actual size.

trochemical (NNIRS) setup. The reflection probe is placed inside the solution at 1.25 mm from the SPE to collect the reflected light. In this way, the thickness of the layer is larger than the diffusion layer and the reaction occurs under semiinfinite diffusion conditions. The reflection probe consists of a bundle of seven 200 μm optical fibers (bundle diameter: 0.625 mm). The optical configuration consists of six illumination fibers around one read fiber that collects the light reflected on the surface of the WE to the QE65000 detector. Absorbance integration time (tintegration) changed between 100 and 220 ms. A volume of 50 μL is placed on the SPE surface for each experiment, covering completely the three electrodes of the SPE.

■

EXPERIMENTAL SECTION Chemicals and Materials. All solutions were prepared with high-quality water (Milli-Q A10 system, Millipore, Bedford, U.S.A.). Buffer solutions were prepared from analytical reagent grade chemicals without further purification. The supporting electrolyte was a phosphate buffer solution (PBS) containing 0.5 M KCl (99%, Panreac) and 0.2 M Na2HPO4 (Merck)−NaH2PO4 (Panreac), pH = 7. Dopamine hydrochloride (3-hydroxytyramine hydrochloride) 99% was obtained from Acros Organics and used as received. All solutions were prepared daily. The solution pH was measured with a pH-glass electrode (micropH 2002, Crison). All experiments were performed at room temperature Instrumentation. Cyclic voltammetry was carried out using a CHI730A potentiostat (CH Instruments). All experiments were performed using commercial screen-printed electrodes (DRP-110, DropSens), which included a three-electrode configuration printed on the same strip. The strips had a 4 mm diameter disk screen-printed carbon working electrode (WE), a carbon counter electrode (CE), and a silver pseudoreference electrode (RE). Each screen-printed electrode was placed in a box connector DSC (DropSens) that held the SPE and allowed us to reproduce its position between measurements. A pseudoreference electrode is useful for this type of analysis, and no significant shift of potential was observed with dopamine concentration, obtaining a good reproducibility between different solutions (±7 mV). Special care has to be taken when compounds that can oxidize silver are analyzed. Spectroelectrochemical experiments were performed in the UV/visible spectral range (270−1000 nm) using a deuterium light source (Ocean Optics, model DH-2000), coupling the potentiostat with a QE65000 spectrometer (Ocean Optics) that consists of a 2D diode array of 1044 × 64 pixels. A PEEK reflection/backscattering UV/vis probe (R200-7-UV/vis, Ocean Optics) with an external diameter of 6.25 mm, was focused directly on the carbon working electrode surface (Figure 1) in a near-normal incidence reflection spectroelec-

■

RESULTS AND DISCUSSION Spectroelectrochemistry of Dopamine Electrooxidation. During the first stages of this study, we have performed experiments to explain the dopamine oxidation mechanism. Figure 2 shows the spectra and the voltammogram recorded

Figure 2. Spectra evolution during oxidation of dopamine 10−3 M in PBS buffer solution (pH = 7). Inset: Linear voltammogram registered during dopamine oxidation; Einitial = −0.50 V, Efinal = +0.70 V, scan rate = 0.05 V s−1, tintegration = 135 ms.

while the potential was scanned at 0.05 V s−1 between −0.50 and +0.70 V, potential range where dopamine 10−3 M in PBS buffer can be oxidized. Linear voltammogram (inset Figure 2) shows only one anodic peak at +0.28 V related to, a seemingly simple oxidation process. However, spectra evolution (Figure 2) evidence the complexity of this oxidation process. From +0.13 V to +0.44 V, the oxidation of DA takes place, as indicated by the anodic peak. Electrical signal does not provide 9147



Article

+0.44 V. The absorption band at 395 nm continues to increase roughly at the same rate, while the band at 470 nm becomes hardly noticeable. From these results we conclude that, at high dopamine concentration, the intramolecular cyclization and later oxidation is hindered during this first oxidation step, where DAQ is generated, and dominates the spectral response. Cyclic voltammetry was used to understand how this oxidation process evolved in subsequent potential cycles. Three consecutive potential cycles between −0.50 V and +0.70 at 0.05 V·s−1 were performed using the two DA concentrations analyzed in Figures 2 and 3, and recording the spectra during the voltammetry. Multivariate curve resolutionalternating least squares (MCR-ALS)50 was used to deconvolve the spectra of the main compounds formed during the dopamine oxidation. By selecting two components and appropriate constraints, as non-negativity of concentration and absorbance, the normalized spectra of the two main components were resolved (Figure 4a,b). An initial estimation of the pure spectra was performed using Independent Component Analysis (ICA) to simplify the deconvolution. After applying MCR-ALS, calculated spectra for the two components (C1 and C2) at the two DA concentrations does not reveal significant differences, with one band at 480 nm for the first component (C1) and other band centered at 395 nm for the second component (C2). The normalized concentration profiles of these two components vs time (Figure 4c,d) and potential (Figure 4e,f) were also calculated, showing significant differences for the two dopamine concentrations. C1 profile increases during the three potential cycles in the case of DA 10−3 M, while for DA 6 × 10−3 M it increases drastically in the first reduction scan and the second oxidation scan. This growth is a function of time (not of the potential applied), reaching an almost constant value in the last scan. From these results, we conclude that C1 profile is related to the generation of a polymer on the electrode surface. The amount of polymer generated increases during the three potential cycles for dopamine solutions 10−3 M, while it reaches a maximum value in two cycles for dopamine solutions 6 × 10−3 M. From these concentration profiles it is also possible to observe that polymerization starts at potentials higher than +0.35 V. C2 profile shows the same maximum value for the three potential cycles in the case of the lowest DA concentration. In the case of the highest DA concentration, C2 profile grows during the first two cycles reaching an almost constant value in the last potential cycle. The second component (C2) for the two DA concentrations starts to increase at around +0.20 V. Therefore, this second component can be related to dopaminoquinone generation whose absorbance maximum is located at 395 nm. It is noteworthy that the absolute C2 values are very small (multiplied by 10 in Figure 4c,e for a clear observation) for 10−3 M DA concentration, indicating that DAQ is readily consumed as it is generated, yielding compounds that form the polymer on the electrode surface. However, in the case of the highest DA concentration, C2 values are about 10 and 30 times higher, meaning that DAQ in solution is quite stable because it is produced in a higher amount than the needed to start the cyclization and the polymerization reaction on the electrode surface. These results explain the higher absorbance ratio (A395/A300 = 0.70) observed in Figure 3 at +0.44 V in comparison with the one (A395/A300 = 0.59) observed in Figure 2 at the same potential. Spectroelectrochemical Determination of Dopamine. Spectroelectrochemical experiments supply not only mecha-

much more information about the process at this potential range. Simultaneously recorded spectral changes are remarkable, with two absorbing bands centered at 305 and 395 nm. These two bands are attributed to an intraligand π → π* transition derived from the 1Lb ← 1A1 dopamine transition typical of a semiquinone46,47 and to the n → π* transition related to the carbonyl group linked to a benzene ring distinctive of the o-quinone,40,47 respectively. This band at 395 nm is attributed to dopaminoquinone (DAQ), the first oxidation product of dopamine.40 New spectral changes were observed in the potential range from +0.44 to +0.70 V, where the oxidation process should be mainly diffusion controlled, as indicated by the electrical signal. While the increment of absorbance at 395 nm is slowed down, a new band emerges centered at 470 nm and overlapped with that at 395 nm. This third band is also assigned to a n → π* transition localized in the carbonyl group linked to the benzene ring of the oxidized product of dopamine that has undergone an intramolecular cyclization to give the leucodopaminochrome (LDAC). This compound is easily oxidized to dopaminochrome (DAC)40,48 at potentials higher than +0.44 V. After subsequent rearrangements and polymerization processes a dark melanin-like polymer is formed on the electrode surface, rendering the working electrode inactive.40,48,49 Spectroelectrochemical experiments indicate that this mechanism depends on dopamine concentration. Figure 3 displays

Figure 3. Spectra evolution during oxidation of dopamine 6 × 10−3 M in PBS buffer solution (pH = 7). Inset: Linear voltammogram registered during dopamine oxidation. Einitial = −0.50 V, Efinal = +0.70 V, scan rate = 0.05 V s−1, tintegration = 135 ms.

the voltammogram and spectra evolution of a 6 × 10−3 M dopamine solution in PBS buffer during its oxidation, keeping constant the rest of experimental variables. The voltammogram waveform changes considerably. The oxidation peak is shifted to more anodic potentials, around +0.43 V, being less abrupt the initial growth of the current. Changes of the electrical signal are concomitant with spectral changes. From +0.13 V to +0.44 V, two absorption bands at 300 and 395 nm are observed, which are better defined than in the previous experiment, as a consequence of the higher concentration of dopamine in solution. It is worth pointing out that the ratio between the absorbance at 395 and at 300 nm (A395/A300) at +0.44 V is higher when the dopamine concentration is 6 × 10−3 M (A395/ A300 = 0.70) than when DA concentration was 10−3 M (A395/ A300 = 0.59), indicating that the relative amount of DAQ increases with dopamine concentration. This increment determines the spectral behavior at potentials higher than 9148



Article

Figure 4. (a, b) MCR-ALS deconvolution of spectra registered during dopamine oxidation; (c, d) concentration profiles vs time of the different species resolved by MCR-ALS; (e, f) concentration profiles vs potential of the different species resolved by MCR-ALS; (g, h) cyclic voltammograms registered during dopamine oxidation between −0.50 V and +0.70 at 0.05 V s−1 (3 cycles). (a, c, e, g) Cdopamine = 10−3 M; (b, d, f, h) Cdopamine = 6 × 10−3 M. All solutions were prepared in PBS buffer media (pH = 7.00); tintegration = 135 ms.

prepared in the range from 9 × 10−5 to 10−3 M. Linear voltammetry was selected to carry out the calibration experiments by scanning the potential between 0.0 V and +0.70 at 0.05 V s−1 and recording both the linear voltammo-

nistic information related to dopamine oxidation but also quantitative information. With this aim, different calibration curves were performed. According to the results explained above, a first calibration set, with seven dopamine samples, was 9149



Article

and A3 were less reliable as can be deduced by the high RSD values (Table 2). Electrochemical regression (VC) also indicates that peak current is a suitable response to carry out this determination. Similarity of results from VC and A1 methods indicate that both methods, indistinctively, can be used to dopamine detection. IUPAC recommends that the trueness of analytical results should be validated by experiments performed with a second and independent method. 51 Spectroelectrochemical experiments supply two different and independent analytical responses. Therefore, once demonstrated that the two methods can be used separately to determine dopamine, the absence of bias have to be verified comparing estimations obtained with the two methods (VC and A1). Being a the slope of the regression and b the intercept, we obtain the following fitting: a = [1.00 ± 0.08] and b = [0.16 ± 4.91] × 10−5. The absence of bias in the analytical results is verified concluding that the two methods can measure dopamine indistinctively. These results show that UV/vis spectroelectrochemistry is an autovalidated technique, increasing its added value. Results presented in Figures 2 and 3, indicates that dopamine oxidation process depends on the concentration. It is wellknown, that calibration parameters can change in different concentration ranges. A second calibration set, with eight dopamine samples, was prepared in the range of 1.5 × 10−3 to 6 × 10−3 M to find out if high concentrations of dopamine have an effect on the calibration model. Other experimental conditions were the same as in the previous calibration experiments performed from 9 × 10−5 M to 10−3 M. Peak current for electrochemical data and absorbance at +0.70 V at 305, 395, and 470 nm were also selected as responses for the calibration. Table 3 summarizes the calibration parameters obtained from the four regressions curves, assessed by least square regressions

gram and the spectra evolution. In a spectroelectrochemical experiment, two independent calibrations can be carried out, one from electrochemical and the other from the spectroscopic data. For the electrochemical calibration we selected the peak current as response, while for the spectrophotometric one, the absorbance values at +0.70 V at the three maxima showed in spectra in Figures 2 and 3 (305, 395, and 470 nm) were chosen as responses. Regression parameters for the four univariate regression curves in this concentration range are shown in Table 1. Outlier Table 1. Calibration Parameters Obtained for the Determination of Dopamine in the Concentration Range from 9 × 10−5 to 10−3 M by UV/Visible Spectroelectrochemistrya method

sensitivity

intercept

residual standard deviation (Syx)

VC A1

0.037 A 50.87 a.u. M−1 16.79 a.u. M−1 21.25 a.u. M−1

−3.6 × 10−6 A 7.4 × 10−3 a.u.

5.60 × 10−7 8.55 × 10−4

0.9973 0.9980

1.2 × 10−5 a.u.

8.94 × 10−4

0.9801

4.0 × 10−4 a.u.

8.67 × 10−4

0.9882

A2 A3

coefficient of determination (R2)

Methods indicated in the first column correlate peak current (VC), +0.70V +0.70V A+0.70V 305nm (A1), A395nm (A2), A470nm (A3) vs dopamine concentration (CDA). a

detection has been carried out by least median of squares (LMS) regressions. After removing the outlier points, a least square regression (LS) was performed to predict the unknown problems. Table 2 shows the concentration prediction of two Table 2. Concentration Estimated from Regression Curves Tabulated in Table 1a CP1 = 5.5 × 10−4 M

method VC A1 A2 A3

%RSD (α = 0.05)

CI (M) [5.3 [5.7 [5.7 [5.2

± ± ± ±

0.4] 0.5] 1.6] 1.2]

× × × ×

10−4 10−4 10−4 10−4

2.4 3.2 10.2 8.5

Table 3. Calibration Parameters Obtained for the Determination of Dopamine in the Concentration Range from 1.5 × 10−3 and 6 × 10−3 M by UV/Visible Spectroelectrochemistrya

CP2 = 2.5 × 10−4 M %RSD (α = 0.05)

CI (M) [2.6 [2.4 [3.0 [2.5

± ± ± ±

0.4] 0.4] 1.2] 1.0]

× × × ×

10−4 10−4 10−4 10−4

4.8 6.1 15.2 14.0

residual standard deviation (Syx)

coefficient of determination (R2)

method

sensitivity

intercept

VC

0.026 a.m.−1 39.83 a.u. M−1 17.45 a.u. M−1 12.91 a.u. M−1

1.7 × 10−5 A

4.0 × 10−6

0.9920

4.3 × 10−2 a.u.

1.17 × 10−2

0.9617

−9.5 × 10−3 a.u.

1.78 × 10−3

0.9950

2.4 × 10−3 a.u.

3.0 × 10−3

0.9803

a

A1

CP1 and CP2 are the concentrations of the two problem samples of dopamine; CI: confidence interval. %RSD: relative standard deviation. Methods indicated in the first column correlate peak current (VC), +0.70V +0.70V A+0.70V 305nm (A1), A395nm (A2), A470nm (A3) vs dopamine concentration (CDA).

A2 A3

Methods indicated in the first column correlate peak current (VC), +0.70V +0.70V A+0.70V 305nm (A1), A395nm (A2), A470nm (A3) vs dopamine concentration (CDA). a

problem samples prepared to evaluate the prediction capability of the four calibration curves. The best results correspond to regressions performed with electrochemical data and the spectroscopic data obtained from 305 nm absorbance values at +0.70 V. These two regressions have the highest coefficients of determination (Table 1) and lead to the lowest relative standard deviation (%RSD) for the estimations (Table 2). From these results, correlated to spectral evolution showed in Figure 1, we conclude that in the case of dopamine concentrations lower than 10−3 M, the best wavelength to perform the optical determination is 305 nm (A1), wavelength at which contribution of cycled-derivatives as dopaminochrome is higher. It can be also concluded that R2 values for methods A2

(LS) after outliers’ detection by least median of squares regressions (LMS). The prediction capability of each calibration curve was calculated for two different problem samples (Table 4). In this case, the best results derived from the voltammetric signal and from the absorbance at 395 nm, with the highest coefficient of determination and the lowest % RSD values were obtained. It can be also deduced that, in this case, method A1 was the less reliable because of the highest RSD value (Table 4). As explained above (results shown in Figure 3), DAQ, the first oxidation product of dopamine, is 9150



Article

generated during the oxidation of dopamine. In multivariate regression, these problems do not affect the final prediction. Comparison between the estimations found with the electrochemical regression (VC) and PLSR absorptometric regression lead to unbiased models, a = [1.00 ± 0.01] and b = [0.08 ± 3.14] × 10−5 (Figure S2). Once again the autovalidation of spectroelectrochemical methods has been demonstrated. Effect of Probable Interfering Species. As has been shown, our spectroelectrochemical setup is useful for quantitative analysis, but the most important feature of spectroelectrochemistry is the large amount of information that can be obtained in a single experiment. Spectroelectrochemistry is an intrinsically trilinear technique that can be used for multiway data analysis. Particularly, we have selected PARAFAC52,53 as chemometric tool to assess the concentrations and signals of the individual components of a complex sample. As a proof of concept, we have selected the detection of dopamine in presence of catechol (CAT). Usually, these two analytes have been resolved using separation techniques such as liquid chromatography or microfluidics. Voltammetry is not a useful technique to resolve this mixture because voltammograms are completely overlapped, Figure S3a. However, spectroelectrochemistry can be an alternative for the resolution of this problem due to the differences found in the spectra, Figure S3b. We have prepared ten samples with concentrations of dopamine between 0 and 4 × 10−4 M and catechol between 0 and 2 × 10−3 M, Table S1. We have calculated a two-factor constrained PARAFAC model assuming the non-negativity of the concentrations, spectra and voltabsorptograms from an array of data of dimension (10 concentrations × 132 wavelengths × 85 potentials). Figure 5 shows the deconvolution of the mixtures (scores, loadings vs potential and loadings vs wavelength). The main advantage of the PARAFAC model is that it is not essential to know the concentration of the interfering compounds because the scores of the concentrations are resolved because of the trilinearity of the data, Figure 5a. We have added a sample identified as outlier (triangle, Figure 5a) to show that it is clearly not linear with the other scores. For this chemical problem, outliers are easily detected using spectroelectrochemistry. We have observed that, in some cases, for replicated experiments in which the electrochemical behavior is very similar, Figure S4a, a new band appears at wavelengths higher than 550 nm. Figure S4b shows the different voltabsorptograms at 600 nm obtained for the two experiments. Outlier samples exhibit an anomalous increment of absorbance that is not observed in samples suitable for prediction. We can affirm that this strange behavior does not depend on possible reactions taking place in solution because the two experiments have been done using the same solution. Further studies would be needed to understand this process, but it is clear that outliers can be detected analyzing the absorbance at 600 nm. Experiments showing this anomalous absorbance cannot be used to calculate the PARAFAC model or to predict the test sample. As can be seen, evolution of the loadings with potential, related to the voltabsorptograms, shows the strong overlapping of the oxidation processes (Figure 5b). Loadings versus wavelength (Figure 5c) are directly related to the absorption spectra of the two components (Figure S3b). We have predicted the concentration of a test sample with concentration 3 × 10−4 M of dopamine in presence of catechol 5 × 10−3 M, obtaining a value of 2.88 × 10−4 M for dopamine (recovery of 96%). Moreover, if the concentrations of the

Table 4. Concentration Estimated from Regression Curves Tabulated in Table 3a CP1 = 4.5 × 10−3 M

method VC A1 A2 A3

%RSD (α = 0.05)

CI (M) [4.7 [4.5 [4.6 [4.2

± ± ± ±

0.4] 0.8] 0.3] 0.6]

× × × ×

CP2 = 2.5 × 10−3 M

10−3 10−3 10−3 10−3

3.6 7.2 2.6 5.9

%RSD (α = 0.05)

CI (M) [2.6 [3.0 [2.3 [2.9

± ± ± ±

0.4] 0.8] 0.3] 0.6]

× × × ×

10−3 10−3 10−3 10−3

6.7 10.5 4.9 8.6

a

CP1 and CP2 are the concentrations of the two problem samples of dopamine; CI is the confidence interval; and %RSD is the relative standard deviation. Methods indicated in the first column correlate +0.70V +0.70V peak current (VC), A+0.70V 305nm (A1), A395nm (A2), A470nm (A3) vs dopamine concentration (CDA).

quite stable at these high DA concentrations and this compound has an absorbance maximum at 395 nm. These two factors explain the results obtained from the spectroscopic data. As in the previous calibration set, the comparison between the estimations found with the electrochemical method (VC) and absorptometric method at 395 nm (A2) yields the absence of bias, a = [0.99 ± 0.18] and b = [0.35 ± 6.65] × 10−4. We conclude that the two methods can measure dopamine without distinction, confirming that spectroelectrochemistry is an autovalidated analytical technique. Absorptometric univariate regression models are not useful for the whole concentration range (from 9 × 10−5 to 6 × 10−3 M). This problem can be circumvented by using multivariate regression, as will be demonstrated below. Full spectra every 135 ms, between 250 and 900 nm, were recorded in a typical spectroelectrochemistry experiment. Taking into account all this spectral information, multivariate regression can be used to obtain more valuable information related to dopamine oxidation. In this case, the calibration set includes all dopamine solutions analyzed separately previously, from 9 × 10−5 to 6 × 10−3 M. Among the different multivariate regression methods, partial least squares regression (PLSR) was selected in this case. From spectra in Figures 2 and 3 it is possible to select the most informative wavelengths, from 290 to 550 nm, because this range displays all the information related to dopamine oxidation. To simplify the analysis, only spectra at vertex potential (+0.70 V) were selected, taking absorbance at these wavelengths as independent variables or predictors for each sample measured. In this case, the number of predictors (330 wavelengths) is much larger than the number of observations or dependent variables (15 samples). Obviously, there are problems of multicollinearity that are resolved by PLSR. The latent variables with the best predictive power to be used in the regression were calculated and selected. In this part of the work, the whole calibration set (from 9 × 10−5 to 6 × 10−3 M) was used to construct the PLSR model. Regression model has been performed selecting two latent variables that are related to the two main compounds generated during dopamine oxidation, DAQ, and DAC, as can be deduced from the loadings plotted in Figure S1. The concentration of two dopamine problem samples has been estimated from the optical data, with the following confidence intervals: [2.6 ± 0.1] × 10−3 (%RSD = 1.8), [6.4 ± 0.8] × 10−4 (%RSD = 5.6). From these data we conclude that, while absorptometric univariate regression models are influenced by dopamine concentration, leading to wrong predictions because of the contribution of the compounds 9151



Article

good response under 290 nm for a better deconvolution of the dopamine.

■

CONCLUSIONS UV/vis absorption spectroelectrochemistry is not only a very good technique to understand reaction mechanisms, but also an excellent technique for quantitative purposes. The two features have been demonstrated in our case for the study and determination of dopamine at physiological pH. Spectroelectrochemistry at screen printed electrodes is very advantageous because small sample volumes are used, while providing a large amount of information in only one experiment. Dopamine concentration has a deep influence on the products formed during its oxidation. If DA concentration is lower than 10−3 M, DAC and other intramolecular cyclization derivatives predominantly contribute to the absorption spectra. This fact is directly related to the wavelengths more suitable for DA determination, centered on 305 nm. However, when DA concentration is higher than 10−3 M, DAQ is the chemical oxidation compound more dominant in the spectral response, 395 nm becoming the most appropriate wavelength for DA determination. Differences with regard to spectral bands more suitable for DA quantification are circumvented by using multivariate regression models. Particularly absorptometric DA determination by PLS regression can be performed in a wider concentration range than in the case of univariate regression, leading to better results. We have also demonstrated that spectroelectrochemistry is an autovalidated technique, which is an added value for quantitative analysis. Finally, our device has been successfully used for the resolution of a mixture of dopamine and catechol that cannot be resolved electrochemically, demonstrating the great capability of spectroelectrochemistry to resolve complex problems, even without knowing the interfering species.

■

ASSOCIATED CONTENT

S Supporting Information *

Supplementary data associated with this article. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +34 947 25 88 31.

Figure 5. Deconvolution of the data matrix. (a) Scores vs concentration, (b) loadings vs potential, and (c) loadings vs wavelength for dopamine−catechol mixtures prepared in PBS buffer solution (pH = 7); Einitial = 0.00 V, Efinal = +0.70 V, scan rate = 0.01 V s−1, tintegration = 100 ms; Concentration of mixtures in Table S1.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Support from Ministerio de Ciencia y Tecnologı ́a (CTQ201017127) and Junta de Castilla y León (GR71, BU197A12-2) is gratefully acknowledged. N.G.D. thanks ECyL and Junta de Castilla y León her research contract.

interfering compound, catechol, were known and introduced in the model, we can also estimate its value (5.5 × 10−3 M) from the scores assessed by the PARAFAC model, obtaining a good concordance with the real catechol concentration (recovery of 110%). This demonstrates that spectroelectrochemistry can be easily used for the resolution of complex samples. The main advantage of spectroelectrochemistry is that the redox potentials of the electroactive couples and their spectra cannot always coincide. However, special care has to be taken in selecting the best instrument to perform the determination. Two aspects have to be considered, the spectrometer resolution to obtain a good separation of the spectra and the good performance of the spectrometer in the spectral region of interest. For example, for the determination of catechol and dopamine it would have been better to use a spectrometer with

■

REFERENCES

(1) Kuwana, T.; Darlington, R. K.; Leedy, D. W. Anal. Chem. 1964, 36, 2023−2025. (2) Hu, X. W.; He, J. B.; You, Y. H.; Zhang, Y. M.; Zhang, S. L. Electrochim. Acta 2011, 56, 1219−1226. (3) Zhang, J.; Zhou, A.; Xie, Q.; Wei, W. Talanta 2000, 52, 903−908. (4) Shi, P.; Fromondi, I.; Shi, Q.; Wang, Z.; Scherson, D. A. Anal. Chem. 2007, 79, 202−207. (5) Ruiz, V.; Colina, A.; Heras, M. A.; López-Palacios, J. Electrochem. Commun. 2007, 9, 255−261. (6) Colina, A.; Ruiz, V.; Heras, A.; Ochoteco, E.; Kauppinen, E.; López-Palacios, J. Electrochim. Acta 2011, 56, 1294−1299.

9152



Article

(43) Van Dyke, D. A. Anal. Chem. 1988, 60, 1256−1260. (44) Mayausky, J. S.; McCreery, R. L. Anal. Chem. 1983, 55, 308− 312. (45) Wang, X.; Baokang, J. I. N.; Xiangqin, L. I. N. Anal. Sci. 2002, 18, 931−933. (46) Barreto, W. J.; Barreto, S. R. G.; Ando, R. A.; Santos, P. S.; DiMauro, E.; Jorge, T. Spectrochim. Acta, Part A 2008, 71, 1419−1424. (47) Barreto, W. J.; Ponzoni, S.; Sassi, P. Spectrochim. Acta, Part A 1998, 55, 65−72. (48) Corona-Avendaño, S.; Alarcón-Angeles, G.; Ramírez-Silva, M. T.; Rosquete-Pina, G.; Romero-Romo, M.; Palomar-Pardavé, M. J. Electroanal. Chem. 2007, 609, 17−26. (49) Lane, R. F.; Hubbard, A. T. Anal. Chem. 1976, 48, 1287−1292. (50) Jaumot, J.; Gargallo, R.; de Juan, A.; Tauler, R. Chemom. Intell. Lab. Syst. 2005, 76, 101−110. (51) Danzer, K.; Currie, L. A. Pure Appl. Chem. 1998, 70, 993−1014. (52) Andersen, C. M.; Bro, R. J. Chemom. 2003, 17, 200−215. (53) Bro, R. Chemom. Intell. Lab. Syst. 1997, 38, 149−171.

(7) Robinson, A. M.; Harroun, S. G.; Bergman, J.; Brosseau, C. L. Anal. Chem. 2012, 84, 1760−1764. (8) Kavan, L.; Dunsch, L. ChemPhysChem 2007, 8, 974−998. (9) Heinen, M.; Chen, Y. X.; Jusys, Z.; Behm, R. J. ChemPhysChem 2007, 8, 2484−2489. (10) Heinen, M.; Chen, Y. X.; Jusys, Z.; Behm, R. J. Electrochim. Acta 2007, 52, 5634−5643. (11) Shih, W. Y.; Yang, J. Electroanalysis 2006, 18, 267−274. (12) Dunsch, L. J. Solid State Electrochem. 2011, 1−16. (13) Dewald, H. D.; Watkins, Ii, J. W.; Elder, R. C.; Heineman, W. R. Anal. Chem. 1986, 58, 2968−2975. (14) Soderholm, L.; Antonio, M. R.; Williams, C.; Wasserman, S. R. Anal. Chem. 1999, 71, 4622−4628. (15) Prenzler, P. D.; Bramley, R.; Downing, S. R.; Heath, G. A. Electrochem. Commun. 2000, 2, 516−521. (16) Klod, S.; Ziegs, F.; Dunsch, L. Anal. Chem. 2009, 81, 10262− 10267. (17) Kamat, P. V.; Barazzouk, S.; Hotchandani, S. Angew. Chem., Int. Ed. 2002, 41, 2764−2767. (18) Palacios, R. E.; Fan, F. R. F.; Bard, A. J.; Barbara, P. F. J. Am. Chem. Soc. 2006, 128, 9028−9029. (19) Brajter-Toth, A.; Dryhurst, G. J. Electroanal. Chem. 1981, 122, 205−213. (20) Lapkowski, M.; Zak, J.; Karon, K.; Marciniec, B.; Prukala, W. Electrochim. Acta 2011, 56, 4105−4111. (21) Balland, V.; Byrdin, M.; Eker, A. P. M.; Ahmad, M.; Brettel, K. J. Am. Chem. Soc. 2009, 131, 426−427. (22) Ping, Z.; Neugebauer, H.; Neckel, A. Electrochim. Acta 1996, 41, 767−772. (23) Burkhardt, S. E.; Conte, S.; Rodriguez-Calero, G. G.; Lowe, M. A.; Qian, H.; Zhou, W.; Gao, J.; Hennig, R. G.; Abruña, H. D. J. Mater. Chem. 2011, 21, 9553−9563. (24) Zanardi, C.; Terzi, F.; Pigani, L.; Heras, A.; Colina, A.; LopezPalacios, J.; Seeber, R. Electrochim. Acta 2008, 53, 3916−3923. (25) Tesarova, E.; Heras, A.; Colina, A.; Ruiz, V.; Svancara, I.; Vytras, K.; López-Palacios, J. Anal. Chim. Acta 2008, 608, 140−146. (26) Liu, X.; Zhang, Z.; Cheng, G.; Dong, S. Electroanalysis 2003, 15, 103−107. (27) Shtoyko, T.; Maghasi, A. T.; Richardson, J. N.; Seliskar, C. J.; Heineman, W. R. Anal. Chem. 2003, 75, 4585−4590. (28) Richardson, J. N.; Dyer, A. L.; Stegemiller, M. L.; Zudans, I.; Seliskar, C. J.; Heineman, W. R. Anal. Chem. 2002, 74, 3330−3335. (29) Tudorache, M.; Bala, C. Anal. Bioanal. Chem. 2007, 388, 565− 578. (30) Honeychurch, K. C.; Hart, J. P. TrAC, Trends Anal. Chem. 2003, 22, 456−469. (31) Hart, J. P.; Wring, S. A. TrAC, Trends Anal. Chem. 1997, 16, 89− 103. (32) Zhang, L.; Li, Y.; Zhang, L.; Li, D. W.; Karpuzov, D.; Long, Y. T. Int. J. Electrochem. Sci. 2011, 6, 819−829. (33) Li, D.; Li, D. W.; Fossey, J. S.; Long, Y. T. Anal. Chem. 2010, 82, 9299−9305. (34) Schroll, C. A.; Chatterjee, S.; Heineman, W. R.; Bryan, S. A. Electroanalysis 2012, 24, 1065−1070. (35) Schroll, C. A.; Chatterjee, S.; Heineman, W. R.; Bryan, S. A. Anal. Chem. 2011, 83, 4214−4219. (36) Venton, B. J.; Wightman, R. M. Anal. Chem. 2003, 75, 414A− 421A. (37) Zachek, M. K.; Takmakov, P.; Moody, B.; Wightman, R. M.; McCarty, G. S. Anal. Chem. 2009, 81, 6258−6265. (38) Hawley, M. D.; Tatawawadi, S. V.; Piekarski, S.; Adams, R. N. J. Am. Chem. Soc. 1967, 89, 447−450. (39) Li, Y.; Liu, M.; Xiang, C.; Xie, Q.; Yao, S. Thin Solid Films 2006, 497, 270−278. (40) Zhang, F.; Dryhurst, G. Bioorg. Chem. 1993, 21, 392−410. (41) Rein, F. N.; Rocha, R. C.; Toma, H. E. J. Inorg. Biochem. 2001, 85, 155−166. (42) Deputy, A.; Wu, H. P.; McCreery, R. L. J. Phys. Chem. 1990, 94, 3620−3624. 9153


Spectroelectrochemistry at Screen-Printed Electrodes: Determination

Recommend Documents