Pulse-Voltammetric Glucose Detection at Gold Junction Electrodes

Aug 5, 2010 - is operated with fixed bias at +0.5 V vs SCE (to eliminate capacitive .... 200 mg of KCN, 200 mg of K2HPO4, 200 mg of K2CO3 in 10 cm3 wa...
1 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 7063–7067

Letters to Analytical Chemistry Pulse-Voltammetric Glucose Detection at Gold Junction Electrodes Liza Rassaei and Frank Marken* Department of Chemistry, University of Bath, Bath BA2 7AY, United Kingdom A novel glucose sensing concept based on the localized change or “modulation” in pH within a symmetric gold-gold junction electrode is proposed. A paired gold-gold junction electrode (average gap size ca. 500 nm) is prepared by simultaneous bipotentiostatic electrodeposition of gold onto two closely spaced platinum disk electrodes. For glucose detection in neutral aqueous solution, the potential of the “pH-modulator” electrode is set to -1.5 V vs saturated calomel reference electrode (SCE) to locally increase the pH, and simultaneously, either cyclic voltammetry or square wave voltammetry experiments are conducted at the sensor electrode. A considerable improvement in the sensor electrode response is observed when a normal pulse voltammetry sequence is applied to the modulator electrode (to generate “hydroxide pulses”) and the glucose sensor electrode is operated with fixed bias at +0.5 V vs SCE (to eliminate capacitive charging currents). Preliminary data suggest good linearity for the glucose response in the medically relevant 1-10 mM concentration range (corresponding to 0.18-1.8 g L-1). Future electroanalytical applications of multidimensional pulse voltammetry in junction electrodes are discussed. Electrochemical glucose sensing is of high importance in pointof-care and clinical diagnostics,1 in the food industry, in fermentation processes, and in biotechnology.2 The first and second generation of commercial glucose sensors are based on immobilization of glucose oxidase enzyme at the electrode surface.3 Mediators are employed for electron transfer shuttling between glucose oxidase and the electrode surface. Direct electron transfer between enzyme and electrode is introduced in third generation glucose sensors.4 The selectivity of these sensors is excellent; however, problems in enzymatic glucose sensors exist such as

avoiding the denaturation of immobilized enzymes and the resulting decrease in reproducibility5 during storage and during decontamination. Therefore, simple nonenzymatic glucose sensors are of considerable interest, but all conventional electro-catalysts such as boron-doped diamond,6 silver,7 or gold8 require alkaline aqueous conditions during measurement, which so far have been inappropriate for most glucose sensor applications. In this report, alkaline conditions are generated locally within a gold-gold junction electrode with submicrometer gap. The two adjacent gold electrodes may be regarded as “generator” and “collector” or as shown in this study as “modulator” and “sensor” electrodes. In the past, electrochemical generatorcollector systems, first introduced by Frumkin and Nekrasov,9 have been developed on the basis of rotating ring-disk electrodes,10 interdigitated microarrays,11,12 flow channels,13 or adjacent or paired microelectrodes.14 The use of two independent electrodes in close proximity and with bipotentiostatic potential control allows reaction intermediates to be investigated,15 analytical processes to be enhanced,16,17 surfaces to be imaged,18 ion transport across liquid/liquid boundaries to be studied,19 and novel analytical processes, for example, based (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)

* To whom correspondence should be addressed. E-mail: F.Marken@ bath.ac.uk. (1) Heller, A.; Feldman, B. Chem. Rev. 2008, 108, 2482. (2) Scully, P. J.; Betancor, L.; Bolyo, J.; Dzyadevych, S.; Guisan, J. M.; Fernandez-Lafuente, R.; Jaffrezic-Renault, N.; Kuncova, G.; Matejec, V.; O’Kennedy, B.; Podrazky, O.; Rose, K.; Sasek, L.; Young, J. S. Measur. Sci. Technol. 2007, 18, 3177. (3) Bartlett, P. N.; Cooper, J. M. J. Electroanal. Chem. 1993, 362, 1. (4) Vaddiraju, S.; Tomazos, I.; Burgess, D. J.; Jain, F. C.; Papadimitrakopoulos, F. Biosens. Bioelectron. 2010, 25, 1553. 10.1021/ac101303s  2010 American Chemical Society Published on Web 08/05/2010

(16) (17) (18) (19)

Liu, H. Y.; Hu, N. F. Electoanalysis 2007, 19, 884. Zhao, J. W.; Wu, L. Z.; Zhi, J. F. Analyst 2009, 134, 794. Quan, H.; Park, S. U.; Park, J. Electrochim. Acta 2010, 55, 2232. Pasta, M.; Ruffo, R.; Falletta, E.; Mari, C. M.; Della Pina, C. Gold Bull. 2010, 43, 57. Frumkin, A. N.; Nekrasov, L. I. Dokl. Akad. Nauk SSSR 1959, 126, 115. Pleskov, Yu. V.; Filinovskii, V. Yu. the Rotating Disc Electrode; Plenum Press: New York, 1976. Aoki, K. Electroanalysis 1993, 5, 627. Goluch, E. D.; Wolfrum, B.; Singh, P. S.; Zevenbergen, M. A. G.; Lemay, S. G. Anal. Bioanal. Chem. 2009, 394, 447. Thompson, M.; Klymenko, E. V.; Compton, R. G. J. Electroanal. Chem. 2005, 576, 333. Menshykau, D.; Del Campo, F. J.; Mun ˜oz, F. X.; Compton, R. G. Sens. Actuators, B 2009, 138, 362. Nei, L. B.; Marken, F.; Hong, Q.; Compton, R. G. J. Electrochem. Soc. 1997, 144, 3019. Paixao, T. R. L. C.; Richter, E. M.; Brito-Neto, J. G. A.; Bertotti, M. Electrochem. Commun. 2006, 8, 9. Zevenbergen, M. A. G.; Wolfrum, B. L.; Goluch, E. D.; Singh, P. S.; Lemay, S. G. J. Am. Chem. Soc. 2008, 131, 11471. Liljeroth, P.; Johans, C.; Slevin, C. J.; Quinn, B. M.; Kontturi, K. Electrochem. Commun. 2002, 4, 67. Vagin, M. Y.; Karyakin, A. A.; Vuorema, A.; Sillanpa¨a¨, M.; Meadows, H.; Del Campo, F. J.; Cortina-Puig, M.; Page, P. C. B.; Chan, Y. H.; Marken, F. Electrochem. Commun. 2010, 12, 455.

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

7063

Figure 1. Schematic drawing of a gold-gold sensor-modulator junction electrode with a negative potential applied to the modulator electrode in order to produce locally alkaline conditions in the micrometer interelectrode gap.

on “time-of-flight” measurements to be developed.20 Recently, gold-gold junction electrodes have been formed in a one-step electro-deposition process employing a bipotentiostatic gold deposition protocol with automated cutoff.21-23 The resulting gold-gold junction electrodes can be fabricated reproducibly with micrometer to submicrometer interelectrode gap24 and employed with a response time depending on the selected interelectrode gap. These junction electrodes now allow new types of electro-analytical procedures to be developed including methods where pulse sequences are applied independently to the two paired electrodes. Pulse sequences may be introduced to scan a potential range, to probe a time domain, or as demonstrated here, to systematically vary the pH in the interelectrode gap. In this study, a nonenzymatic glucose sensor for neutral media is proposed on the basis of a symmetric pair of adjacent gold electrodes. The interelectrode gap is approximately 500 nm.24 The electrochemical oxidation of glucose in aqueous media, which is possible at gold electrodes under bulk alkaline conditions,25 is demonstrated here in neutral solution. One electrode (the “modulator” electrode) is employed to generate hydroxide from water (see eq 1), and the second electrode (the “sensor” electrode) is employed to oxidize glucose to gluconic acid25 (see eq 2, Figure 1). 2H2O + 2 e- f H2 + 2 OH-

(1)

glucose + 2 OH- f gluconic acid + H2O + 2 e-

(2)

For small electrode junctions, the localized production of hydroxide is feasible when the diffusional transport of OH- across the interelectrode gap occurs rapidly on the time scale of a cyclic voltammetry or pulse voltammetry experiment. Preliminary data are presented to show that pulse voltammetry in junction electrodes will be feasible in robust miniaturized glucose sensors and of interest in a wider range of sensing problems. Amatore, C.; Sella, C.; Thouin, L. J. Electroanal. Chem. 2006, 593, 194. French, R. W.; Collins, A. M.; Marken, F. Electroanalysis 2008, 20, 2403. French, R. W.; Marken, F. J. Solid State Electrochem. 2009, 13, 609. Rassaei, L.; French, R. W.; Compton, R. G.; Marken, F. Analyst 2009, 134, 887. (24) French, R. W.; Gordeev, S. N.; Raithby, P. R.; Marken, F. J. Electroanal. Chem. 2009, 632, 206. (25) Ghanem, M. A.; Compton, R. G.; Coles, B. A.; Canals, A.; Vuorema, A.; John, P.; Marken, F. Phys. Chem. Chem. Phys. 2005, 7, 3552. (20) (21) (22) (23)

7064

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

EXPERIMENTAL SECTION Chemical Reagents. Potassium hydroxide, potassium gold(I) dicyanide, potassium cyanide, potassium dihydrogen phosphate, dipotassium hydrogen phosphate, potassium carbonate, poly(diallyldimehtylammonium chloride) or PDDAC, potassium chloride, and D-glucose were obtained from Sigma-Aldrich and used without further purification. Demineralized and filtered water was taken from a Thermo Scientific water purification system (Barnstead Nanopure) with not less than 18 MΩ cm resistivity. Argon (BOC) was employed to deaerate electrolyte solutions. Experiments were conducted at 20 ± 2 °C. Instrumentation and Procedures. A PGSTAT12 biopotentiostat system (Autolab, EcoChemie, The Netherlands) with GPES and NOVA software options was employed for electrochemical measurements. The NOVA software allows bipotentiostatic pulse voltammetry experiments to be designed for paired electrode systems. A conventional four-electrode cell with platinum counter and saturated calomel reference electrode (SCE, Radiometer, Copenhagen) and two working electrodes was employed. The gold-gold junction electrodes were grown by bipotentiostatic electro-deposition following a literature procedure.21 Briefly, two platinum wires (100 µm diameter) were sealed into glass with approximately 100 µm separation. The electrode surface was polished/renewed with an aqueous slurry of 1 µm of alumina (Buehler). The electro-plating solution was based on an alkaline cyanide bath (120 mg of KAu(CN)2, 200 mg of KCN, 200 mg of K2HPO4, 200 mg of K2CO3 in 10 cm3 water with 60 µL of 0.35 wt % PDDAC) and held at ca. 58 °C. Bipotentiostatic growth was carried out in a one-step process until automatic cutoff using the chronoamperometric method (GPES) at a potential of -0.88 V vs SCE and -0.8805 V vs SCE for the two electrodes. The resulting gold-gold junction electrodes are produced with an approximate gap size of ca. 0.5 µm. Caution: Experiments with this alkaline cyanide gold plating bath should always be carried out under a fume cupboard and handling/disposal of solutions should follow strict rules for work with cyanide solutions. RESULTS AND DISCUSSION Glucose Detection at Gold-Gold Junction Electrodes I: Effect of Applied Modulator Potential. Electro-oxidation of glucose at gold26 and gold alloys27 is a well-known reaction and has been studied widely in alkaline media. However, in most applied glucose sensing scenarios, alkaline pH conditions are not an option, and therefore, little progress has been made with the direct use of gold in glucose sensors. The availability of gold-gold junction sensors changes this situation. It is demonstrated here that, in a junction with interelectrode gap of ca. 0.5 µm, hydroxide can be generated locally and glucose oxidation performed as soon as the concentration of hydroxide in the interelectrode gap is sufficiently high. Figure 2A shows a set of typical voltammetric responses recorded for the gold sensor electrode when the gold modulator electrode in the junction is biased to negative potentials. The oxidation of 10 mM glucose in aqueous 0.1 M KCl is monitored, (26) Cherevko, S.; Chung, C. H. Sens. Actuators, B 2009, 142, 216. (27) Yi, Q. F.; Yu, W. Q.; Niu, F. J. Electroanalysis 2010, 22, 556.

Figure 2. (A) Cyclic voltammograms (scan rate of 20 mVs-1) for oxidation of 10 mM glucose in aqueous 0.1 M KCl obtained at a gold sensor electrode when the modulator potential was -1.1, -1.5, and -1.7 V vs SCE. (B) Plot of the main glucose oxidation peak current (P1, observed during the negative going scan) versus the modulator potential. (C) Plot of the main glucose oxidation peak current (P1, observed during the negative going scan, modulator potential of -1.5 V vs SCE) versus scan rate. (D) Plot of the main glucose oxidation peak current (P1, 20 mVs-1, modulator potential of -1.5 V vs SCE) versus glucose concentration.

and the potential of the sensor electrode is scanned from -0.3 V to +0.8 V vs SCE at a scan rate of 20 mV s-1. At modulator potentials of -1.1 V vs SCE and more negative, the oxidation of glucose appears as a sensor current on both the forward and backward potential scan. As the modulator potential is biased more negative, the oxidation responses increase. At -1.7 V vs SCE, a well-defined oxidation peak (see P1 at ca. 0.4 V vs SCE) is observed on the scan toward negative potentials (in agreement with literature reports28). The plot in Figure 2B confirms the increase in the peak current P1 with more negative potential applied to the modulator. The current for the formation of hydroxide at the modulator electrode is expected to increase exponentially when going to more negative potentials, and the resulting increase in local hydroxide concentration at the sensor electrode can explain the observed increase in peak current P1. The scan rate applied during the glucose oxidation does affect the peak current for process P1 but due to the complex junction geometry and the predominantly kinetic reaction control only in a complex manner (see Figure 2C). Good voltammetric responses were obtained with a scan rate of 20 mVs-1. After optimization of the voltammetric signal and employing a modulator potential of -1.5 V vs SCE, the concentration of glucose (for a range of 1-20 mM, see Figure 2D) can be shown to be approximately linearly related to the peak current P1. Glucose Detection at Gold-Gold Junction Electrodes II: Square Wave Voltammetry with Applied Modulator Potential. In order to explore the sensitivity enhancements possible with pulse voltammetry in a gold-gold junction electrode, experiments were conducted with a fixed bias applied to the modulator electrode and with the sensor electrode in square wave voltammetry mode scanning in positive and in negative scan direction. (28) Kurniawan, F.; Tsakova, V.; Mirsky, V. M. Electroanalysis 2006, 18, 1937.

Figure 3. (A) Schematic drawing of the potential applied to the sensor electrode S and the modulator electrode M in square wave voltammetry mode. The gray line shows the data collection periods for the forward, F, and the backward, B, currents. The square wave voltammetry current response is calculated as the difference IF - IB.29 (B) Experimental square wave voltammograms with the modulator electrode generating hydroxide (potential fixed at -1.5 V vs SCE) for the oxidation of glucose (concentration indicated in the plot) in 0.1 M KCl at a gold sensor electrode (square wave voltammetry mode with step potential of 2 mV, amplitude of 20 mV, frequency of 8 Hz, scan rate of 16 mVs-1, and scan from -0.2 to +0.8 V vs SCE). (C) The same experiment with opposite scan direction (step potential of -2 mV and scan from +0.8 to -0.2 V vs SCE).

Figure 3A shows a schematic drawing of the square wave voltammetry potential program (applied to the sensor electrode S) with a small potential step superimposed on a bigger potential staircase. The gray zones indicate measurements under forward (F) and backward (B) bias, and the resulting square wave voltammogram is obtained as the difference in forward and backward current. This methodology is sensitive to reversible electrode processes, and it discriminates against Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

7065

charging currents.30 For the chemically irreversible glucose oxidation process and with a fixed modulator bias at -1.5 V vs SCE, complex voltammetric responses as a function of glucose concentration are obtained (see Figure 3B,C). The square wave voltammetry data for positive and negative scan direction are very different, and sensitivity toward lower glucose levels is observed in particular for the negative scan direction (see Figure 3C). At more negative modulator bias and in particular for the positive scan direction, this methodology is plagued by high noise levels originating probably from the gas evolution (H2 evolution) during hydroxide formation. A constant bias applied to the modulator electrode is likely to lead to gas bubble formation, and this can upset the measurement at the sensor electrode. Therefore, a better glucose sensing protocol may be developed on the basis of a pulsed modulator potential (pulsed hydroxide formation) and a fixed sensor bias. Glucose Detection at Gold-Gold Junction Electrodes III: Normal Pulse Modulation of Hydroxide Levels with Constant Sensor Potential. Next, it is possible to employ only short pulses of hydroxide generated at the modulator electrode. This methodology can be realized in the normal pulse voltammetry mode when negative potential pulses are applied to the modulator electrode and the sensor electrode potential is fixed at +0.5 V vs SCE for the detection of glucose. This method benefits from the absence of capacitive current components in the sensor current and from the minimization of interference from hydrogen evolution. Figure 4A shows a schematic drawing of the measurement conditions where pulses to more and more negative potentials are applied to the modulator electrode (M) while the sensor electrode bias (S) remains at the glucose oxidation potential. Only the sensor current is of interest in this case, and a typical set of plots of sensor currents is shown in Figure 4B. A clear progression of analytically useful current responses is observed when the concentration of glucose is gradually increased. A plot of the peak current versus glucose concentration is linear (see Figure 4C, analytical equation Ip/nA ) 1.37 * [glucose]/mM + 0.03, with R2 ) 0.98) over a glucose concentration range of 1-10 mM. This concentration range is medically relevant. Blood glucose levels are typically in the 3-10 mM range (corresponding to 0.5-1.8 g L-1) and need monitoring, for example, in diabetic patients. The origin of the background current observed during sensor-modulator experiments is intriguing. The mechanism for the background current must be Faradaic in nature (the sensor electrode potential is constant) and likely to be associated with oxidation of hydrogen (which is formed at the modulator electrode). The presence of glucose in the solution appears to suppress this background current as well as result in a distinct peak which is shifting to more negative potentials as the glucose concentration is increased. The mechanism for both background and peak processes need further investigation. The new modulator-sensor methodology based on traditional pulse methods applied to a junction electrode system is highly versatile and applicable to a wider range of sensing (29) Southampton Electrochemistry Group, Instrumental Methods in Electrochemistry; Horwood Publishing Ltd.: Chichester, 2001, p 72. (30) Lovric, M. In Electroanalytical Methods; Scholz, F. Ed.; Springer: Berlin, 2005; p 111.

7066

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

Figure 4. (A) Schematic drawing of the potential applied to the sensor electrode S and the modulator electrode M in normal pulse voltammetry mode. The gray line shows the data collection time. (B) Experimental data for the oxidation of glucose (glucose concentration (i) 1 mM, (ii) 2 mM, (iii) 4 mM, and (iv) 8 mM in aqueous 0.1 M KCl) at a sensor electrode (+0.5 V vs SCE) with the modulator electrode scanning in normal pulse voltammetry mode (potential step of -5 mV, pulse time of 70 ms, and scan rate of 10 mVs-1). (C) Plot of the peak current versus glucose concentration.

problems. The multiparameter nature of the experiment allows “multi-dimensional” voltammograms to be obtained and optimized. Traditional pulse voltammetric methods can be applied for the modulator or for the sensor electrode, and in the future, new types of customized pulse sequences will be desirable for the junction sensor electrode. The determination of glucose has been selected here to highlight the potential and importance of this experimental tool and further applications will include, for example, determining organohalides,31 ammonia,32 (31) Wiyaratn, W.; Somasundrum, M.; Surareungchai, W. Anal. Chem. 2004, 76, 859. (32) Takahashi, M.; Nakamura, K.; Jin, J. Electroanalysis 2008, 20, 2205.

buffer capacity, or pKA values in aqueous media. The miniaturization of junction sensors and the optimization of the interelectrode gap will be of considerable practical importance in electroanalysis. CONCLUSIONS A robust, simple, and sensitive electrochemical sensor concept for nonenzymatic glucose monitoring is proposed on the basis of locally changing the pH within the interelectrode gap of a gold-gold junction electrode. The “modulator” electrode is employed to generate hydroxide at negative applied potentials, and the “sensor” electrode allows oxidation of glucose to be observed. Although the data presented in this report are preliminary in nature, it can be concluded that the most promising results are observed when a normal pulse

sequence is employed for the modulator electrode and a fixed bias is applied to the sensor electrode. The multiparameter nature of this kind of voltammetric experiment in junction electrodes allows future 2D- or 3D-voltammetric experiments to be designed and new approaches to be developed for analytical problems such as low analyte concentration or interferences. ACKNOWLEDGMENT L.R. thanks the EPSRC (EP/F025726/1) for financial support of this work. Received for review May 19, 2010. Accepted August 2, 2010. AC101303S

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

7067