Electrode geometry plays a critical role in electrochemical sensors, and therefore multiple sensor designs incorporating the three electrode cell with different electrode geometries were explored. Photographs illustrating the evolution of the sensor layout from the first to the last design are shown in . In each case the sensor consisted of a three-electrode system, with a Bi working electrode, an Ag/AgCl reference electrode, and an Au auxiliary electrode inside a micro-channel. The desired potential was applied between the working and the reference electrodes. The auxiliary electrode provided the current required to sustain electrolysis at the working electrode. The three electrode system is advantageous because it prevents the reference electrode from driving the current which could change its potential.
Fig. 3 Photographs of the earlier (a) and the later (b) generation of microfluidic sensor for heavy metal determination. Panels (c) and (d) illustrate close-up images of the sensors. AE auxiliary gold electrode, RE Ag/AgCl reference electrode, WE working Bi (more ...)
Optimization of electrode geometry with respect to size and layout was especially important since generation of gas bubbles at the Au auxiliary electrode due to electrolysis of water was observed in the first generation sensor when applying negative potentials (<−1.5 V) in attempts to detect highly electronegative metals. shows an anodic stripping voltammogram of our early attempt to simultaneously detect four heavy metals (Mn, Zn, Cd and Pb) contained in a single solution. At the first glance, it appears the sensor is working as four distinct peaks are clearly apparent. However, while the stripping peaks of Cd and Pb are unaffected, the highly electronegative Mn shows heavy influence of solvent electrolysis and the accompanying bubble formation at the auxiliary electrode disrupting the path for charge flow in solution between the auxiliary and working electrodes. This caused the Mn stripping peak to broaden and to be of poor quality, preventing any quantification. In these early experiments, excess gas production resulted in poor sensor stability and reproducibility. Gas bubbles that formed on the auxiliary electrode surface led to surface masking and thus stability issues. Bubbles that dislodged and drifted to the working electrode caused disruptions in metal concentrations near the electrode surface resulting in poor reproducibility. Excessive electrolysis also disrupted the solution connection between the three electrodes in the small channel of the microfluidics system, causing high resistance and loss of potential control of the working electrode giving erratic behavior of the sensor system. In a number of severe cases, loss of the auxiliary electrode was observed due to excessive and aggressive bubble formation that caused Au film adhesion loss and flaking.
Fig. 4 ASV measurement of Mn, Zn, Cd and Pb at concentrations of approximately 20 μM, 15 μM, 9 μM, 5 μM, respectively. Analytes were pre-concentrated at −1.8 V for 600 s on an electrodeposited bismuth working electrode (more ...)
The conventional strategy for minimizing hydrolysis is to limit the potential scan to above −1.5 V, which minimizes the current in the sensor and the amount of electrolysis at the auxiliary electrode. However, this approach is not suitable for detection of highly electronegative metals such as Mn, which strips at approximately −1.47 V (vs. 3 M KCl Ag/AgCl), as the potential must be much more negative during the metal pre-concentration step. Our most successful strategy was to increase the area of the auxiliary electrode and the distance between the working and the auxiliary electrodes. This is clearly evident from , which illustrates the final design with a 20× larger auxiliary electrode area from the initial design (10 mm2 vs. 0.5 mm2). Increasing the area of the auxiliary electrode while maintaining the area of the working electrode reduced the current density required to sustain the electrochemical reaction at the auxiliary electrode. In essence, while the same gas production occurred, the larger area made the gas bubbles less concentrated and therefore reduced their effect on the sensor. The distance between the working and auxiliary electrode was increased from 400 μm in the earlier designs to 1.65 mm for the final design. By increasing the distance between the auxiliary and working electrode, less interference due to gas production was observed due to the increased distance the bubbles must traverse to affect the working electrode.
A stable reference electrode is an important component of an integrated electrochemical system and is a prerequisite for achieving reliable performance. The key advantages of using Ag/AgCl reference electrodes are the desirable electrochemical characteristics and process compatibility. The Ag/AgCl electrode is based on a very high exchange-current density reaction, which means that at low current densities the electrode is not polarized and the potential at the electrode/electrolyte interface is a function of Cl−
activity only. One factor known to limit stability of the Ag/AgCl electrodes is the well-known solubility of AgCl in solutions. This problem, although trivial for millimeter scale electrodes, can become critical for thin-film electrodes where only a few hundred nanometers of AgCl are present. This problem was overcome by depositing a relatively thick 350 nm layer of silver chloride. We found this thickness to be sufficient for repeated sensor operation over a 1-week period (over 4 h of continuous operation). For longer lifetimes, thicker layers of Ag and AgCl can be deposited if necessary (thicknesses exceeding 1 μm have been reported (Bousse et al. 1986
The working electrode plays an essential role in ASV sensors because this is where the metals are detected by pre-concentration and stripping. In this work, bismuth was used as the working electrode material due to its demonstrated suitability for stripping voltammetry of electronegative metals and substantially reduced toxicity over the standard mercury electrode (Jorge et al. 2007
; Krolicka et al. 2003
; Pauliukaite et al. 2004
; Wang et al. 2000
). The working potential window of bismuth is limited by the potential required for oxidation of Bi0
, which leads to an upper limit for the bismuth electrode of approximately −0.3 V. On the negative side, the potential window is flanked by the onset of hydrolysis, at approximately −1.7 V. This negative working potential window is ideally suited to detect strongly electronegative metals such as Mn and Zn (stripping peak at approximately −1.35 V), as well as other electronegative metals such as Cd (stripping peak at approximately −0.85 V) and Pb (stripping peak at approximately −0.55 V). An added benefit of using bismuth is that its films can be electro-deposited on a number of conductive surfaces, making it compatible with standard lab-on-a-chip fabrication methods. The low toxicity of bismuth allows for the development of environmentally-friendly disposable devices(Rodilla et al. 1998
; Sun et al. 1999
), which can be utilized for a variety of point-of-care applications.
In stripping analysis, the working electrode surface plays a critical role as it influences the ability of the analyte to be stripped and therefore detected. Thus, working electrode surface roughness (or rather surface smoothness) is important and was investigated. We compared controlled-current and controlled-potential methods of Bi deposition and found the latter to be the better procedure for preparing the electrode. A controlled −0.8 V potential worked best as compared to the controlled current at 5 mA/cm2 for 4 min of Bi film deposition. compares the surfaces of the deposited Bi using the two methods with a plain Au surface. During electrodeposition, a Bi film covers the Au seed layer first and then starts to grow grain clusters. AFM analysis indicated that the formation of the grain clusters was much more dense for the controlled potential–deposition, with surface roughness of εRMS ~ 230 ± 49 nm, as compared with the controlled-current condition at 5 mA/cm2 (εRMS ~ 610 ± 78 nm). The current density increased constantly during the controlled potential deposition and was often higher than the current density used for the controlled-current deposition. Thus the controlled-potential method not only had increased surface roughness and thus overall surface area, but also exhibited the ability to deposit thicker films for the time intervals selected. An electrode-position rate of ~51 nm/min for the controlled-potential (vs. ~34 nm/min for the controlled-current condition) was measured for the Bi electrodeposition cell (), which offers excellent control of the deposited film thickness.
Fig. 5 Comparison of Bi films electrodeposited by controlled-current (5 mA/cm2) and controlled-potential (−0.8 V). SEM images of the (a) Au seed layer, (b) controlled current Bi film, and c controlled potential Bi film illustrate surface quality. AFM (more ...)
Ultimately, the surface properties of the working electrode affect the working potential window of the electrochemical sensor and the onset of hydrolysis. Cyclic voltammograms of Bi films deposited by the controlled-current and controlled-potential methods in acetate buffer (pH 5.75) are compared in . The Bi film formed by controlled-current exhibited extensive hydrolysis with a potential window ranging from −1.7 V to −0.3 V. The hydrolysis was reduced substantially when the Bi electrode was formed by controlled-potential, with a much wider negative potential window ranging from −1.9 V to −0.3 V. This is a critically important result for detection of highly electronegative metals, as the favorable shift of the hydrolysis away from the detection region reduces the background noise. Thus, owing to the larger surface area and the wider potential window of the potential-controlled Bi film, the measured peak currents of the anodic stripping analysis proved to be higher, with sharper and smoother stripping peaks. Thus, the controlled-potential method of Bi deposition is clearly superior and was selected for the results presented herein.
Fig. 6 Cyclic voltammetry comparison of the working potential windows of the Bi working electrodes prepared by controlled-current (5 mA/cm2) and controlled-potential (−0.8 V). Dotted lines indicate extension of the working potential window for the Bi (more ...)
The final sensor design exhibited a much improved detection capability, as illustrated by results presented in . The voltammogram in was the measurement obtained from a solution containing both Pb and Cd ions. At first glance, the stripping waveforms of Cd (9 μM) and Pb (5 μM) appear to be the same as the results obtained with the earlier sensor design in . However, the measured peak currents and quality have improved. Specifically, for the same metal concentration in the sample and 600 s pre-concentration time, the peak current for Pb increased tenfold from ~3.35 μA in the earlier design to ~9 μA, while the peak full width at half maximum (FWHM) remained the same ~88 mV. Similarly, the stripping waveform for Cd increased in amplitude from ~2.39 μA to ~4 μA for a 9 μM sample (a ~2× increase in signal for the same concentration), while the peak FWHM increased slightly from ~73 mV to ~100 mV.
Fig. 7 Measurements of metal ions with the final sensor design. (a) ASV waveforms illustrating measurement of Cd (9 μM) and Pb (5 μM) metals. (b) Comparison of ASV waveforms for measurement of Mn with the final and initial designs. Metal concentrations (more ...)
Most importantly, however, the stripping waveform of Mn showed a remarkable improvement. compares the stripping peaks we obtained with our initial and final designs at similar Mn concentrations (20 μM vs
. 5 μM). The most obvious improvement is the smooth, noise-free shape of the peak measured with the final design, which can now be quantified. Another distinct improvement is the amplitude of the current change—in the initial design the waveform shows a ~5 μA current change for the 20 μMMn sample, while the final design yields a ~5 μA stripping peak but for the 5 μMMn sample. The peaks became sharper as well, with the FWHM decreasing from ~228 mV in the initial design to ~145 mV in the final design. From these results it is clear that the final sensor design is capable of detecting highly electronegative metals such as Mn and offers superior measurement capabilities for even mildly electronegative metals such as Pb and Cd. These results also show the sensor is capable of multi-analyte detection in the low μM concentration range, which is the relevant range for assessment of heavy metal exposure in physiological samples such as blood or urine (in a healthy adult blood levels of Pb, Zn and Mn are typicall at or below 20 μg/dL (1 μM), 90 μg/dL (18 μM), and 50 μg/L (1 μM), respectively(Ellingsen et al. 2006
; Goullé et al. 2005
; Hotz et al. 2003
To demonstrate the ability of the sensor to measure metals in biological samples, Pb and Cd were measured in blood serum. For this, blood was drawn from a healthy adult male into a metal free, sterile, Royal Blue Top tube (BD Vacutainer) and centrifuged (Eppendorf 5810R) at 4,000 rpm for 8 min at 35 C. Serum was taken off with a pipette into a tube, diluted in a 1:5 ratio with 100 mM sodium acetate buffer (pH 5.75), and spiked with 5 μM of Pb and 9 μM of Cd. shows a representative voltammogram, illustrating the stripping peaks for both metals. The peaks shifted slightly in potential, approximately +100 mV, which is attributed to the increased Cl−
concentration in the serum sample. It is important to note that the stripping peaks in serum samples yielded peak currents comparable to the peak currents obtained in buffer only (), although the background was slightly higher. This difference in background could be due a number of factors, including biofouling of the bismuth electrode (as some protein deposition was observed at the working electrode). Nevertheless, the initial 3~5 measurements with each sensor yielded reproducible results, which is more than adequate for disposable devices. If sensor operation for a longer period is necessary, strategies will need to be developed to address biofouling. One possible approach that has proven to be successful is to coat electrodes with a Nafion film, which is permeable to metal ions (Hurst and Bruland 2005
Fig. 8 ASV waveform illustrating measurement of Cd (9 μM) and Pb (5 μM) metals in blood serum. The sample was diluted with 0.1 M sodium acetate buffer (pH 5.75) in 1:5 ratio. Analytes were pre-concentrated at −1.8 V for 600 s on an electrodeposited (more ...)
The sensor exhibited high reproducibility over multiple days, with hours of continuous operation. The measurement reported in is an average from three separate experiments on the same sample, with the standard deviations of 1.36 μA and 0.47 μA for Pb and Cd respectively. Similarly smallvariabilitywasobserved in blood serum, with an average from four separate experiments on the same sample exhibiting standard deviations of 1.72 μA and 0.34 μA for Pb and Cd respectively (). To further assess reproducibility, multiple sensors were used to perform repeated measurements of 5 μM Pb and 9 μM Cd in 0.1 M sodium acetate buffer (pH 5.75) over a 5 day period (n=24). Following each measurement, sensors were rinsed with DI water, blow-dried with nitrogen, and stored in ambient air until then next measurement. Representative results are shown in (for clarity, data for only two sensors for days 1, 3, and 5 are shown). The mean peak currents for each metal were 9.6 μA for Pb and 3.6 μA for Cd, with standard deviations of 1.37 μA and 1.26 μA respectively. Continued sensor use beyond 5 days, up to 36 sequential measurements, resulted in signal drift and ultimately sensor failure, most commonly due to the eventual loss of the auxiliary electrode to hydrolysis. Nevertheless, considering that the sensor is ultimately intended to be disposable, failure after repeated use is not a performance issue. Overall, these results indicate that sensors performed consistently, regardless of being used immediately following fabrication or being stored prior to or between measurements.
Fig. 9 Representative results of repeated measurements of Cd (9 μM) and Pb (5 μM) metals in sodium acetate buffer over a 5 day period by two different sensors. Between measurements sensors were cleaned and stored in ambient air. All analytes (more ...)
The described sensor is the first lab-on-a-chip capable of measuring highly electronegative Mn, and its design has several distinct advantages. First, the modifications in the configuration of the electrodes delayed the onset of hydrolysis and permitted detection of highly electronegative metals. The quality of the stripping peaks improved, with stripping waveforms exhibiting increase in peak current, peak sharpness, and elimination of peak noise due to hydrolysis. Second, the bismuth electrodes offer an environmentally-friendly solution to the conventionally used mercury electrode. The other commonly used electrodes such as Au, Pt and carbon do not have sufficiently wide negative potential windows to be used with Mn and other highly electronegative metals. Third, electrodepositing bismuth in situ (inside microfluidic chip) offers a renewed fresh sensing surface for repeated analyses or for multi-metal analysis. This also increases reliability and repeatability of the sensor for repeated analyses as well as the durability and lifetime of the sensor increased as Bi films can be reproduced repeatedly.