Trends Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 Impedance methods for electrochemical sensors using nanomaterials Ian I. Suni This article presents an overview of electrochemical sensors that employ nanomaterials and utilize electrochemical impedance spectroscopy for analyte detection. The most widely utilized nanomaterials in impedance sensors are gold (Au) nanoparticles and carbon nanotubes (CNTs). Au nanoparticles have been employed in impedance sensors to form electrodes from nanoparticle ensembles and to amplify impedance signals by forming nanoparticle-biomolecule conjugates in the solution phase. CNTs have been employed for impedance sensors within composite electrodes and as nanoelectrode arrays. The advantages of nanomaterials in impedance sensors include increased sensor surface area, electrical conductivity and connectivity, chemical accessibility and electrocatalysis. ª 2008 Elsevier Ltd. All rights reserved. Keywords: Analyte detection; Biosensor; Carbon nanotube; Electrochemical impedance spectroscopy; Electrochemical sensor; Gold nanoparticle; Immunosensor; Impedance; Nanoparticle; Nanomaterial Ian I. Suni* Department of Chemical and Biomolecular Engineering, Center for Advanced Materials Processing (CAMP), Clarkson University, Potsdam, NY 13699-5705, USA * Tel.: +1 315 269 4471; E-mail: isuni@clarkson.edu 1. Introduction 1.1. Electrochemical impedance spectroscopy – background Electrochemical impedance spectroscopy (EIS) has long been employed for studying electrochemical systems [1], including those involved in corrosion, electrodeposition [2], batteries [3] and fuel cells [4]. For impedance measurements, a small sinusoidal AC voltage probe (typically 2–10 mV) is applied, and the current response is determined. The in-phase current response determines the real (resistive) component of the impedance, while the out-of-phase current response determines the imaginary (capacitive) component. The AC probe voltage should be small enough so that the system response is linear, allowing simple equivalent circuit analysis. Impedance methods are quite powerful, in that they are capable of characterizing physicochemical processes of widely differing time constants, sampling electron transfer at high 0165-9936/$ - see front matter ª 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.trac.2008.03.012 frequency and mass transfer at low frequency. Impedance results are commonly fitted to equivalent circuits of resistors and capacitors, such as the Randles circuit shown in Fig. 1 [5], which is often used to interpret simple electrochemical systems. This equivalent circuit yields the Nyquist plot shown in Fig. 2, which provides visual insight into the system dynamics. In Figs. 1 and 2, Rct is the charge-transfer resistance, which is inversely proportional to the rate of electron transfer; Cd is the double-layer capacitance; Rs is the solution-phase resistance; and, Zw is the Warburg impedance, which arises from mass-transfer limitations. If an analyte affects one or more of these equivalent circuit parameters and these parameters are not affected by interfering species, then impedance methods can be used for analyte detection. Rs arises primarily from the electrolyte resistance and is analytically useful mainly in conductivity sensors, which I will not discuss in this article. The Warburg impedance, which can be used to measure effective diffusion coefficients, is seldom useful for analytical applications. The equivalent circuit elements in Figs. 1 and 2 that are most often useful for analyte detection are Rct and Cd. The measured capacitance usually arises from the series combination of several elements, such as analyte binding (Canal) to a sensing layer (Csens) on an Au electrode (CAu). In this case, the measured capacitance is: 1 1 1 1 ¼ þ þ Cd CAu Csens Canal ð1Þ for a sensing layer and analyte layer that are continuous. In many cases, the 604 Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 Trends Rct ¼ RAu þ Rsens þ Ranal Cd Rs Zw Rct Figure 1. Randles equivalent circuit for a simple electrochemical system. -Z imag Slope=unity g sin w a cre de Rs R ct Z real Figure 2. Nyquist plot arising from the Randles circuit shown in Fig. 1. capacitance at the Au electrode-sensing layer interface is large and can be neglected. The sensitivity is then determined by the relative capacitance of the analyte layer and the sensing layer. For each dielectric layer, the capacitance per unit area depends on the layer thickness (t) according to: C ed ¼ A t ð2Þ where ed is the dielectric constant of the dielectric layer, so capacitance is most sensitive to binding of large analytes, such as proteins or cells. One difficulty with capacitive sensors is that their sensitivity depends on obtaining the proper thickness of the original sensing layer [6]. If the original sensing layer is too thin, then the underlying electrode surface may be partially exposed, allowing for non-specific interactions from interfering species. However, if the original sensing layer is too thick, then the AC impedance current that is detected is dramatically reduced, as is the change in capacitance upon analyte binding. Rct can also be quite sensitive to analyte binding, particularly for detection of large species, such as proteins or cells, which significantly impede electron transfer. For analyte binding (Ranal) to a sensing layer (Rsens) on an Au electrode (RAu), the measured resistance is: ð3Þ The resistance at the interface between the Au electrode and the sensing layer is typically negligible. Measurement of Rct requires the presence of redox-active species in the electrolyte. Impedance sensing is most useful for large species that significantly perturb the sensing interface, although impedance detection of glucose was recently reported [7]. Many of the examples of impedance sensors that I discuss later in this article monitor Rct as a measure of analyte concentration. 1.2. Electrochemical impedance spectroscopy – sensing applications For biosensors, EIS has some important advantages over amperometry. For direct amperometric biosensors, an oxidoreductase enzyme is immobilized at a conductive electrode, and electron transfer is detected during a biologically-mediated oxidation/reduction reaction. However, the active site must be both in close proximity to the electrode surface and easily accessible to the analyte solution. In many cases, electron transfer occurs far from the electrode surface, and electron-transfer rates drop exponentially with distance [8]. This problem can be reduced through the use of redox mediators, but detection then becomes limited by mediator mass transfer. Indirect amperometric biosensors detect the product of a biologically-catalyzed reaction, often hydrogen peroxide. However, the analyte often contains additional species (e.g., ureate or ascorbate) that can also be electrochemically oxidized or reduced, so indirect amperometric biosensors are not selective. One of the most significant advantages of impedance detection for biosensing is that antibody-antigen binding can be directly detected, allowing the development of immunosensors. The main drawback of impedance methods for biosensors is the need for interfacial engineering to reduce non-specific adsorption. One well-studied method to minimize non-specific interactions is to embed the probe agent into a composite film that contains the biomolecule of interest interspersed with a protein-resistant species, such as molecules containing ethylene-glycol moieties. This approach has been widely touted by the research group of George Whitesides [9–11], and such reagents are now commercially available. The use of impedance methods for biosensors has been recently reviewed [12], but not with a focus on the use of nanomaterials. Limits of detection (LODs) have been reported for impedance biosensors in the nM–pM range in controlled laboratory conditions [13–17]. It should be acknowledged that Au-nanoparticle conjugation to biomolecules has been employed in biosensors using several other electrochemical detection methods, predominantly anodic stripping voltammetry (ASV), anodic Au surface oxidation and quartz crystal http://www.elsevier.com/locate/trac 605 Trends Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 microbalance (QCM) [18,19]. However, impedance detection has some significant advantages over these methods. Impedance sensing does not require the voltage scanning needed for ASV and anodic oxidation, which is time consuming and may degrade the electrochemical interface during wide potential sweeps. In addition, impedance methods are largely insensitive to environmental disturbance, which is often problematic for QCM sensors. 2. Nanomaterials for sensing applications Nanomaterials are generally defined as involving the length scale from 1–100 nm; in other words, materials intermediate between the atomic and molecular scale and the bulk scale. The chemical, electronic, and optical properties of nanomaterials generally depend on both their dimensions and their morphology. Although a wide variety of nanomaterials for sensors have been reported in the literature, the most widely employed nanomaterials are carbon nanotubes (CNTs) and Au nanoparticles, in part because of their commercial availability. In addition, both materials are considered to be biocompatible. 2.1. Au nanoparticles Au nanoparticles are generally synthesized by chemical reduction of Au salts in aqueous-phase, organic-phase, or mixed-phase solutions [20]. The most difficult aspect of this synthesis is to control the reactivity of the Aunanoparticle surface during particle growth, since the surface energy is quite high. Controlled synthesis of Au nanoparticles requires the use of stabilizing agents, such as citrate or thiolated species, that bind to the particle surface to control growth and to prevent aggregation. Numerous methods have been reported for creation of biomolecule-Au-nanoparticle conjugates either during or after Au-nanoparticle synthesis [20]. Commercial reagents are now available for conjugation of biomolecules to Au nanoparticles of several different sizes. One of the primary reasons for the intensive research into biomolecule-Au-nanoparticle conjugates is that biomolecules in this environment are generally stable and retain their biological activity. Depending on the application, different Au-nanoparticle sizes may be optimal [21]. 2.2. Carbon nanotubes CNTs, which are allotropes of carbon from the fullerene structural family, can be conceived as sp2 carbon atoms arranged in grapheme sheets that have been rolled up into hollow tubes. Multi-walled CNTs (MWCNTs) behave as conductors and have electrical conductivities greater than metals, suggesting their incorporation into sensing electrodes may be beneficial. However, depending on the tube diameter and chirality, single-walled CNTs (SWCNTs) can 606 http://www.elsevier.com/locate/trac behave electronically as either metals or semiconductors [22], complicating their use in sensing electrodes. CNTsynthesis methods create a mixture that includes amorphous carbon, graphite particles and CNTs, so synthesis is typically followed by a difficult separation process. For electrochemical applications, CNTs are typically activated in strong acids, which opens the CNT ends and forms oxygenated species, making the ends hydrophilic and increasing the aqueous solubility of CNTs [22]. The electrochemical behavior of CNTs varies considerably with the methods used for purification and preparation, including oxidation treatment [22]. For analytical applications, CNTs are most often used to modify other electrode materials, or as part of a composite electrode, in part due to difficulties in handling them. 3. Impedance sensors using Au nanoparticles 3.1. Au-nanoparticle substrates – impedance detection The most widely reported use of Au nanoparticles in impedance sensors involves their incorporation into an ensemble substrate onto which a protein, oligonucleotide, or other probe molecule is immobilized [23–34]. Most studies involve construction of a sensing interface that contains one layer of Au nanoparticles on a conductive electrode, although, in a few cases, Au nanoparticles are incorporated into a ceramic sol gel or polymer film. The Au nanoparticles are sometimes made using colloidal techniques, and sometimes by electrodeposition. The Au nanoparticles can be conjugated with probe reagents (antibodies or ssDNA) either before or after the Au-nanoparticle ensemble is formed. The advantages of sensing interfaces that contain Aunanoparticle networks, compared to sensing interfaces based on flat Au surfaces, include the increased surface area for sensing, improved electrical connectivity through the Au-nanoparticle network, and chemical accessibility to the analyte through these networks. The advantages, compared to non-Au surfaces, also include electrocatalysis. One potentially powerful method for using Au nanoparticles to enhance impedance detection in biosensors involves the construction of three-dimensional networks with Au nanoparticles dispersed throughout the sensing interface. This can be accomplished through repeated use of a bifunctional reagent, such as cysteineamine or 4-aminothiophenol, where the thiol group can bind to a biomolecule and the amine group can bind to Au nanoparticles, for layer-by-layer formation of an Aunanoparticle network. Impedance detection of human immunoglobulin (hIgG) using such a three-dimensional Au-nanoparticle network was recently reported using 6nm diameter Au nanoparticles and cysteamine as the bifunctional reagent [23]. Fig. 3 shows the sensorpreparation process. Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 Trends Figure 3. Au-nanoparticle-multilayer preparation onto an Au electrode (a), and the immobilization of antibody and the interaction of antigen and biotin-conjugated antibody (b) (from [23]). These authors also studied the nature of the sensing interface as a function of the number of Au-nanoparticle layers using both cyclic voltammetry and EIS. As the number of Au-nanoparticle layers increased, the 3=4 FeðCNÞ6 oxidation/reduction peak height increased and the peak separation decreased, demonstrating increased reversibility. Similarly, Rct decreased continuously as the number of Au-nanoparticle layers increased. Following electrostatic binding of goat anti-human IgG antibody, the sensing interface was able to detect the presence of hIgG, with Rct increasing with an increase in human-IgG concentration. Amplification of the impedance signal was accomplished by further binding biotinconjugated goat anti-human IgG, resulting in a detection range of 5–400 lg/L. The LOD was then estimated to be 0.5 lg/L [23]. In this study, antibody was immobilized only on the outer layer of Au nanoparticles to ensure chemical accessibility of the analyte, a protein. For small-molecule analytes that can be detected by impedance methods, such multilayer Au-nanoparticle net- works may be invaluable for sensing. They could allow dramatic increases in the electrode surface area without introducing mass-transfer limitation. Impedance detection of carcinoembryonic antigen (CEA), a glycoprotein involved in cell adhesion produced only during fetal development, was recently reported [31]. The CEA antibody was first bound through its surface amino groups to glutathione-modified Au nanoparticles of diameter 15 ± 1.5 nm by amide-bond formation using N-(3-dimethylaminopropyl)-N 0 -ethylcarbodiimide hydrochloride (EDC) and N-hydroxysulfylsuccinimide sodium salt (NHSS). The sensing interface was then formed by co-polymerizing a mixture of o-aminophenol and the Au-nanoparticle-conjugated CEA antibodies. An interesting feature of their study was the direct comparison between the antibody-containing sensing interface, with and without Au-nanoparticle conjugation. They reported that Rct increased by only 0.59 · 105 X (35%) for the sensing interface without Au nanoparticles, but by 6.3 · 105 X (7%) with Au http://www.elsevier.com/locate/trac 607 Trends Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 nanoparticles [31]. The authors tested their sensing interface in both model lysozyme solutions and serum samples and reported no false positives arising from non-specific interactions. They estimated an LOD for CEA of 0.1 ng/mL. Impedance detection was recently demonstrated for an intriguing application, detection of the IgE antibody to a protein allergen from dust mites [29,30]. Au nanoparticles were deposited onto a glassy-carbon electrode (GCE) either by electrodeposition, or by immersion in (3-mercaptopropyl)trimethoxysilane (MPTS), followed by immersion in a colloid solution containing 16-nm diameter Au nanoparticles. For Au electrodeposition, 30 s of deposition from 0.1% HAuCl4 produced Au nanoparticles of average diameter 40 ± 8 nm. The Aunanoparticle-modified GCE was then immersed in recombinant dust-mite allergen Der f2 to form a protein film, and this interface was employed for impedance detection of the murine monoclonal antibody to Der F2 over the range 2–300 lg/ml. At relatively low antibody concentrations, Rct increased continuously with antibody concentration. At higher antibody concentrations, Rct became relatively insensitive to changes in antibody concentration, probably due to surface saturation. This type of impedance sensor might be employed for allergy screening of patients, where allergen-specific IgE is detected for a wide range of allergens. In addition to using antibodies as probe reagents, impedance sensors have been demonstrated using DNA or oligonucleotides bound to Au-nanoparticle arrays to detect complementary target molecules [26,34]. In addition, the incorporation of CdS nanoparticles conjugated to ssDNA into the sensing interface of an impedance sensor has been reported [35]. One group reports forming a sensing interface by binding thiol-derivatized oligonucleotides onto Au surfaces modified by Au electrodeposition, followed by impedance detection of two minor DNA groove-binding agents, mythramycin and netropsin, and a DNA intercalator, nogalamycin [26]. The advantage of using Au electrodeposition to modify the Au substrate is that the effects of surface roughness, which is related to the Au-nanoparticle size, can be studied quantitatively by measuring the surface area by voltammetric reduction of Au oxide. Substrates were prepared with a total surface area up to 90% greater than that of the original flat Au substrate. The greatest sensitivity was observed for an Au-electrodeposition process that produced Au nanoparticles in the 20–80nm range. The authors estimated that this allowed a reduction in the LOD by a factor of 20–40x, down to 5 nM for nogalamycin [26]. Au nanoparticles and carbon nanofibers have also been reported to be useful in composite substrates for impedance sensing of cells [36,37]. In these studies using EIS, the binding of K562 leukemia cells was monitored as an increase in Rct. These authors reported 608 http://www.elsevier.com/locate/trac that incorporating Au nanoparticles increased the sensitivity to cell binding, which was attributed to increased electrode-surface area. Au nanoparticles were first synthesized using chitosan as a combined reducing and stabilizing agent, then reacted with ammonia to create a sol-gel film atop a GCE with embedded Au nanoparticles of 12-nm diameter. Adhesion of K562 leukemia cells was then monitored in situ by EIS. Cell adhesion could be detected only by the combination of chitosan and Au nanoparticles atop a GCE. Rct was reported to correlate to the logarithm of the cell concentration over the range 104–108 cells/mL with an LOD of 8.7 · 102 cells/mL. 3.2. Au-nanoparticle conjugation in solution Several recent studies described different strategies for the use of Au nanoparticles for impedance sensing that involved Au-nanoparticle conjugation in the solution phase rather than prior modification of the sensing interface. In one approach, impedance sensing included an extra step of analyte conjugation to 10-nm diameter Au nanoparticles, with signal amplification occurring only when the Au nanoparticles become embedded in the sensing interface [38]. This approach was demonstrated using the model system fluorescein/anti-fluorescein, with fluorescein bound to the flat Au substrate using EDC/NHSS linker chemistry. The analyte (goat antifluorescein) was conjugated to Au nanoparticles in solution prior to detection. A change in the impedance at the sensing interface was observed only when the antibody was conjugated to Au nanoparticles, but not for the bare antibody [38]. Signal amplification was significantly higher with a redox probe (impedance detection) than without a redox probe (capacitance detection). This is believed to reflect the substantial electrochemistry that can occur on the Au nanoparticles embedded within the sensing interface, which is otherwise essentially a polymer film. As a result, Rct is substantially reduced upon analyte binding, which embeds Au nanoparticles within the sensing interface. A similar detection scheme was recently reported to detect DNA hybridization, with the target ssDNA conjugated to 5-nm diameter CdS nanoparticles [39]. Probe ssDNA was immobilized onto an Au electrode using selfassembly chemistry and amide-bond formation with EDC/NHSS coupling. CdS nanoparticles were prepared by precipitation from CdCl2 and Na2S using mercaptoacetic acid as a stabilizer, then conjugated to the complementary ssDNA. The authors reported that conjugation to CdS nanoparticles increased the sensitivity by about two orders of magnitude. Interestingly, unlike the results observed with Au-nanoparticle conjugation [38], here analyte binding was accompanied by a dramatic increase in Rct [39]. The difference between these two studies can be explained by the different rates of electron transfer on Au and CdS, and by the different sensing Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 interfaces. For CdS-nanoparticle conjugation, the interfacial Rct prior to analyte detection was about two orders of magnitude lower than that in the study with Aunanoparticle conjugation. For this less well-passivated sensing interface, the dominant effect upon binding of ssDNA-CdS is obscuration of the underlying conductive electrode, rather than enhanced rates of electron transfer due to embedding of CdS nanoparticles. However, when Au nanoparticles are embedded into a sensing interface that is completely polymer coated, the dominant effect is the improved rates of electron transfer on the Au nanoparticles. In biosensors, the use of nanomaterials has been envisioned to create successive amplification steps [40]. This type of approach was recently demonstrated with a different type of solution-phase Au-nanoparticle conjugation, utilizing a strategy that might be termed an impedance-sandwich assay [41]. In this approach, antiprotein A IgG was bound to an Au-electrode surface, and then exposed to protein A of varying concentrations. Following protein A binding, the sensing interface was exposed to a solution containing IgG antibodies conjugated to 13-nm diameter Au nanoparticles. Without this amplification step, the LOD of protein A was reported to be 1.0 ng/mL, and the LOD was reduced by one order of Trends magnitude by the amplification step. The authors reported that their sensitivity was about 100x better than that obtained with conventional ELISAs. One advantage of this approach is that the protein-antibody conjugate can be prepared in advance and stored for about one month without loss of activity. Another group recently reported the use of solutionphase Au-nanoparticle conjugation for amplifying the signal from an impedance biosensor. The sensing interface was an Au electrode onto which Au nanoparticles were attached using 1,6-hexanedithiol, followed by immobilization of rabbit anti-IgG [28]. After binding the hIgG analyte, and blocking non-reacted surface sites with bovine serum albumin (BSA), the impedance signal was amplified by binding Au-colloid-labeled goat antihIgG that was synthesized in advance [28]. This approach (Fig. 4) was motivated by the relatively small impedance change sometimes observed upon antigen recognition by a surface-immobilized antibody. Without amplification, the impedance change upon binding of hIgG was about 100 X-cm2, whereas, with amplification, the impedance change was several thousand X-cm2. The authors reported an LOD for human IgG of 4.1 ng/L and a linear concentration range of about 15–330 ng/L. Figure 4. The process of immobilization of rabbit anti-hIgG antibody onto an Au electrode, followed by analyte binding and amplification by the Au-nanoparticle-labeled antibody (from [28]). http://www.elsevier.com/locate/trac 609 Trends Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 4. Impedance sensors using carbon nanotubes 4.1. Carbon-nanotube substrates – impedance detection The most detailed studies of impedance sensors that employ CNTs do not employ SWCNTs or MWCNTs, but instead employ electrodes constructed from CNT towers grown by chemical-vapor deposition (CVD) [42–44]. Starting with bare Si wafers, Al was deposited by electron-beam evaporation and then oxidized, followed by deposition of a Fe-catalyst film through a shadow mask. CNT towers several mm thick were then grown by CVD at 750C from a mixture of ethylene, water, and hydrogen. The CNT tower was peeled from the Si substrate, cast in epoxy, and polished to reveal the underlying CNTs. The average CNT diameter is 20 nm, the average spacing is about 200 nm, and the aspect ratio is approximately 2 · 105. A significant advantage of this method for creating an electrochemical-sensing interface is that purification of the CNTs is not needed. The electrochemical characteristics of these CNTtower electrodes have been most fully characterized by voltammetry. Voltammetry of CNT towers in both FeðCNÞ63=4 and RuðNH3 Þ3þ show a sigmoidal shape, 6 without clear current peaks, at scan rates of up to 100 mV/s, and show current peaks for scan rates of 500 mV/s and above [42,43]. These results are similar to results for MWCNT arrays that exhibit sigmoidal voltammograms for large nanotube spacing, where the diffusion fields from individual nanotubes do not fully overlap, and peak-shaped voltammograms for small nanotube spacing, where diffusion fields overlap [45,46]. As has been widely reported for micro-electrodes [47], arrays of nanotube electrodes have enhanced diffusion rates relative to macroscopic electrodes, and reduced capacitance per unit area, which can significantly improve their sensitivity. Given the high electron-transfer rates observed, CNT towers might be useful for characterizing rapid redox processes [42]. CNT-tower electrodes have been employed for impedance detection of both mouse IgG and prostatecancer cells [43,44]. Prior to immobilization of antimouse IgG, the open end of the CNTs were oxidized in strong acid or electrochemically to form active carboxylate groups [43]. This allowed the use of standard EDC/ NHSS coupling chemistry for amide-bond formation to anti-mouse IgG. Both antibody immobilization and analyte binding were monitored by the extent to which they increased Rct, providing a non-linear calibration curve. The LOD for mouse IgG was reported as 200 ng/mL, with a dynamic range of up to 100 lg/mL. Preliminary results for impedance detection of prostrate-cancer cells involved somewhat more complex electrode preparation, including Au electrodeposition onto the CNT-tower electrode [44]. As for protein detection, cell binding is detected as an increase in Rct. 610 http://www.elsevier.com/locate/trac CNTs have also been incorporated into composite electrodes used for impedance detection of DNA hybridization [48,49]. In these studies, MWCNTs were copolymerized with polypyrrole atop a GCE. EDC/NHSS linker chemistry was used to form an amide bond and immobilize ssDNA. The complementary oligonucleotide could be detected by the accompanying change in Rct, both with [48] and without [49] subsequent metallization. DNA metallization is a widely-studied technique, whereby metal ions that bind to the center of the DNA double helix greatly increase the conductivity of the sensing interface, and that could be detected as a reduction in Rct [48]. However, DNA hybridization without metallization could be detected as an increase in Rct [49]. CNTs were incorporated within the sensing interface due to their high conductivity and their effect of increasing the active surface area. 5. Future outlook EIS has been widely used to study a variety of other electrochemical systems, including corrosion, electrodeposition, batteries and fuel cells. However, only recently have impedance methods been applied in the field of biosensors. Given their ability to sense Rct and Cd, application should be possible for several different types of sensing schemes, with numerous recognition agents. Electrochemical impedance sensors are particularly promising for portable and implantable applications. Commercialization will depend on improvements in several different areas, including minimization of the effects of non-specific adsorption. References [1] A. Lasia, Electrochemical impedance spectroscopy and its applications, in: B.E. Conway, J. OÕM. Bockris, R. White (Editors), Modern Aspects of Electrochemistry, vol. 32, Plenum Press, New York, USA, 1999, p. 143. [2] R. Wiart, Electrochim. Acta 35 (1990) 1587. [3] F. Huet, J. Power Sources 70 (1998) 59. [4] C.Y. Yuh, J.R. Selman, AIChE J. 34 (2004) 1949. [5] J.E.B. Randles, Discuss. Faraday Soc. 1 (1947) 11. [6] S.Q. Hu, Z.Y. Wu, Y.M. Zhou, Z.X. Cao, G.L. Shen, R.Q. Yu, Anal. Chim. Acta 458 (2002) 297. [7] J. Wang, K.A. Carmon, L.A. Luck, I.I. Suni, Electrochem. Solidstate Lett. 8 (2005) H61. [8] F.A. Armstrong, G.S. Wilson, Electrochim. Acta 45 (2000) 2623. [9] J. Lahiri, L. Isaacs, J. Tien, G.M. Whitesides, Anal. Chem. 71 (1999) 777. [10] E. Ostuni, R.G. Chapman, R.E. Holmlin, S. Takayama, G.M. Whitesides, Langmuir 17 (2001) 5605. [11] X. Qian, S.J. Metallo, I.S. Choi, H. Wu, M.N. Liang, G.M. Whitesides, Anal. Chem. 74 (2002) 1805. [12] E. Katz, I. Willner, Electroanalysis (N.Y.) 15 (2003) 913. [13] C. Berggren, G. Johansson, Anal. Chem. 69 (1997) 3651. Trends in Analytical Chemistry, Vol. 27, No. 7, 2008 [14] V.M. Mirsky, M. Riepl, O.S. Wolfbeis, Biosens. Bioelectron. 12 (1997) 977. [15] M. Zayats, O.A. Raitman, V.I. Chegel, A.B. Kharitonov, I. Willner, Anal. Chem. 74 (2002) 4763. [16] F. Lucarelli, G. Marrazza, M. Mascini, Biosens. Bioelectron. 20 (2005) 2001. [17] H. Cai, T.M.H. Lee, I.M. Hsing, Sens. Actuators, B 114 (2006) 433. [18] J. Wang, Anal. Chim. Acta 500 (2003) 247. [19] W. Fritzsche, T.A. Tatton, Nanotechnology 14 (2003) R63. [20] S. Guo, E. Wang, Anal. Chim. Acta 598 (2007) 181. [21] J.F. Hainfield, R.D. Powell, J. Histochem. Cytochem. 48 (2000) 471. [22] J.J. Gooding, Electrochim. Acta 50 (2005) 3049. [23] M. Wang, L. Wang, H. Yuan, X. Ji, C. Sun, L. Ma, Y. Bai, T. Li, J. Li, Electroanalysis (N.Y.) 16 (2004) 757. [24] M. Wang, L. Wang, G. Wang, X. Ji, Y. Bai, T. Li, S. Gong, J. Li, Biosens. Bioelectron. 19 (2004) 575. [25] D. Tang, R. Yuan, Y. Chai, J. Dai, X. Zhong, Y. Liu, Bioelectrochemistry 65 (2004) 15. [26] C.Z. Li, J.H.T. Luong, Anal. Chem. 77 (2005) 478. [27] Z.S. Wu, J.S. Li, M.H. Luo, G.L. Shen, R.Q. Yu, Anal. Chim. Acta 528 (2005) 235. [28] H. Chen, J.H. Jiang, Y. Huang, T. Deng, J.S. Li, G.L. Shen, R.Q. Yu, Sens. Actuators, B 117 (2006) 211. [29] H. Huang, Z. Liu, X. Yang, Anal. Biochem. 356 (2006) 208. [30] H. Huang, P. Ran, Z. Liu, Bioelectrochemistry 70 (2007) 257. [31] H. Tang, J. Chen, L. Nie, Y. Kuang, S. Yao, Biosens. Bioelectron. 22 (2007) 1061. [32] I. Szymanska, H. Radecka, J. Radecki, R. Kaliszan, Biosens. Bioelectron. 22 (2007) 1955. Trends [33] S. Zhang, F. Huang, B. Liu, J. Ding, X. Xu, J. Kong, Talanta 71 (2007) 874. [34] J. Yang, T. Yang, Y. Feng, K. Jiao, Anal. Biochem. 365 (2007) 24. [35] H. Peng, C. Soeller, M.B. Camnell, G.A. Bowmaker, R.P. Cooney, J. Travas-Sejdic, Biosens. Bioelectron. 21 (2006) 1727. [36] C. Hao, L. Ding, X. Zhang, H. Ju, Anal. Chem. 79 (2007) 4442. [37] L. Ding, C. Hao, Y. Xue, H. Ju, Biomacromolecules 8 (2007) 1341. [38] J. Wang, J.A. Proffitt, M.J. Pugia, I.I. Suni, Anal. Chem. 78 (2006) 1769. [39] Y. Xu, H. Cai, P.G. He, Y.Z. Fang, Electroanalysis (N.Y.) 16 (2004) 150. [40] J. Wang, Small 1 (2005) 1036. [41] M. Li, Y.C. Lin, K.C. Su, Y.T. Wang, T.C. Chang, H.P. Lin, Sens. Actuators, B 117 (2006) 451. [42] Y.H. Yun, V.N. Shanov, M.J. Shulz, Z. Dong, A. Jazieh, W.R. Heineman, H.B. Halsall, D.K.Y. Wong, A. Bunge, Y. Tu, S. Subramanian, Sens. Actuators, B 120 (2006) 298. [43] Y.H. Yun, A. Bunge, W.R. Heineman, H.B. Halsall, V.N. Shanov, Z. Dong, S. Pixley, M. Behbehani, A. Jazieh, D.K.Y. Wong, A. Bhattacharya, M.J. Shulz, Sens. Actuators, B 123 (2007) 177. [44] Y.H. Yun, Z. Dong, V.N. Shanov, M.J. Shulz, Nanotechnology 18 (2007) 465505. [45] J. Li, H.T. Ng, A. Cassell, W. Fan, H. Chen, Q. Ye, J. Koehne, J. Han, M. Meyyapan, Nano Lett. 3 (2003) 597. [46] J. Koehne, J. Li, A.M. Cassell, H. Chen, Q. Ye, H.T. Ng, J. Han, M. Meyyapan, J. Mater. Chem. 14 (2004) 676. [47] A.M. Bond, Analyst (Cambridge, U.K.) 119 (1994) 1R. [48] Y. Xu, Y. Jiang, H. Cai, P.G. He, Y.Z. Fang, Anal. Chim. Acta 516 (2004) 19. [49] Y. Xu, X. Ye, L. Yang, P.G. He, Y.Z. Fang, Electroanalysis (N.Y.) 18 (2006) 1471. http://www.elsevier.com/locate/trac 611