Introduction

From nanopore battery arrays to diamond nanoelectrode arrays, advances in nanoelectrochemistry have been both rapid and extensive due to the beneficial properties gained from reducing electrode size to the nanoscale. [1–4] Recent reviews have covered the current state of nanoelectrochemistry. [3, 5–7] Nanoelectrodes are of particular interest due to properties such as decreased charging currents and increased spatial resolution. In recent years, research in nanoelectrochemistry has begun to address highly multiplexed measurements [8], electrochemistry in nanoscopic volumes [9] and nanoporous electrodes. [10] These interests intersect with the development of nanoelectrode arrays, particularly nanopore-electrode based arrays and ensembles. Interest in nanoelectrode arrays began with the work of Martin and coworkers, who developed nanoelectrode ensembles based on the nanoporous structure of track-etched membranes. Collections (arrays, ensembles) of nanoelectrodes such as these are of interest for many reasons; they exhibit macroscopic electrode area, and they support large currents, while retaining the unique mass transport features that accrue to nanoscale electrodes.

In this review, consistent with general use in the literature, array will be used to describe spatially ordered collections of nanoelectrodes, whereas ensemble will be used to describe random or stochastic collections of nanoelectrodes. Arrays and ensembles of electrodes are of particular interest due to the way that their properties depend on geometry, size, spacing, etc. [11–19] In particular, mass transport can be drastically affected by these properties. For example, the behavior of nanoelectrode arrays varies with electrode size, interelectrode distance, timescale, and diffusion zone overlap, the latter determining whether the array behaves as one large electrode or each nanoelectrode experiences enhanced radial diffusion. The size and geometry of the electrode also present interesting opportunities to control mass transport by manipulating electrostatic screening and coupling to electrokinetic flow.

A subset of nanoelectrodes are those that are integrated with solid-state nanopores, a class of structures which has demonstrated a versatile set of capabilities based on manipulating mass transport, ion transfer, and multiplexing with spectroscopic measurements. This particularly active area of current nanoelectrode array research, especially arrays and ensembles of nanopore electrodes, will be the primary focus of this review. In nanopore electrode arrays, electrodes are embedded in nanochannels or nanopores oriented either vertically or horizontally, depending on the intended end use. Nanopore electrode arrays exhibit interesting and useful properties, especially for applications in electroanalytical chemistry, e.g. redox cycling, and electrochemical processing, e.g. cascade reactions.

Fabrication

In current practice, solid-state nanopore electrode arrays are fabricated using either serial, e.g. direct-write electron beam lithography (EBL), ion beam milling, [20–24] or parallel, e.g. lithography/etching-based methods, [25–27] in contrast to early nanoelectrode arrays that were typically developed from adventitious substrates, such as porous alumina membranes. [28–31] While powerful, these fabrication methods are expensive and demand exacting conditions. To circumvent these challenges, a great deal of recent research has targeted the development of cheap, fault-tolerant routes to nanopore electrode fabrication. A common method involves colloidal (nanosphere) lithography, as depicted in Fig. 1. [32] In this approach, organized self-assembled layers of microparticles (typically polystyrene) are used as a mask to pattern the electrode array. While simple and robust, more complex structures typically demand advanced fabrication approaches. Figure 2 shows an example of multiple electrode layers fabricated into nanopore electrode arrays with pore sizes <20 nm in diameter, [33] using integrated circuit fabrication techniques. While the challenges and approaches to fabrication could justify its own review, the remainder of the review will focus on applications of nanoelectrode arrays - particularly redox cycling, coupled spectroscopy-electrochemistry, and biosensing.

Fig. 1
figure 1

Schematic illustration of the fabrication of a nanotemplate by nanosphere or colloidal lithography. a A monolayer of polystyrene (PS) microbeads formed at the substrate surface is subjected to an O2 plasma etch to reduce the size of the microbeads without changing their position; b A masking layer is applied, here by plasma-enhanced chemical vapor deposition of SiOx; c The microbeads are removed by mechanical or chemical lift-off, after which the remaining SiOx layer may be used as a process mask for further steps. Reprinted with permission from Anal. Chem. 2006, 78, 7588. Copyright (2006) American Chemical Society

Full size image
Fig. 2
figure 2

a Schematic illustration of pore-first fabrication of 3-metal layers embedded in a nanopore membrane by reactive ion etching (RIE); top - substrate with nitride layers on top and bottom sides; 2nd patterned electrodes; 3rd - nanopore formation by RIE; 4th - protection layers encapsulating the nanopore; 5th - formation of the membrane by removal of the Si substrate. b Plan view SEM image of the region consisting of three metal layers - M3, M2, and M1, top to bottom. c Cross-sectional TEM image along the black line in (b) showing the uniformity of the layers over 400 nm. d Zoom-in of the red rectangular region in (c) shows the precise (±5 nm) control of layer thickness. e TEM image of a single nanopore formed by RIE. f SEM image showing an array of nanopores fabricated by RIE; average nanopore diameter ~ 18 ± 2 nm. g Integration of functional nanopores on an 8 in. wafer. Reproduced from Nanoscale 2014, 6, 8900. with permission of The Royal Society of Chemistry

Full size image

Chemical and biochemical sensors

The properties of nanoelectrode arrays have naturally been exploited in developing new types of sensors with lower limits of detection (LOD). Sensors to address biomolecules have been targeted to glutamate [34], glucose [35–38], H2O2 [39–43], DNA, prostate specific antigen [44], dopamine [45] and lipids [46] as well as to study enzyme reactivity [36, 47]. Carbon nanofiber arrays are among the most popular electrode structures for these types of measurements due to the ease with which the surface chemistry can be tuned. [34, 45, 48, 49] For example, sensors capable of DNA detection have been fabricated as nanoelectrode arrays with DNA-selective surface functionalization. In these structures, the interaction of DNA with surface molecules limits diffusion or ion migration resulting in a decrease in electrochemical current. [50] Alternatively, monitoring ion flux though arrays of nanochannels has permitted DNA detection down to 0.1 nM. [51] Progressively lower detection limits of small metabolites, e.g. H2O2, down to 10−9 M, have also been achieved with nanoelectrode arrays, which exhibit a linear response over a large concentration range as well. [39–43] In addition to using faradaic electrochemistry, other properties of nanoelectrode arrays can be exploited for biomolecular analysis. For example, Fig. 3 shows the change in resistance of a carbon nanofiber array as a function of cardiac troponin-I concentration. Label free carbon fiber nanoelectrode arrays have demonstrated LODs (0.2 ng/mL) lower than conventional ELISA. [52]

Fig. 3
figure 3

Change in electron transfer resistance, ΔR et , for different concentrations of human cardiac troponin-I (Δ), a specific target protein analyte, and human myoglobin (□), an unrelated nonspecific analyte for the immunosensor electrode. Reprinted with permission from Anal. Chem. 2013, 85, 3858. Copyright (2013) American Chemical Society

Full size image

On the non-biological side, environmental monitoring of water pollutants has been achieved with membrane-based electrochemical nanoarray sensors. Differential pulse anodic stripping is able to resolve both Pb and Cu while maintaining excellent limits of detection (0.4 μM). Furthermore, these sensors are able to quantify Cu and formaldehyde, common water source pollutants, directly in environmental samples without further clean up. [53] Similarly, carbon nanotube electrode arrays have been exploited to detect heavy metals such Cd and Pb at levels <1 ppb. [54].

Nanochannel arrays with embedded nanoelectrodes

Embedded nanoelectrodes in nanochannels are nanopore devices that are particularly interesting for lab-on-a-chip applications. Arrays on elongated nanopore (or nanochannel) arrays with embedded annular nanoband electrodes (EANE) can be fabricated with layer-by-layer deposition, followed by FIB milling. [55] Images of a typical device are shown in Fig. 4(a-c), while Fig. 4(d-e) shows SEM cross sections of the array. This class of nanopore array exhibits unique behavior in which mass transport is coupled to electrochemical processing. Alternating layers of Au and polymer permit voltage across the separate Au layers to drive electrokinetic flow, which increases the analyte delivery to the electrode surface, while the same small voltage can affect electron transfer at the EANE working electrode. By separating the EANE and a quasireference electrode, constituted by a separate metal layer (i.e. top layer of structure in Fig. 4d), in a nanopore by approximately 10 μm, electric fields, E ~ 103 V cm−1 can be generated from ~1 V potentials and has been shown to enhance the steady state current by as much as 50× compared to redox processes in the absence of electroosmotic flow (EOF). These array-based devices show promise for analyte processing. Due to the extremely short diffusion lengths, approximately 100 % conversion efficiency can be obtained at suitably large dimensionless electrode width, W, and Peclet number, Pe, defined as, [56]

$$ W=\frac{w}{R} $$
(1)
Fig. 4
figure 4

Embedded annular nanoband electrode arrays. a Reflected light micrograph of nine pore arrays; b SEM image of a single 11 × 11 pore array; c same array at increased magnification; d SEM image of a sacrificial sample at 15° tilt, cross-sectioned by FIB milling; e magnified view of the EANE at 52° tilt. Reprinted with permission from J. Am. Chem. Soc. 2012, 134, 8617. Copyright (2012) American Chemical Society

Full size image

and

$$ Pe=\frac{u_{avg}R}{D} $$
(2)

where w is the physical electrode width, R is the pore radius, u avg is the linear fluid velocity averaged over the radial dimension of the pore, and D is the diffusion coefficient of the redox-active species. When EOF is turned on in these structures, the electrochemical processing is rapid and nearly perfectly efficient, resulting in the collapse of the usual diffusion layer, since redox active species are reacted as fast as they can be delivered to the EANE. Clearly under these circumstances, the reaction velocity is limited by how fast reactants can be delivered by EOF, which, in turn, is limited by the magnitude of the driving potential. To decouple EOF from the potential driving electron transfer and achieve the same high overall reaction velocities, electrokinetic flow can be controlled with a second working electrode, located external to the nanopore, leading to current increase of ~30×. [57] Furthermore, this scheme can be used with separate functions assigned to the different electrodes. For example, the upstream electrode can be used for reagent generation, as demonstrated by the upstream in situ generation of H2 for downstream catalysis. [58].

Recessed nanodisk electrode arrays

The nanopore arrays discussed above are examples of 1D nanostructures, since there is one dimension, length, which is not nanoscale. Another interesting class is composed of 0D nanostructures, in which all three critical dimensions are nanoscale. An example of 0D nanostructures are recessed nanodisk electrode arrays, which consist of arrays non-conductive nanopores with a recessed metal disk at the bottom of each pore functioning as the electrode. These devices are of particular interest due to their unique mass transport properties, namely they provide the ability to study and exploit confinement. [59–63] Common strategies of fabrication of these types of structures include preparing them from tracked-etch membranes, processing diblock copolymers and using nanosphere lithography (vide supra). [31, 32, 64] Recessed nanopores with diameters down to 10 nm have been produced from track-etched membranes. [65] Arrays of nanopores can access a variety of diffusional regimes at different time scales, as illustrated schematically in Fig. 5. [66] At fast scan rates, peak shaped voltammograms are expected, as linear diffusion within the pores occurs without any overlap of diffusion zones between the pores. As the scan rate is decreased, sigmoidal voltammograms are observed, characteristic of radial diffusion to the pore, and as the scan rate is further decreased, diffusion zone overlap occurs between the nanopores, creating a linear diffusion regime, which again produces peak shaped voltammograms.

Fig. 5
figure 5

Schematic illustration showing the diffusion patterns at nanopore electrodes: a linear diffusion to planar electrode, b overlapping radial diffusion to porous nanoelectrodes, c radial diffusion to porous nanoelectrodes, and d linear-active diffusion to porous nanoelectrodes. Reprinted with permission from Langmuir 2001, 17, 6396. Copyright (2001) American Chemical Society

Full size image

At fast scan rates the peak current is given by, [67]

$$ {i}_p=0.446 nF\left(N\pi {a}^2\right)C\sqrt{\frac{nFD\nu}{RT}} $$
(3)

where N is the pore density, C is concentration, n is the number of electrons, D is the diffusion coefficient, F is Faraday’s constant, v is scan rate, T is temperature, a is pore radius, and R is the gas constant. Similarly, the limiting current at slow scan rates is given by, [17, 68]

$$ {i}_{\lim }=\frac{4{\uppi \mathrm{nFCDa}}^2N}{4L+\pi a} $$
(4)

where L is the pore length. Recessed nanodisk electrode arrays have numerous applications in sensing and especially in redox cycling. Emphasizing the difference in behavior relative to macroscopic planar electrodes, the small diameter of the pores and recessed nature of the electrodes make it possible to use charge screening to control mass transport to the electrode. Figure 6 shows the effect electrolyte concentration and pH have on electrostatic charge screening in arrays of 10 nm recessed nanopore electrodes. When the total electrolyte concentration is low, the opposed double layers overlap, and electrostatic interactions can either decrease or increase faradaic current, depending on the sign of the permselective behavior. At high pH, when the pore surface is negatively charged, ζ < 0, an increase in current is seen from cationic analytes and a decrease is observed with anionic analytes. The dimensions of the devices are also useful in size-exclusion applications. [61] Similar types of devices, prepared from porous anodic alumina membranes, have been used to simulate the effects of confinement on biochemical reaction kinetics, [36, 69] allowing the activity of confined enzymes to be characterized using nanoporous recessed electrodes.

Fig. 6
figure 6

pH dependence of the ratio i lim /C, comparing charged redox species, X = Fe(CN)6 3− (triangles) or trimethylaminoferrocene (TMAFc+, circles) to unchargedferrocenedimethanol, FcDM. i lim values were measured at 0.01 V s−1 on recessed nanopore electrodes (10 nm in pore diameter) in solutions containing 0.1 M (filled symbols) or 0.01 M (open symbols) KNO3 in addition to 0.01 M KH2PO4/K2HPO4. The ratios of the diffusion coefficients in 0.1 M KNO3 + 0.01 M KH2PO4/K2HPO4 (pH 6.3) are shown as dashed lines. Reproduced from Analyst 2010, 135, 172. with permission of The Royal Society of Chemistry

Full size image

Redox cycling in nanopore electrode arrays

Electrode density plays an important role in the design of electrode arrays for amperometric sensing, since it governs the mass transport behavior and determines the ratio of faradic to charging current. [66] In their pioneering work, Martin and coworkers used template synthesis to fabricate nanoelectrode ensembles, which exhibited significantly decreased charging current and enhanced sensitivity. [28] In a typical amperometric measurement, diffusion behavior depends strongly on the electrode density of the electrode array or ensemble, as illustrated in Fig. 5. In order to take advantage of the array size to achieve maximum sensitivity, an appropriate interelectrode distance should be chosen to achieve diffusional independence for individual electrodes. [67, 70] Further increasing electrode density leads to competition for analyte between adjacent electrodes due to the overlapping of diffusion zones. As a result, the limiting current decreases for an individual electrode, which can be attributed to a shielding effect and characterized by a shielding factor. [71, 72].

On the other hand, by building a structure in which it is possible to hold two adjacent electrodes negative and positive of the equilibrium potential, E eq , of an electroactive species, positive feedback and an increase in faradic current can be achieved. [73–77] In such a platform, electroactive species oxidized at one electrode can diffuse to, and be reduced at, the other electrode. The cycling of redox species between the two electrodes multiplies the electron transfer reaction, resulting in current amplification at both electrodes. This redox cycling (RC) effect, which is similar to the positive feedback mode in scanning electrochemical microscopy, [78] has been observed in interdigitated electrode arrays (IDEA), [73–77] thin-layer-cells (TLC), [79–81] and planar/ring-recessed disk electrode (P/R-RDE) arrays. [26, 82–88] Here, we focus on redox cycling at P/R-RDE arrays and present some unique properties that have not been observed in other electrode geometries.

Figure 7 gives an example of the fabrication of these electrode arrays. [26, 87] In a typical electrode geometry, two thin-film gold electrodes are separated by an insulating layer, such as a polymer, SiNx, or SiO2 film which determines the interelectrode distance. Reliable insulation can be achieved with polymer films, but only at μm thicknesses, [89] whereas SiNx and SiO2 films as thin as 100 nm can provide sufficient insulation to ensure independent electrode operation. [26, 88, 90] In order to provide solution access to the bottom electrode, different lithography methods have been employed for the mask to etch the top electrode and the insulating layer. Well-defined μm-scale pores can be generated photolithographically, [89] while e-beam and nanosphere lithography can produce pores down to ~100 nm. [26, 88, 90] Multistep reactive ion etching processes can then be performed using the lithographically defined mask to access various electrode areas. [26, 87, 88].

Fig. 7
figure 7

a Schematic cross section showing the fabrication procedure for an R-RDE array, including (i) layer-by-layer deposition, (ii) nanosphere lithography, and (iii) multistep reactive ion etching. The colors represent different layers: gray (glass slide), yellow (Au), red (SiNx), pink (SiO2), purple (polystyrene spheres), and green (Cr). b Schematic diagram of the ring–disk geometry of the array. c An SEM image of the array at 50° tilt. d An optical image of the array integrated with SiNx channels. e Schematic diagram of the macroscopic layout of the R-RDE array (dark blue) integrated with channels (black) and covered by a piece of PDMS (light blue) with two circular wells. Reprinted with permission from Anal. Chem. 2013, 85, 9882. Copyright (2013) American Chemical Society

Full size image

Current amplification in RC devices

The most obvious advantage of using RC is the enhancement of faradic current (Fig. 8), which can be evaluated by an amplification factor, AF. [74, 87] In a dual electrode system consisting of two working electrodes (WE1 and WE2), AF is typically calculated by eqn. (5),

$$ AF=\frac{i_{cyc}}{i_{off}} $$
(5)
Fig. 8
figure 8

Comparison of generator-collector (GC) and non-GC mode cyclic voltammograms of 1 mM Fe(CN)6 3−on a high density array with a 200 nm insulator separating the ring and disk electrodes. (a) Non-GC mode: ring electrodes are floating, and disk electrodes are swept at 100 mV/s (black) and 5 mV/s (red). (b) GC mode: disk (solid) and ring (dashed) current for ring electrodes held 0.5 V and disk electrodes swept at 100 mV/s. Reprinted with permission from ACS Nano 2013, 7, 5483. Copyright (2013) American Chemical Society

Full size image

where i cyc and i off are the faradic currents obtained at the first working electrode, WE1, in RC mode with WE2 held at a potential supporting the opposite reaction, and non-RC mode (WE2 is floating), respectively. This equation can be used to evaluate the performance of P/R-RDE arrays as well as other electrode geometries.

The AF, a direct result of the number of redox events per unit time, depends largely on the diffusion of the redox molecules from one electrode to the other. For example, in a TLC electrode AF can be estimated by, [80, 91, 92]

$$ AF=\frac{\pi r}{4h} $$
(6)

where r and h are the radius and interelectrode distance of the TLC electrode, respectively. Thus, decreasing the interelectrode distance, h, for a TLC electrode of a given size, r, produces higher AF. Similarly, decreasing the thickness of the insulating layer in a P/R-RDE array can also produce higher current amplification. [26, 89, 93] The dependence of AF on the interelectrode distance has been demonstrated from previous studies, which reported an increase of AF ranging from several-fold for ~μm scale h, [89, 93] to ~30-fold as h decreases to ~100 nm. [90, 94] As indicated from eqn. 6, increasing the electrode size can also produce larger current amplification. This size dependence of the AF was demonstrated on the P/R-RDE array, where increasing electrode density results in higher collection efficiency and larger faradic current. Using smaller pores with higher density maximized the collection efficiency, resulting in current amplification up to ~50-fold. [90].

Steady-state response with small charging current

Another important characteristic of the RC device is the steady-state response, Fig. 8(b), of the CVs, which is independent of the scan rate as well as the array size. [74, 89] This property is attributed to the fast mass transport occurring locally between closely-spaced electrodes. Accordingly, the number of molecules reaching the electrode surface is not limited by a diffusional boundary layer. A 2 mm × 2 mm P-RDE array has been reported to exhibit steady-state response operating in RC mode, whereas a macroscopic electrode of comparable size would exhibit diffusion limited currents. [89, 95] Furthermore, in a well-designed RC device with high collection efficiency, WE2 produces almost the same faradic current as WE1, but with much smaller charging current. For example, in an R-RDE array, charging current at the ring electrodes (WE2) held at constant potential is ~10-fold smaller than that of the disk electrode. [26] This property is especially important for high scan rate CV measurements.

Potential modulation and selectivity

Since the reaction at WE1 in an RC device relies strongly on the species diffusing from WE2, it is possible to control the redox form of the species diffusing to WE1 by controlling the potential at WE2, [26, 91] as shown in Fig. 9. When the ring electrode (WE2) is held at a potential positive of E eq for Fe(CN)6 3/4-, e.g. 0.5 V vs. Ag/AgCl, oxidation of Fe(CN)6 4− to Fe(CN)6 3− occurs at WE2, whereas reduction Fe(CN)6 3− to Fe(CN)6 4− is observed on the disk electrodes (WE1). In contrast, holding WE2 at a potential negative of E eq , e.g. -0.1 V, reverses redox reactions occurring at both the ring and disk electrodes. Finally, intermediate behavior with both anodic and cathodic current collected at the rings and disk electrodes is generated if WE2 is set at an intermediate potential, e.g. +0.25 V vs. Ag/AgCl.

Fig. 9
figure 9

a Ring potential dependence of the voltammetric response of 1 mM Fe(CN)6 3−on a high density array with 200 nm ring-disk separation. Disk electrodes are swept at 100 mV/s, and ring electrodes are held at 0.5 V (black), 0.25 V (red), and −0.1 V (blue); disk current (solid) and ring current (dashed). b Finite element simulated CV response of an array of 10 electrodes in 1 mM Fe(CN)6 3− solution on a high density array with the ring electrode held at 0.5 V (black), 0.25 V (red), and −0.1 V (blue). (c–e) Concentration profiles of Fe(CN)6 3− in adjacent pores with a radius of 250 nm taken from simulations of an array of 10 electrodes. The height of the insulator layers (gray) and the ring electrode (yellow) is 200 nm. Reprinted with permission from ACS Nano 2013, 7, 5483. Copyright (2013) American Chemical Society

Full size image

One application of this potential modulation is to improve measurement selectivity, which is especially important in applications that are not coupled to a separation technique. Use of the RC effect to improve selectivity depends on the reversibility of the redox reaction. Since irreversible species are not regenerated, they produce smaller faradic currents compared to reversible species under the same conditions. [80, 87, 89] In addition, an irreversible interfering species can be further depleted at the WE2 by holding it at a constant potential, enabling determination of reversible analytes at the WE1 with little interference. On the basis of this principle, selective measurement of redox reversible dopamine in the presence of irreversible ascorbic acid has been demonstrated on RC devices of different geometries. Nanoscale P/R-RDE arrays are better suited for this purpose than microscale arrays, because nanopore electrodes deplete the irreversible species more efficiently before they diffuse to and react on the disk electrode. Selectivity can be further enhanced by integrating the electrode array with micro/nanofluidic channels, which limit the amount of redox species capable of diffusing to the electrode surface, [80, 87] thereby further reducing the irreversible species in RC but not the reversible species. [26, 80, 87] Using these strategies, the measured selectivity for dopamine against ascorbic acid in an R-RDE array coupled with nanochannels was improved by 103x, allowing interference-free detection of dopamine in the presence of 100-fold excess of ascorbic acid. [87].

Modulation of the redox reaction can also be used to address the more challenging situation in which two reversible species with similar redox potentials must be differentiated. Although the RC effect cannot be applied directly in this situation, the potential of WE2 can be adjusted such that one species exhibits a cathodic response, while the other gives an anodic response at WE1. This principle was successfully applied to differentiate reversible species with overlapping redox potentials, such as dopamine/Fe(CN)6 3− and ferrocenemethanol/Fe(CN)6 4−. [87, 96] Again, nanoscale R-RDE arrays outperform their microscale counterparts due to higher conversion ratios.

Ion accumulation and permselectivity in nanopores

In RC devices, larger current amplification can be achieved by decreasing interelectrode distance or increasing electrode density. On the other hand, the pore size of P/R-RDE array has a great impact on redox cycling performance. As the diameter of the pores is reduced to a size commensurate with that of the electrical double layer, unique behaviors such as ion permselectivity and ion concentration polarization are observed. [97–100] Indeed, as a result of high surface-to-volume ratio, these effects can increase ion concentrations by several orders of magnitude. [97, 99] The ion enrichment observed in charged nanopores at low ionic strength requires very low analyte concentration and the absence of supporting electrolyte (SE) - novel conditions for electrochemical measurements. For this reason, the accumulation effect has not been employed to improve sensitivity in previous nanopore array measurements without the RC effect. Even more challenging, non-RC measurements performed in nanopores would deplete the accumulated ions before reaching steady-state current, since the number of ions, in the nanopores even at accumulated concentrations, is comparable small to those diffusing from bulk solution.

However, because the RC effect relies on redox species cycling locally between two electrodes, the localized analyte concentration, which determines the number of redox events, can benefit from ion accumulation to further improve measurement sensitivity. This concept has been demonstrated on an R-RDE array [88] using Ru(NH3)6 3+, which is expected to accumulate in nanopores with negative-ζ SiO2 surfaces. Figure 10 illustrates this ion accumulation/ion permselectivity effect in increasing faradic current at low analyte concentration in the absence of SE. In contrast to the canonical electrochemical protocol which uses SE, three distinct regimes are observed in the absence of SE, revealing strong ionic strength dependence. At low analyte concentration, current enhancements as large as 100-fold arise from a combination of ion enrichment and ion migration effects. Combining these with the ~20-fold enhancement due to redox cycling yields a total current amplification ~2000-fold on the R-RDE array, making it interesting for electrochemical analysis.

Fig. 10
figure 10

Limiting ring current from voltammograms of Ru(NH3)6 3+ as a function of concentration both with (0.1 M KCl, blue) and without (red) supporting electrolyte. Black curve: value of 1/κh calculated at the corresponding Ru(NH3)6 3+ concentration in the absence of supporting electrolyte. Reprinted with permission from J. Am. Chem. Soc. 2014, 136, 7225. Copyright (2014) American Chemical Society

Full size image

Self-induced redox cycling

In typical RC measurements, two electrodes are held at potentials positive and negative of E eq to initiate and sustain electrochemical cycling. However, redox cycling can also occur when a powered electrode is placed adjacent to an unbiased electrode. [89, 95, 101–106] This so-called self-induced redox cycling (SIRC) relies on the bipolar behavior of the floating electrode that is coupled to the biased WE. Under these conditions, depletion of the redox species by the biased WE produces a spatially-dependent concentration polarization relative to the adjacent floating electrode, [95, 101] which is sufficient to drive the oxidation and reduction reactions occurring respectively at opposite poles of the unbiased bipolar electrode.

Electrons flow within the bipolar electrode to balance the charge induced by the localized redox reactions, as confirmed by voltage measurements and optically observed by electrogenerated chemiluminescence. [71, 72] This electron flow can be used to couple two independent redox reactions, for example using one reaction to monitor the other. Indeed, significant sensitivity improvement was demonstrated by Yasukawa et al., who coupled the oxidation of 4-aminophenol with the reductive deposition of Ag+, the accumulation of which was then determined by anodic stripping voltammetry. [102, 104] Recently, we explored this technique to couple RC reactions of target analytes on the P-RDE array with the electrochemical reaction of fluorigenic redox reporter systems, e.g. dihydroresorufin/resorufin. [94] The change in fluorescence intensity in the reporter system is determined by the redox reaction of the target analyte, Fig. 11. Electrochemical reduction of 1 nM Ru(NH3)6 3+ on the nanoelectrode array was detected by monitoring fluorescence at the remote reporter site, demonstrating the possibilities for high sensitivity electroanalytical measurements by coupling fluorigenic reporter redox reactions to self-induced redox cycling in open bipolar electrode topologies.

Fig. 11
figure 11

a Schematic diagram showing a top view of the macroscopic layout of the bipolar RDE (BRDE) array and remote electrode separated by mm-distances; b schematic side view of the BRDE array and illustration of the mechanism of self-induced redox cycling for Ru(NH3)6 2/3+ (abbreviated as Ru2/3+). SIRC in the BRDE at OA is monitored by a reporter redox couple (Repox/red) at a remote location (OP); c SEM image of the array at 45° tilt. Reproduced from from Chem. Sci. 2015, 6, 3173. with permission of The Royal Society of Chemistry

Full size image

Novel nanoarrays for electrochemistry

Nanoscale spectroelectrochemistry

Multi-modal measurements are particularly powerful since they allow information from two different sources to be correlated in a single measurement. Notably, coupling fluorescence to electrochemistry has led to a number of new measurement modalities (vide supra) with attendant improvements in sensitivity. In this regard, zeptoliter-volume metal-clad 0D nanopores are of special interest, because they can exhibit both nanoelectrochemical and nanophotonic properties simultaneously. Nanopores in an opaque metal film below a certain size cutoff exhibit zero-mode waveguide (ZMW) behavior, which produces strongly-confined, exponentially decaying evanescent fields within the pore. Simultaneously, if the ZMW is fabricated from a metal with a large ideally polarizable region, it can be used as a working electrode. Under the proper conditions, single molecule spectroelectrochemistry can be performed by coupling the electron transfer events from fluorigenic reactions to the optical behavior of the ZMW.

In an initial demonstration of the electrochemical possibilities of ZMWs, single molecules of flavin adenine dinucleotide (FAD) were functionalized in individual Au ZMW nanopores and investigated as a function of potential with wide field epi-illumination microscopy. [107] By controlling the potential, single immobilized FAD molecules were modulated between emitting (oxidized) and non-emitting (reduced) states. Examining the statistical distribution of emitting/non-emitting states over many single molecules gives insight into the single molecule dispersion in electrochemical behavior, e.g. E eq variation among individual molecules. This work was extended in similar ZMWs but with freely diffusing FAD and flavin mononucleotide (FMN) by moving to confocal fluorescence geometry. Because single molecule fluorescence bursts could be measured down to the μs-timescale, the problem posed by μs residence times of freely diffusing molecules in single ZMW nanopores was circumvented, and single molecule fluorescence analogs of scanned potential experiments, in particular chronofluorometry and cyclic potential sweep fluorescence, were realized, showing the power single molecule spectroscopy applied to electrochemical experiments. [73].

Nanoscale ion transfer at immiscible interfaces

The electrochemical behavior of immiscible liquid-liquid interfaces, especially the ability to study ion transfer across the interface, has been of interest for some time. Arrigan and coworkers have pioneered the translation of this experiment to nanoscale interfaces between two immiscible electrolyte solutions, i.e. nanoITIES arrays. [108–111] Produced primarily with FIB milling, ion transfer can be electrochemically monitored across the nanoarray. Figure 12 illustrates the tight relationship between nanopore geometry and electrochemical performance by showing the ion transfer current density for various arrays and the relationship between sensitivity and pore size. The sensitivity of ion transfer across the phase boundary increases as the size of the nanopores decreases, providing ample motivation for moving to nanoscale interfaces. [112, 113] Similar to other nanoelectrode array devices, maintaining diffusional independence between electrodes is an important consideration in order to maximize signal. [114] The expected limiting current for nanoITIES arrays is given by,

$$ I=4 zFDcrN $$
(7)
Fig. 12
figure 12

a Forward scan ion transfer current density (J, μA cm−2) vs. TEA+ concentration (μM) for each nanointerface array, based on nanopore design: 1 (■), 2(▲), 3(×), 4(♦) and 5 (+). b Sensitivity of the nanointerface array (μA cm−2 μM−1) as a function of the nanopore radius r a (nm) used to form the nanointerface. c Sensitivity (μA cm−2 μM−1) as a function of r a −1, compared to theory (dashed line). The error bars in (b) and c were calculated from the standard deviations of the slopes of the calibration curves in (a). Reproduced from Analyst 2011, 136, 4674. with permission of The Royal Society of Chemistry

Full size image

where z is the charge, D is the diffusion coefficient, c is the bulk concentration, r is the radius of the nanopore, and N is total number of pores in the array. [114–116] Recently, independent radial diffusion for nanoITIES array devices was shown to occur when the interpore distance to pore diameter ratio is greater than ~50. These devices show promise as new types of electrochemical sensors.

Future outlook

Nanopore- and nanochannel-based electrode arrays are particularly promising for electrochemical applications with enhanced performance, e.g. lower limits of detection and improved selectivity, by taking advantage of novel modes of operation, such as coupled single molecule spectroscopy and redox cycling that are uniquely enabled in nanoscale structures. The ultimate level of sensitivity, allowing the electrochemical behavior of single redox-active molecules to be studied, is attainable in these types of nanopore-based devices. Thus, the enhanced analytical performance metrics of these nanoelectrode arrays presage great advances for the future. The greatest remaining challenge in the development of nanoelectrode arrays is fabrication. In particular, the translation of small batch fabrication to large-scale, cost-effective production remains a barrier to widespread adaption of nanopore- and nanochannel-based electrode arrays for sensing and testing applications, but this too should yield to ever more clever and intricate experimental approaches.