Self-induced redox cycling coupled luminescence on nanopore recessed disk-multiscale bipolar electrodes

Self-induced redox cycling at nanopore ring-disk electrodes is coupled, through a bipolar electrode, to a remote fluorigenic reporter reaction.


Introduction
The combination of uorescence spectroscopy with electrochemistry presents new avenues for the study of redox reaction events, with potential for enhanced throughput, sensitivity, and spatial resolution. [1][2][3][4][5][6] For example, complementary optical and electrochemical information provided by coupling amperometry and total internal refection uorescence microscopy has been employed to unravel the mechanism of exocytosis events on single cells. 7,8 Fluorescence-detected cyclic voltammetry has also been used to study thermodynamic and kinetic characteristics of electron transfer in immobilized proteins, exhibiting variation in uorescence due to surface heterogeneity. 6,9 The coupling of electron transfer events to uorescence allows the inherently low background of uorescence measurements to be exploited, as shown by the demonstration of single molecule sensitivity in monitoring charge-transfer events in reversible redox dyes that switch between uorescent and nonuorescent states upon potential modulation. [10][11][12][13] Nonuorescent redox couples may also be studied by monitoring the interaction between the couple and a uorescent dye. For example, pH-sensitive uorescent indicators have been used to monitor electrochemically generated H + . 10,14 This simple, but effective, strategy has been developed into a combinatorial method for simultaneous parallel screening of electrocatalysts for the oxidation of methanol. 5 Alternatively, electrochemical reactions of nonuorescent analytes can be coupled to electrogenerated chemiluminescence (ECL) using a bipolar electrode which acts as anode and cathode simultaneously. 15,16 This conguration allows the redox reaction of the target couple to be monitored through ECL intensity changes at the indicator. 17 Zhang and coworkers have exploited this type of coupling reaction using bipolar electrodes 18,19 to develop uorescence-enabled electrochemical microscopy, which can monitor a large number of parallel redox events by uorescence imaging. 2 In an effort to improve the sensitivity of electrochemical detection, dual electrodes with mmto nm-scale spacing have been fabricated to take advantage of redox cycling (RC) to amplify redox events and enhance the measured current. [20][21][22][23][24][25] The RC effect, relying on the cycling of the redox species between two closely spaced electrodes, can provide up to 1000fold current amplication, achieving single molecule detection in favorable circumstances. [23][24][25][26] In addition, RC electrodes can be integrated within microuidic systems to execute hydrodynamic voltammetry thereby avoiding the loss of sensitivity due to transport across a diffusive boundary layer. 22,[26][27][28] The RC effect can also improve selectivity to species exhibiting different degrees of reversibility, as well as different redox potentials. 22,27,29,30 In addition, RC is compatible with microuidic systems, thus holding promise for lab-on-a-chip devices. 20,22,24,27,28 Therefore, it is reasonable to ask whether coupling of RC to uorescence detection could combine the advantages of the individual techniques to achieve singular sensitivity and selectivity of in the study of redox reactions.
In a typical RC measurement, two electrodes are held at potentials negative and positive of the analyte redox potential to initiate and sustain electrochemical cycling. 21,24,26 Surprisingly, self-induced redox cycling (SIRC) can be observed when a powered electrode is placed adjacent to a nearby unbiased (oating) electrode. [31][32][33][34] In this situation, depletion of redox species at the working electrode produces a location-dependent concentration polarization relative to the adjacent electrode. These local concentration differences can be sufficient to drive oxidation and reduction reactions occurring on the unbiased electrode. 31,34 This constitutes a bipolar electrode so that electrons owing to the other end can balance the charge induced by the localized redox reaction, as has been previously monitored using simple voltage measurements and conrmed by stripping voltammetry. [30][31][32]35,36 In the present study, the SIRC effect is employed to couple redox reactions of a target redox couple, at a nanopore-conned recessed disk electrode, with inherently low background uorescence measurements at the remote end of a multiscale bipolar electrode. The exposed bipolar electrode is multiscale, because the nanopore portion is separated from the recessed disk electrode by 100 nm, while the remote end of the electrode is in contact with a different solution located $1 mm away, viz. Fig. 1. The SIRC effect is studied by measuring the cyclic voltammetry (CV) of a model redox couple, Ru(NH 3 ) 6 2/3+ , using the recessed disk electrodes as working electrodes (WE) and the nanopore portion of the bipolar electrode as the collector in a generator-collector arrangement. In order to conrm the participation of the oating nanopore bipolar electrode in SIRC, electrogenerated chemiluminescence (ECL) 17,37-39 from Ru(bpy) 3 2+ and tri-n-propylamine, was monitored at the remote end of the multiscale bipolar electrode. To accomplish SIRCcoupled uorescence, the multiscale bipolar electrode was placed in contact with a redox-switchable uorescent species in the remote cell. SIRC in the nanopore bipolar recessed disk electrode (BRDE) array was detected by the oxidation of non-uorescent dihydroresorun to uorescent resorun and corresponding increase in uorescence in the remote cell. 2,40 Increasing uorescence due to electrochemical oxidation of non-uorescent dihydroresorun to uorescent resorun indicates an oxidation reaction at the recessed disk WE, while reduction of resorun to dihydroresorun and decrease of uorescence indicates reduction at the WE. 18 This scheme was validated using Ru(NH 3 ) 6 3+ , demonstrating sensitivity to concentrations as low as 1.0 nM Ru(NH 3 ) 6 3+ .

Device fabrication
Similar to the procedure used for developing recessed ring-disk electrode arrays, 20 above the BRDE array is 100 mm Â 100 mm. The second open area (OP) on the remote end of the bipolar electrode, varying with size, is 500 mm or 6 mm away from OA.

Electrochemical and uorescence measurements
Electrochemical experiments were conducted on a CHI bipotentiostat (842c, CH Instruments Inc.) using a Ag/AgCl reference electrode or thin lm Au quasi-reference electrode (CE/QRE, Fig. 1). Tri-n-propylamine (TPA) (1.0 M) was dissolved in 1.0 M HCl prior to mixing with Ru(bpy) 3 Cl 2 aqueous solution. The solutions of analytes and resorun were prepared in 0.2 M phosphate buffer (pH 7) and purged with N 2 for $5 min prior to use. Dihydroresorun (0.1 mM) was obtained by mixing 0.1 mM resorun with 0.5 M NaOH and 50 mM glucose. 41 For all electrochemical and ECL measurements, $30 mL solution was added to a PDMS well covering the electrodes. In the uorescence measurements, analyte of Ru(NH 3 ) 6 3+ or ferrocenemethanol was added to a PDMS well covering the OA portion, while 2 mL of resorun or dihydroresorun solution was added to the remote electrode area (OP, Fig. 1) and covered by a coverslip. Chronoamperometry was performed by applying potential steps for 1 s, followed by a return to a rest potential for a 5 s recovery period between steps. Fluorescence and ECL measurements were performed on an epiuorescence microscope (IX71, Olympus) equipped with an X-Cite 120 PC illumination system (Exfo) and a TRITC lter set (Chroma). All images were acquired using a 10Â, 0.25 NA lens and recorded at 10 frames per second using an electron-multiplier CCD camera (PhotonMax512, Princeton Instruments). WinView (Princeton Instruments) and ImageJ (NIH) soware packages were used for acquiring images and data analysis, respectively.

Results and discussion
Self-induced redox cycling on the BRDE array The SIRC effect has been observed previously on interdigitated electrode arrays and planar-recessed disk electrode arrays with mm-scale spacing. 31,34 In order to increase the efficiency of redox cycling and generate larger current amplication, multiscale BRDE arrays, with a nanopore interelectrode spacing of 100 nm, were fabricated in this study. Fig. 1 illustrates the layout of the BRDE array prepared with two openingsone on the BRDE array (OA) and another on the remote portion of the bipolar electrode (OP). Fig. 2 shows the results of CV measurements of 1 mM Ru(NH 3 ) 6 3+ on the BRDE array. In order to reveal the effect of the bipolar electrode in redox cycling, voltammograms obtained on a recessed microdisk electrode array of the same size, i.e. with the same geometry as the BRDE array but without a top Au layer, are used for comparison. With the planar electrode oating and only OA in contact with the solution, a pseudo-steady-state response with limiting current of 120 nA was observed, a value 4-fold larger than the peak current (30 nA) obtained on the recessed microdisk electrode array. This change in CV response and increase in faradaic current are attributed to the SIRC effect mediated by the redox reactions occurring on the nanopore portion of the oating bipolar electrode. The SIRC effect is more pronounced when the remote portion of the bipolar electrode (OP) is exposed to Ru(NH 3 ) 6 3+ solution, i.e. Ru(NH 3 ) 6 2/3+ is the reporter species.
Indeed, the resulting current amplication increases with the area of OP on the planar electrode that contacts the solution (Fig. S2, ESI †). With a sufficiently large OP area (1 mm Â 5 mm), a steady-state response (Fig. 2, red curve) with maximized SIRC effect is achieved with SIRC current equivalent to that obtained by biased RC with the collector electrode at +0.1 V vs. Ag/AgCl (Fig. 2, green curve). The limiting current obtained on the BRDE array under these conditions is $870 nA, yielding an amplication factor of $30 and conrming the participation of the oating bipolar electrode in redox cycling. The proposed mechanism for SIRC on the BRDE array is illustrated by Fig. 1(b), which is similar to those suggested previously for interdigitated electrode and microscale planar recessed-disk electrode arrays. 31,34 The reduction of Ru(NH 3 ) 6 3+ to Ru(NH 3 ) 6 2+ on the disk electrode produces a locationdependent concentration polarization of Ru(NH 3 ) 6 3/2+ at the nanopore bipolar electrode. Specically, with a reducing potential applied at the disk electrode, the ratio of Ru(NH 3 ) 6

3+
to Ru(NH 3 ) 6 2+ is much larger at OP than at OA. As a result, reduction of Ru(NH 3 ) 6 3+ at OP is coupled to oxidation of Ru(NH 3 ) 6 2+ at OA, Fig. 1(b), in order to maintain electroneutrality of the bipolar electrode. This then facilitates SIRC on the BRDE array.

SIRC effect observed by ECL
Electron transfer at a oating electrode with SIRC has been previously observed on interdigitated electrode arrays. 31,32,34,36 In the present study, ECL was used to monitor the induced redox reaction at the unbiased bipolar electrode with OP located 500 mm away from OA. Potential steps were applied to the disk electrode with the bipolar electrode oating and both electrodes in contact with 5 mM Ru(bpy) 3 2+ and varying concentrations of  Fig. 3(a), potentially modulated ECL was observed on OA due to oxidation and subsequent reaction of Ru(bpy) 3 2+ with TPA, thereby generating chemiluminescence. [37][38][39] Similar to the mechanism given in Fig. 1, the oxidation of Ru(bpy) 3 2+ to Ru(bpy) 3 3+ at recessed disk electrode at OA leads to a gradient in the concentration of Ru(bpy) 3 3/2+ between the exposed portions of the bipolar electrode at OP and OA. Accordingly, reduction of Ru(bpy) 3 3+ and oxidation of Ru(bpy) 3 2+ occur on the OA and OP areas of the bipolar electrode, respectively, to counteract this difference and maintain electroneutrality. ECL with a similar potential dependence to that observed at OA, Fig. 3(a) is observed at OP, thus conrming the participation of the oating bipolar electrode in redox cycling. The slight delay of the self-induced ECL response is assigned to the accumulation of Ru(bpy) 3 3+ at OA, which is required to trigger the oxidation of Ru(bpy) 3 2+ and TPA at OP, since TPA is oxidized at a more positive potential. The induced ECL intensity at OP is signicantly affected by the concentration of TPA, Fig. 3 (Fig. S4 †), the latter exhibiting relatively strong, TPA concentration-independent emission above 50 mM TPA. These results indicate that a concentration difference across the two ends of the bipolar electrode is crucial for induced reactions on the OP portion of the bipolar electrode and SIRC at the nanopore electrode array (OA).

SIRC coupling with uorescence microscopy
Electron transfer at the oating bipolar electrode, conrmed by the above ECL results, can be employed to couple SIRC to uorescence emission, such that a redox reaction at the nanopore-recessed disk electrode (OA) can be monitored by uorescence at the OP portion of the multiscale bipolar electrode. Separation of the target redox couple from the reporter is invaluable, for example, in eliminating potential interferences in analytical measurements. To test this hypothesis, the SIRC effect was rst measured by contacting OA and OP with two different solutions. OA was lled with 1 mM Ru(NH 3 ) 6 3+ , while OP was lled with Ru(NH 3 ) 6 3+ of different concentrations. With the introduction of the target solution to OP, an obvious increase of faradic current can be seen at the disk electrode at OA (Fig. S5 †). With a xed OP area (100 mm Â 100 mm), larger current amplications are obtained on the array with higher Ru(NH 3 ) 6 3+ concentrations at OP, since these produce larger concentration differences between OP and OA. A similar structure was then used to investigate the coupling of uorescence to voltammetry. The exposed spot OP was lled with 0.1 mM dihydroresorun, a nonuorescent phenoxazine dye that can be oxidized to resorun, the latter exhibiting strong uorescence. 41,42 The potential modulated change of uorescence intensity of dihydroresorun has been observed previously 41 and was conrmed here by direct electrochemical oxidation, Fig. 4 (red curve). Similar switching of uorescence was observed on the oating bipolar electrode at OP when potential steps were applied to the disk electrode sufficient to oxidize ferrocenemethanol (Fc) to ferriceniumethanol (Fc + ). This observation is consistent with the SIRC effects reported above; Fc + generated at the recessed nanopore disk WE is reduced to Fc at the OA portion of the bipolar electrode, which is then coupled to the oxidation of dihy-droresorun (H 2 RF) to resorun (RF) at OP. The overall reaction, should occur spontaneously, since Fc + /Fc has a more positive standard reduction potential ($0.2 V vs. Ag/AgCl) 18,43 than the resorun couple ($À0.1 V vs. Ag/AgCl). 42 The degree of reaction and the resulting uorescence intensity should therefore depend on the concentration of Fc, which is conrmed by comparison of the 100 mM and 1 mM results in Fig. 4. It is interesting to note that the recovery of the uorescence to baseline is slower in the presence of SIRC than with direct potential-modulation. This observation is reasonable, since no reaction occurs at the recessed disk WE at 0 V in the presence of Fc. Instead, the recovery from uorescent state to nonuorescent state (baseline) relies on the diffusion of H 2 RF from the bulk solution to replace the RF on the electrode surface at OP, which is a slow process compared to the potential-driven reaction.
Having established the efficient coupling of redox reactions in the nanopore array at OA to uorigenic reactions at the remote end of the bipolar electrode, the approach was used to monitor the presence of analyte at OA, using the H 2 RF/RF reporter system at OP. The results for determination of Ru(NH 3 ) 6 3+ at concentrations in the range 1 mM to 1 nM are given in Fig. 5. Applying a reducing potential to the recessed disk WE results in decreased uorescence, similar to the potential-driven behavior observed with the Fc/Fc + couple. The overall coupling reaction is, Similar to Reaction (1), this reaction occurs spontaneously, since the standard reduction potential of H 2 RF/RF is more positive than that of Ru(NH 3 ) 6 2/3+ (À0.2 V vs. Ag/AgCl). 22,34 Also similar to the behavior of the Fc/Fc + couple at OA, recovery of the uorescence baseline intensity at 0 V is a slow, diffusioncontrolled process. Again, this is reasonable, since reduction of Ru(NH 3 ) 6 3+ at the WE at 0 V is not spontaneous.
Since one goal of this study was to exploit the coupling of uorescence emission and electrochemistry to improve the sensitivity of redox cycling measurements, Ru(NH 3 ) 6 3+ determinations were also performed at lower concentrations using the same electrode geometry. Similar behavior with detectable change in uorescence intensity could be observed down to 1 nM Ru(NH 3 ) 6 3+ (purple curve, Fig. 5). This excellent sensitivity is attributed to the combination of redox cycling, that amplies the redox event, and the inherently low background of uorescence measurements. Given the direction of the uorigenic reaction, reductions at the WE are accompanied by a highto-low uorescence transition, as seen in Fig. 5, which is not ideal from a measurement perspective. Thus, although a complete delineation of the analytical gures-of-merit is beyond the scope of the present work, limits of detection below 1 nM for this model analyte are clearly achievable by SIRC-coupled uorescence measurements. Furthermore, oxidation reactions at WE, which are signied by increases in uorescence emission (cf. Fig. 4), should provide even better performance.

Conclusion
These experiments introduce a unique multiscale bipolar electrode scheme to couple self-induced redox cycling in nanopore arrays to electrochemistry and uorescence of a reporter redox pair at a location remote from the target redox couple. The nanopore end of the bipolar electrode is placed adjacent (100 nm) to a recessed disk electrode, allowing efficient transport between the two electrodes in order to achieve redox cycling. In contrast, the remote end of the bipolar electrode is placed in a cell far from the BRDE to implement highly efficient reporter measurements, e.g. using redox-coupled uorigenic reactions. Together these characteristics permit the effective coupling of redox cycling with uorescence measurements for ultrasensitive electrochemical detection. SIRC with current amplication up to 30-fold was observed on the BRDE array with an unbiased planar electrode in CV measurements. Although not as large amplication factors achievable with biased collector electrodes, the participation of the bipolar electrode in redox cycling and its electrical connection to the remote end of the electrode permit the target event and the reporter reaction to spatially separated, greatly reducing measurement background.