Toward the Detection Limit of Electrochemistry: Studying Anodic Processes with a Fluorogenic Reporting Reaction

Recently, shot noise has been shown to be an inherent part of all charge-transfer processes, leading to a practical limit of quantification of 2100 electrons (≈0.34 fC) [Curr. Opin. Electrochem.2020, 22, 170−177]. Attainable limits of quantification are made much larger by greater background currents and insufficient instrumentation, which restricts progress in sensing and single-entity applications. This limitation can be overcome by converting electrochemical charges into photons, which can be detected with much greater sensitivity, even down to a single-photon level. In this work, we demonstrate the use of fluorescence, induced through a closed bipolar setup, to monitor charge-transfer processes below the detection limit of electrochemical workstations. During this process, the oxidation of ferrocenemethanol (FcMeOH) in one cell is used to concurrently drive the oxidation of Amplex Red (AR), a fluorogenic redox molecule, in another cell. The spectroelectrochemistry of AR is investigated and new insights on the commonplace practice of using deprotonated glucose to limit AR photooxidation are presented. The closed bipolar setup is used to produce fluorescence signals corresponding to the steady-state voltammetry of FcMeOH on a microelectrode. Chronopotentiometry is then used to show a linear relationship between the charge passed through FcMeOH oxidation and the integrated AR fluorescence signal. The sensitivity of the measurements obtained at different timescales varies between 2200 and 500 electrons per detected photon. The electrochemical detection limit is approached using a diluted FcMeOH solution in which no faradaic current signal is observed. Nevertheless, a fluorescence signal corresponding to FcMeOH oxidation is still seen, and the detection of charges down to 300 fC is demonstrated.


Design of the closed bipolar electrochemical cell
Custom electrochemical cells for closed bipolar electrochemistry were prepared by bonding two separate molded poly(dimethylsiloxane) (PDMS) structures to a glass microscope slide (76 mm × 26 mm × 1 mm) from Carl Roth GmbH.  Figure S1 illustrates the design, with two cells bound to each glass slide (one cell for detection and one for reporting). The dimensions of the templates resulted in each cell having a volume of ≈ 1.33mL. The molded PDMS structures were prepared using a Sylgard™ 184 Silicone Elastomer kit from Dow Europe GmbH. A 10:1 ratio of elastomer to curing agent (by weight) was vigorously mixed (2-5 minutes) before removal of gas bubbles under vacuum (20-30 minutes). The mixture was then poured into a Teflon mold, covered with a glass slide (to provide a smooth upper surface), and left to set in a 75 • C oven overnight. The PDMS cell was then removed from the Teflon mold, washed with isopropanol, and left to dry.
To bond the PDMS to the microscope slide, both the slide and the PDMS structure (smooth side) were cleaned with isopropanol, dried, and placed face up on a sheet of aluminium foil in the chamber of a Harrick Plasma pdc-002-ce plasma cleaner. The surfaces were exposed to plasma for one minute to generate free radicals, then quickly placed on top of each other. To finish, the foil with the microscope slide was transferred to a 125 • C hotplate (15 minutes Figure S2: Absorbance (BLUE) and fluorescence (RED) spectra (excitation @ 550 nm) recorded using an aerobic solution of 0.5 M NaOH + 10 µM Amplex Red. Figure S2 shows excitation and emission spectra recorded in a solution of Amplex Red in sodium hydroxide. In these conditions, the Amplex Red was rapidly photooxidised to resorufin, allowing the spectra to be recorded. The peak excitation and peak emission wavelengths were found to be in agreement with literature values of 571 and 585 nm respectively. It was determined from this that the TRITC-A filter set (λ excitation = 543-566 nm, λ emission = 582-636 nm) could be both used to sufficiently excite the resorufin and to measure the fluoresced light.

Influence and rate of photooxidation
The photooxidation of Amplex Red was demonstrated in a phosphate buffer saline solution.
Fluorescence intensity profiles were recorded using a freshly made solution of Amplex Red in PBS, then the same intensity profiles were recorded after the solution had been left on a lab bench to photooxidise for two hours. In both cases, an electrode (either glassy carbon or carbon fibre) which was not connected to any potentiostat was held in the solution, and the microscope was focussed on the electrode surface to ensure that results were comparable.   Figure S4 shows the fluorescence intensity measured at a carbon fibre electrode through a 20× objective. Due to the higher numerical aperture of this objective (0.5), the photooxidation is clearly visible by the shifting background intensity. It should also be noted that the amount of photooxidation during the measurement is much greater in the old solution.

Background electrochemistry of glucose on Pt and Au electrodes
Carbon electrodes were used to avoid the influence of glucose oxidation on the reporting electrode.  Figure S5 shows the significant activity of glucose observed on both Pt and Au electrodes, compared to the current observed in just sodium hydroxide. In the closed bipolar setup, some of the charge passed during ferrocenemethanol oxidation would be used to drive the oxidation of glucose on these materials, making them less appropriate as reporting electrodes.

S6
Evidence for the passage of current when switching to OCP The cell was kept on after most experiments, with the potential of the working electrode set to the value required for ferrocenemethanol reduction (0 V vs. Ag/AgCl). When switching to open circuit potential after an experiment, spikes in fluorescence intensity (as seen in Figure 4) were observed. This was investigated further using chronopotentiometry.  Figure S6 shows the fluorescence intensity profiles recorded during chronopotentiometry at extremely low currents. The intensity profile seen at 0.1 pA is similar to those seen at currents below 1 pA in Figure 9. The profile at 0.05 pA is closer to the profile observed when there is supposedly no current flowing. This indicates a limitation of the potentiostat in setting the current values, but more importantly suggests that there is still some current flowing at OCP. On this basis, it is theorised that switching to OCP can cause some temporary current flow which induces the fluorescence spikes observed experimentally.

Maximising the collection of fluorescence
Maximising the fluorescence signal naturally increases the sensitivity towards the detection of electrons. This can be achieved by increasing the incident light intensity and by using higher numerical apertures.   Figure S8 shows the fluorescence intensity measured through different microscope objectives during voltammetry of Amplex Red. Objectives with greater magnification also had higher numerical apertures, resulting in increased fluorescence intensities measured during the voltammetry.

S9
Example calculation of the gradients from Figure 7 The gradient of integrated fluorescent count vs. charge passed has units of counts C −1 . This can be converted to photons C −1 using Equation 1, which was provided by Hamamatsu in the documentation alongside the CMOS camera.
where CF is the conversion factor of the camera, ∆I is the measured intensity of pixels (background subtracted), and Q(λ) is the quantum efficiency of the camera. This can be modified to give Equation 2 Photons per electron = CF In this equation, the ∆I term is replaced with our gradient in counts C −1 and then multiplied  Figure S9 shows more evidence of this, measured during a pulsed amperommetric detection procedure. When the working electrode was pulsed to a potential for ferrocenemethanol oxidation (+0.5 V vs. Ag/AgCl), a peak in the fluorescence intensity was observed, but the width of this peak did not correlate to the duration of the pulse, suggesting the fluorophore lingers in solution.

S12
Fluorescence signal in K 2 SO 4 with and without FcMeOH In Figure Figure S10 shows the difference between the fluorescent signals when chronoamperommetry of the working electrode was recorded with and without FcMeOH in the detection cell. Even without the FcMeOH, there was a noticeable fluorescence signal. This is due to Amplex Red oxidation caused by the passage of electrons from double layer charging, electromagnetic noise, and oxidation of trace impurities. These contributions are present in all experiments and can be reduced to some extent, but not fully removed. However, since single photons S13 can be resolved, it is unlikely that any background fluorescence will limit the capabilities of remote optical reporting to monitor electrochemistry below the detection limit.  Figure S11 shows the same relationship between a solution containing only K 2 SO 4 and a solution containing both FcMeOH and K 2 SO 4 , except this time recorded with cyclic voltammetry. Once again the signal recorded in the presence of FcMeOH is greater than that recorded only in K 2 SO 4 and the difference between the two signals could be used to extract some information on the amount of FcMeOH that was oxidised. It is also clear that the background fluorescence signal is driven by processes such as double layer charging and electromagnetic noise.