High spatial resolution electrochemical biosensing using reflected light microscopy

If the analyte does not only change the electrochemical but also the optical properties of the electrode/solution interface, the spatial resolution of an electrochemical sensor can be substantially enhanced by combining the electrochemical sensor with optical microscopy. In order to demonstrate this, electrochemical biosensors for the detection of hydrogen peroxide and glucose were developed by drop casting enzyme and redox polymer mixtures onto planar, optically transparent electrodes. These biosensors generate current signals proportional to the analyte concentration via a reaction sequence which ultimately changes the oxidation state of the redox polymer. Images of the interface of these biosensors were acquired using bright field reflected light microscopy (BFRLM). Analysis showed that the intensity of these images is higher when the redox polymer is oxidized than when it is reduced. It also revealed that the time needed for the redox polymer to change oxidation state can be assayed optically and is dependent on the concentration of the analyte. By combining the biosensor for hydrogen peroxide detection with BFRLM, it was possible to determine hydrogen peroxide in concentrations as low as 12.5 µM with a spatial resolution of 12 µm × 12 µm, without the need for the fabrication of microelectrodes of these dimensions.


Details on the FTO and ITO electrodes
The planar, optically transparent FTO and ITO electrodes used throughout the present work were fabricated starting from metal oxide-coated, 3 inch diameter, glass wafers and using the following sequence of six steps: 1.) ITO / FTO deposition, 2.) photolithography, 3.) wet etching of transparent oxide, 4.) silicon nitride deposition, 5.) photolithography and 6.) silicon nitride dry etching (Fig. S1a). Optical microscopy images of the central parts of typical electrode arrays are shown in Fig.   S1b. As one can observe in Fig. S1b, while the lateral dimension of the electrodes was kept constant (500 µm × 500 µm), the inter-electrode spacing and the design of electrochemically inactive hole structure was modified. However, these parameters carry no significant importance in the current stage of our study. SEM images, revealing the higher surface roughness of the FTO electrodes as compared to the ITO electrodes, are shown in Fig. S1c and S1d. These SEM images confirm similar findings using Atomic Force Microscopy (Fig. 1b).

Electrochemical cell for opto-electrochemical measurements
To the best of our knowledge, there is no commercially available electrochemical cell that allows observing electrodes fabricated onto ~170 µm thick glass through the high magnification objectives (e.g. 63×) of an inverted microscope. Therefore, we constructed the electrochemical cell shown in Fig. S2. This electrochemical cell was built using a plastic microscope slide and it is compatible with common microscope stages. It holds the 2 cm × 2 cm × 170 µm glass slide carrying the metal oxide electrodes without hindering the access of the objectives of the inverted microscope to the electrode/solution interface. The cell can hold around 400 µL of solution. Due to this small volume of the cell, an Ag/AgCl wire was used as quasi-reference electrode. The stability of such a quasi-reference electrode was recently investigated and found satisfactory (~ 1 mV h -1 ) 1 .

Reaction cascades providing the analytical useful signals
The current signals of the hydrogen peroxide and glucose biosensors investigated in the present work arise as a result of the biochemical and electrochemical processes schematically depicted in Fig. S3a and S3b. These reactions are presented into more details elsewhere 2,3 .
Very important to note, the horseradish peroxidase (HRP) -hydrogen peroxide pair oxidizes the osmium complex-based redox polymer while the glucose oxidase (GOx) -glucose pair reduces the osmium complex-based redox polymer. Equally important to note, the processes schematically depicted in Fig. S3a and S3b cause also the BFRLM signal to change (because the refractive index of the redox polymer is different when the polymer is oxidized as compared to when it is reduced). However, the analyte concentration proportional optical signal is recorded with the potential of the electrode set to the OCP (i.e. in the absence of the electron transfer step in between the electrode and the redox polymer). Figure S3. Redox cycles occurring in the used redox hydrogels when these are prepared with either HRP (a) or GOx (b) and are exposed to either hydrogen peroxide (a) or glucose (b).

HRP-based redox hydrogel
The ability of redox hydrogel-modified metal oxide electrodes to detect hydrogen peroxide or glucose was always tested in combination with purely electrochemical methods. with the remaining combinations of metal oxide electrodes (FTO and ITO) and redox hydrogels (with HRP and with GOx). Figure S4 depicts the results obtained using an ITO electrode modified with a HRP-based redox hydrogel in combination with purely electrochemical methods (i.e. cyclic voltammetry and chronoamperometry).  only when using purely electrochemical methods, but also in our opto-electrochemical approach. The coefficient of variation characterizing the sensitivity was 27% for the electrochemical sensors for hydrogen peroxide detection made with ITO and 15% for similar sensors made with FTO. The detection limit and the limit of quantitation were found to be 90 µM and 310 µM, respectively. These limits are larger than those obtained for similar sensors made with FTO (15 µM and 70 µM, respectively).  These results of the cyclic voltammetry study are in perfect line with the reaction cascade depicted in Fig. S3b, according to which the GOx -glucose pair reduces the redox polymer and the electrode oxidizes the redox polymer. During the chronoamperometric experiments, the potential of the FTO electrode modified with GOx-based redox hydrogel was set to either +0.300 V (to oxidize the redox polymer) or to -0.150 V (to reduce the redox polymer) and the concentration of the glucose in the solution bathing the electrode was increased stepwise from 0 to 1.6 mM. The resulting currents were observed to depend on the glucose concentration (Fig. S5b). The anodic currents, recorded at the end of the second anodic pulse (i.e. at t = 256 s), allowed building the calibration curve shown in Fig. S5c  Interestingly enough, the calibration curve characterizing such sensors presented the same two regions even when built using the optical signals of a single region of interest (ROI) defined on the sensor/solution interface (instead of the current signals of the whole sensor -solution interface, see Fig. S5c vs. Fig. 4c). This sustains the robustness of our opto-electrochemical approach. The coefficient of variation characterizing the sensitivity of electrochemical sensors for glucose detection made with FTO was as small as 4%. The detection limit and the limit of quantitation were found to be 450 µM and 760 µM, respectively. Figure S6 shows the results obtained with an ITO electrode modified with a GOx-based redox hydrogel in combination with cyclic voltammetry and chronoamperometry. The cyclic voltammograms of the ITO electrode modified with GOx-based redox hydrogel display two current peaks (Fig. S6a). The peak at +0.244 V is due to the oxidation of the redox polymer found in the hydrogel and the current peak at +0.190 V is due to the reduction of the same polymer. In the presence of glucose, the current corresponding to the oxidation of the redox polymer increased while the current due to the reduction of the redox polymer decreased.

Electrochemical detection of glucose using an ITO electrode modified with GOx-based redox hydrogel
These results of the cyclic voltammetry study are in agreement with the reaction cascade depicted in Fig. S3b. During the chronoamperometric experiments, the potential of the FTO electrode modified with GOx-based redox hydrogel was set to either +0.300 V (to oxidize the Page 10 of 15 redox polymer) or to -0.150 V (to reduce the redox polymer) and the concentration of the glucose in the solution bathing the electrode was stepwise increased from 0 to 1.6 mM. The resulting currents were observed to depend on the glucose concentration (Fig. S6b). The currents recorded at the end of the second anodic pulse allowed building the calibration curve shown in Fig. S6c (average of 9 experiments with 9 different hydrogel-modified ITO electrodes). The sensitivity of this glucose biosensor built on ITO (calculated as the slope of the linear range of the calibration curve observed from 0.4 mM to 1.6 mM glucose) was ~ 19% higher than that of similar biosensors built on FTO (0.108 µA mM -1 vs. 0.087 µA mM -1 ). This confirms that ITO electrodes allow building better biosensors than FTO electrodes for both hydrogen peroxide and glucose detection. Moreover, this behaviour is maintained both when the biosensors are interrogated with purely electrochemical methods (as described in this Supplementary Information) and when they are interrogated with our opto-electrochemical approach (as described in the main text). The coefficient of variation characterizing the sensitivity of electrochemical sensors for glucose detection made with ITO was 12% (vs. 4% for similar sensors made using FTO). The detection limit and the limit of quantitation were found to be 170 µM and 510 µM, respectively. These limits are smaller than those obtained for similar sensors made with FTO (450 µM and 760 µM, respectively).   S7b vs. Fig. 3b). However, the evolution of the corrected mean intensity while changing applied potentials and hydrogen peroxide concentrations is somewhat noisier due to the smaller physical dimensions of the ROIs (12 µm × 12 µm vs. 18 µm × 18 µm). The sensitivity of the ROIs to hydrogen peroxide was nevertheless preserved in spite of the noisier optical signals.  . S8b vs. Fig. 4b).

Selectivity of the developed opto-electrochemical sensors
Redox hydrogel-based electrochemical biosensors are most often characterized by good selectivity because they work with low applied potentials 4 . As a result, they were already used in complex environments such as brain tissue 5,6 . Moreover, due to its excellent analytical performances, a redox hydrogel-based glucose sensor was also integrated into a commercial glucose meter 7  As one can observe in Fig. S9a and S9b, 1.6 mM glucose produced a signal corresponding to only ~ 5 µM hydrogen peroxide on the opto-electrochemical hydrogen peroxide sensor. In the same time, 0.9 mM hydrogen peroxide (a very high concentration taking into account the Page 15 of 15 envisaged applications, e.g., measurements at cellular level) produced a signal corresponding to ~ 140 µM glucose on the opto-electrochemical glucose sensor ( Fig. S9c and S9d). Both signals produced by the non-target compounds correspond to analyte concentrations smaller than the detection limits of the opto-electrochemical sensors (which were calculated to be 20 µM and 320 µM for the opto-electrochemical hydrogen peroxide sensor and the optoelectrochemical glucose sensor, respectively). Both signals can be further decreased by using relatively simple methods described in the literature for improving the selectivity of redox hydrogel-based biosensors (e.g., by using an additional Nafion layer deposited on top of the redox hydrogel 5,6 ).