Direct Quantitative Redox of 8 Zeptomoles of Molecules on a Macroscale Electrode

Abstract

A home-built opto-electrochemical differential reflectometer, SEED, has been developed to push the measurability of electrochemical detection on macroscopic electrodes closer to its fundamental limit determined by shot noise. By positioning a redox-active dye, methylene blue (MB), close to the electrode in a DNA self-assembled monolayer (SAM), the quantitative sensitivity of SEED is measured. By performing cyclic voltammetry (CV) and SEED on more than 50 SAMs immobilized on gold electrodes without perturbing the chip, the signal of SEED at 400 V/s is shown to be linearly proportional to the local redox current measured by CV at 1 V/s. The differential reflectivity at the formal redox potential corresponding to ∼9,375 electrons was 4.5-fold above the ultimate practical shot-noise limit of counting 2,100 electrons in an electrochemical measurement, irrespective of the transduction method. The equivalent peak current of ∼0.12 pA is comparable to single-molecule electrochemistry. Combinatorial electrochemistry on a monolithic electrode is obtained by scanning the laser beam. SEED is potentially a broad platform for combinatorial electrochemistry on monolithic electrodes that, for example, can be applied to MB-tethered ssDNA SAM systems that are of pervasive interest to sense a variety of analytes, from atomic ions to biomolecules.

Electrochemical detection of analytes is a pervasive chemical sensing method because it is intrinsically quantitative: the redox current is linearly proportional to analyte concentration. As molecular understanding of disease biology increases and biomarkers are identified, there is strong motivation to quantify analytes at femtomolar levels. For example, methods such as electrochemical, , CRISPR, , nanopore, , and surface plasmon are being developed to quantify specific microRNA targets in the femtomolar (fM) to nanomolar (nM) range − for the early diagnosis of cancer , neurological, and cardiovascular , and other diseases. In the last three decades, strategies have been developed to amplify the current due to discrete molecular reactions by rapid temporal and/or spatial averaging to count single-molecule reactions. In the first successful approach, the medium is confined over a nanoscale electrode in a space ∼10⁶ fold smaller than diffusion length so that a single (or few) molecule(s) react at the electrode at high rate to generate measurable currents. Several approaches involving spatial confinement coupled with single-molecule spectroscopy have been developed since. , A second approach for “molecule counting” has been the generation (or blocking) of discrete electrochemical current bursts arising from the rapid, stochastic collisions of nanoparticles with an ultramicroelectrode (UME). , This principle has been applied to measure proteins at the single-molecule level as passive and active , entities, and to detect heavy metal ions at fM levels. Orthogonal spectroscopic methods applied during electrochemical reactions have been shown to achieve single-molecule resolution using local surface plasmon-enhanced spectroscopy enabled by nanoparticles. , All these methods require sophisticated sample geometries, such as UMEs, making them difficult to scale, especially where multiple analytes are of interest.

Regardless of the electrochemical analytical method, a common theme in the above-mentioned high-sensitivity methods is to maximize the frequency of redox turnover (which is proportional to scan rate) to allow for the largest electron-transfer rate at the electrode. To measure analytes at the fM-level, the discrete nature of electron transfers, rather than a continuous current, becomes an important consideration. Thus, the shot noise due to the quantized nature of electrons, irrespective of electrochemical analytical method, limits measurability. Analyzing 44 electrochemical experiments from various laboratories that directly or indirectly (for example, by spectroscopy) measured charge transfer at the electrode, it was estimated that a practical limit of quantification (including background) is ∼2,100 electrons in the measurement time window, τ. Here we show a method to measure ∼9.38 × 10³ electrons in τ ∼0.02 s on a macroscale (16 mm²) electrode. The signal is shown to be 4.5-fold above the shot noise limit with background, making the method robust at its limits. The method is scalable with practical electrode area and sample size, and importantly, multiple analytes on a monolithic electrode can be measured.

Experimental Section

Materials

All chemicals used in this study were obtained from commercial suppliers and used without further purification unless otherwise noted (see SI, Section S1). Nuclease-free, molecular biology-grade water (HyClone, Fisher Scientific) was used to prepare all solutions used in the studies, including SEED and CV. The 1 M and 1 mM K₃PO₄ buffers (Fisher Scientific) were prepared in-house. All ssDNA oligonucleotides were purchased from Integrated DNA Technologies (IDT, USA). The anchor, A_n is an ssDNA with a thiol modification at the 5′ end and a flexible −[(CH₂)₆]– spacer to facilitate binding to R_p, and the reporter, R_p a ssDNA was functionalized with MB at the 3′ end (see SI, Section S2, Table S1). All oligonucleotides were handled and stored according to the manufacturer’s recommendations to maintain stability and prevent degradation prior to experimentation (see SI, Section S1).

DNA Chip Fabrication

Silicon substrates (1 cm × 1 cm, SiO₂/Si) containing three patterned gold (Au) electrodes (300 nm thickness; 1 × 8 mm) were used for device fabrication (see SI, Section S1). The metallization was performed in-house using standard microfabrication methods of photolithography and sputtering. Each chip was sequentially sonicated for 5 min in acetone, followed by rinsing with deionized (DI) water and ethanol. The Au electrodes were then subjected to wet chemical cleaning using freshly prepared piranha solution (sulfuric acid: hydrogen peroxide= 3:1 v/v) to remove residual organic material, followed by a DI water rinse and compressed air drying through a 0.2 μm filter (Millipore). Caution: Piranha solution (concentrated H₂SO₄/H₂O₂, 3:1 v/v) is a strongly oxidizing and exothermic mixture that reacts violently with organic materials. It must be prepared and handled in a chemical fume hood using appropriate personal protective equipment (face shield, acid-resistant gloves, lab coat). Never add piranha solution to organic solvents. Dispose of cooled waste according to institutional hazardous waste protocols.

DNA Immobilization

Prior to immobilization of the ssDNA anchor, the chip was exposed to oxygen plasma (60 W, 2 min) to remove residual organic species, followed by hydrogen plasma (80 W, 2 min) to reduce the electrode surface. Approximately 100 μL of a solution of thiol-modified ssDNA anchor at concentrations ranging from 1 μM to 10 μM was dispensed onto individual electrodes and was allowed to incubate for 120 min at 25 °C to allow the DNA to immobilize via an Au–S bond. The sequence is given in SI, Section S2, Table S1. The chip was maintained in a humid environment to prevent the water from evaporating.

The chips were rinsed in 1 mM K₃PO₄ buffer for 5 min to remove unbound ssDNA anchor, followed by backfilling (BF) to passivate unoccupied regions of the gold surfaces. For BF, the chips were immersed in a 1 mM solution of 1-mercapto-6-hexanol (MCH) prepared in 1 mM tripotassium phosphate buffer for 10 min with gentle stirring. After backfilling, the chips were rinsed in 1 mM K₃PO₄ buffer for 5 min and air-dried.

The ssDNA reporter with MB was hybridized to the anchor (see Figure b). The sequence is provided in SI, Section S2, Table S1. To ensure that all anchors bound, the concentration of R_p was 50% higher than the A_n used for ssDNA immobilization. The reporter solution was dispensed over the electrode, followed by 2 h of incubation. The high concentration of R_p ensured that all A_n were hybridized to form the binary structure shown in Figure b. The chips were thoroughly rinsed with 1.0 M K₃PO₄ buffer to remove unbound reporter strands. The chips were then ready for subsequent characterization.

1.

Experimental setup to measure the redox of MB positioned close to the electrode. (a) Schematic of the SEED setup showing the laser beam incident on the Au working electrode (WE) in solution with Ag/AgCl as the reference electrode (RE) and the counter electrode (CE) being the Pt-coated chamber; the details of the optics (i.e., “to detector”) are described in SI, Figure S1a and b. Adapted with permission under a Creative Commons [CC BY] from [6346030931517]; Copyright [2021] [Wiley-VCH]. (b) The binary DNA system to position MB close to the electrode. (c) The CV on a 16 mm² electrode (SI, Figure S1c) for [A_n] = 10 μM. (d) The corresponding raw SEED data for 3 of the 10 DC-ramp cycles (all 10 cycles in SI, Figure S2). (e) The R for all 10 CV ramp cycles is superimposed to obtain <R>; (solid line), with the superimposed R forming the error halo due to cycle-to-cycle variations. (f) R_max for three different electrodes with different coverages of A_n were measured at seven spots in the middle and close to the edges.

Electrochemical Analysis

Electrochemical measurements were performed using chips with three gold electrodes, in which two of the outer ones were functionalized with DNA probes comprising A_n followed by R_p. The chip was mounted in a home-built chamber with an approximate capacity of 0.1 mL (see Figure S1b). The 1 M K₃PO₄ buffer solution was injected into the chamber with a syringe. Air bubbles were removed using a second syringe inserted into another port on the chamber. The electrochemical analysis was performed using an Autolab potentiostat (model PGSTAT303N, equipped with a FRA32 module). The chamber had four Ag/AgCl reference electrodes (RE), Pt-coated chamber walls as the counter electrode (CE), and one of the chip electrodes as the working electrode (WE) (see SI, Figure S1b). The large area of the CE and the four RE ensured that the analysis had minimal ionic ohmic drop, even at low electrolyte concentrations. The details of the SEED setup are described in SI, Section S1. Typically, CV measurements were carried out from −0.5 V to +0.1 V versus Ag/AgCl. The scan rate v was fixed at 1 V s^–1. For AC–CV an AC potential of 0.2 V at 500 Hz was superimposed on the same CV potential (i.e., the DC-ramp).

Results and Discussion

An opto-electrochemical approach, called the Scanning Electrometer for Electrical Double-layer (SEED), has been developed to amalgamate spatial confinement and temporal averaging by optically measuring local reactions on a 20 μm spot on a macroscale (16 mm²) working electrode (WE) at high scan rates (400 V/s) (Figure a). The beam can scan over the electrode to obtain redox signals at multiple spots. Details of the SEED setup are in the Supporting Information (SI), Section S1 (Figure S1). The ∼0.1 mL chamber plated with Pt serves as the large-area counter electrode (CE), and four Ag/AgCl reference electrodes are inserted to avoid ionic potential drops at fast scan rates and to ensure current uniformity over the WE. The “model redox system” is an ssDNA anchor (A_n) immobilized on an Au electrode that is specifically bound to a reporter (R_p) labeled with the redox-active dye methylene blue (MB), and the surrounding electrode area around A_n–R_p is passivated by tethering a monolayer of the hydrophobic surfactant 6-mercapto-1-hexanol (MCH) (Figure b). A_n and R_p are 15 and 28 nucleotides (nt) long, respectively. R_p binds over the entire length of A_n. Details for sample preparation are in the Supporting Information (SI), Section S2. The coverage of A_n–R_p is controlled by the concentration of A_n used for immobilization, while R_p is in excess to ensure that all A_n strands are hybridized (see SI, Section 2). As A_n has a flexible hexamethylene chain as a spacer, the MB is ∼0.7 nm from the electrode surface, making the reaction kinetically limited with negligible capacitive current, I_CV (see Figure c). DNA brush systems composed of linear and folded conformations with a redox dye at the free-end are widely studied to develop chemical sensors to electrochemically quantify specific target nucleic acids, , proteins, − small molecules, and heavy ions.

For SEED, an AC potential of amplitude 200 mV at ω = 500 Hz (i.e., 400 V/s) is superimposed on a slow cyclic voltammetry (CV) ramp, E (at v = 1 V/s), and the differential reflectivity of a 20 μm, 2 mW He–Ne laser spot at ω is measured. As the MB (blue) reduces to LMB (colorless) and back (Figure b), the flip-flop between the two states causes the reflectivity to oscillate due to the AC potential (discussed below in eq ). The amplitude of oscillation in reflectivity is measured by a lock-in-amplifier (LIA) tuned at ω to obtain a high sensitivity, low-noise signal (see SI, Figure S1). The reflectivity oscillation, corresponding to a scan rate of 400 V/s (for ω = 500 Hz), will be highly enhanced compared to CV, as expected owing to the high-speed scan rate, but it is also prone to inevitable stochastic noise due to the short reaction time of ∼1/ω ∼ 0.002 s. By setting the low-pass filter bandwidth to ∼6 Hz (i.e., a time constant of ∼27 ms), the amplitude is averaged over roughly ten AC cycles to read an average R for every τ ∼0.02 s, i.e., where τ is the sampling or measurement time. Thus, R is an average over E ± ΔE/2, where ΔE = v/τ ∼ 20 mV. This R is subsequently referred to as the SEED signal. R peaks close to the redox potential, as shown for ∼3 cycles for better visualization (Figure d). (The complete raw data for all 10 cycles are shown in the SI, Section S3, Figure S2.) The measured R has low noise, indicating good averaging for the fast scan rate. By superimposing R for each of the 10 CV cycles, the average reflectivity amplitude, <R> as a function of E, is obtained (Figure e). The “error halo” is due to cycle-to-cycle variation. The small halo indicates negligible hysteresis and indicates that τ (i.e., the filter setting) of 0.02 s for ω = 500 Hz is appropriate. R_max is averaged over seven spots on the electrode (Figure f). The small spot-to-spot variations for low to high probe coverage indicate good sample uniformity. All R_max reported in the study are averaged over multiple spots covering the whole electrode. We note in passing that this ability to scan demonstrates the ability of SEED to perform combinatorial electrochemistry on a monolithic electrode.

To understand the origin of R, the total current (i.e., AC–CV), I_AC on the 16 mm² electrode is concomitantly measured with SEED (Figure ). The respective oxidation and reduction (envelope) peaks of AC–CV observed for both up- and down-DC- ramp indicate rapid flip-flops between oxidized (MB) and reduced (LMB) states due to the AC potential. The relative amplitude of flip-flops is maximum close to the formal potential. The large capacitance band (compared to none for CV) is attributed to the high frequency of 500 Hz. The AC–CV (envelope) peaks are aligned for both up- and down-DC-ramp indicating negligible diffusion effects, as expected due to close proximity to the electrode. Furthermore, the AC–CV envelope peaks are fairly aligned with CV peaks on the same electrode at the same ramp rate of 1 V/s, indicating that the kinetics of the MB/LMB flip-flop are fast. The ratio of AC–CV and CV peak currents of ∼0.69 mA and ∼1.54 μA, respectively, is ∼450 that is comparable to their scan-rate ratio of 400. The proportionality to scan rate (in contrast to the square-root of scan rate) further indicates that the reaction is under kinetic control rather than diffusion-controlled.

2.

Comparison of all the electrochemical signals performed on a typical electrode. Synchronized profiles of the SEED signal (red), CV ramp (green), and cyclic voltammetry current (blue) as a function of time. The gray oscillating signal represents the AC–CV current, with the dashed black lines indicating the redox envelope. The shaded region highlights the peak SEED signal correlation with the maximum redox current during oxidation and reduction.

As MB/LMB flip-flop between the blue/colorless states, respectively, at ω, the imaginary refractive index of the interface oscillates at the same frequency. The amplitude of reflectivity R, for δf fraction of redox molecules undergoing flip-flop (at a given E), is shown to be (see SI, Section S4, eq S4),

$$\mathrm{R}\approx 4.06\times {10}^{-15}\varphi \delta \mathrm{f}$$ 1

where, the coverage ϕ is the number of molecules per cm² and 4.06 × 10¹⁵ accounts for the optical constants. As CV and SEED, over 16 mm² and 20 μm spot areas, respectively, have similar signal quality, SEED is at least ∼10⁴ fold more sensitive.

The R_max corresponds to E at which the maximum turnover fraction (δf)_max occurs between oxidation and reduction, i.e., where both redox envelopes are nominally aligned (shaded region in Figure ). Thus, from eq ,

$${\mathrm{R}}_{\max }=4.06\times {10}^{-15}\varphi {(\delta \mathrm{f})}_{\max }$$ 2

As the A-R is in UP-conformation, where the distance between MB (or LMB) and the electrode is nominally fixed (except for small thermal fluctuations) and the buffer concentration (1 M) and pH are also fixed, (δf)_max should depend only on the reaction rate, independent of ϕ. Thus, from (2), R_max versus ϕ should be a straight line.

ϕ is calculated directly from CV. ϕ was varied by dispensing A_n at concentrations ranging from 1 μM to 10 μM, whereas the R concentration was 1.5-fold higher than A_n to ensure that all the immobilized A are bound to an R sequence (see SI, Section S2). Because MB is tethered close to the electrode, in a full CV cycle all molecules react. Thus, the area under the CV curve for reduction and oxidation will correspond to the total numbers of MB and LMB undergoing reduction and oxidation, respectively. Although the two areas are similar, confirming reversibility, the larger area is used to calculate ϕ. From the CV peak area, the number of MB molecules N is N = Q/2e, where Q = ∫I­(t)­dt = ∫I­(t)­dE/v is the area under the CV peak (Figure c) above the small capacitive baseline, e is the charge of the electron, and a factor of 2 accounts for the redox process being two-electron (Figure a). The area is averaged between the oxidation and reduction peaks, which are similar, typically within 5%. The average R_max is obtained by performing SEED on multiple spots covering the whole 8 × 2 mm electrode. The coverage ϕ (from CV) versus average R_max (from SEED) is linear, with a fitness of 0.923 (Figure ) leading to

$${\mathrm{R}}_{\max }=3.51\times {10}^{-16}\varphi$$ 3

3.

Correlation between R_max (from SEED) and ϕ (from CV). The blue line is linear regression constrained to zero intercept; black, red, and blue data points represent distinct experimental trials with different researchers, demonstrating reproducibility across different sensor chips; points highlighted in green squares indicate specific data shown in Figure , and the purple square shows the lowest data used in Figure a.

Three typical SEEDs and CVs (marked in Figure ) for the same electrode are compared over a range of coverage for a single DC-ramp cycle (Figure ). For ϕ = 2.5 × 10¹¹/cm² the CV peak is weak (see Figure a), while the corresponding SEED remains strong (see Figure (b)). The corresponding individual sets of SEEDs and CVs are plotted in the same graph in the SI, Section S5, Figure S3 to show that the peaks align and the error halo remains reasonably low even at low coverage.

4.

Electrochemical (CV) and optical (SEED) responses at varying surface coverages. (a) Typical cyclic voltammetry current (I_CV) for three different ϕ over a single ramp cycle, i.e., as a function of E (gray line). (b) SEED on the corresponding electrodes.

Next, we estimate the total reaction measured by SEED. From eqs and , (δf)_max = 8.65 × 10^–2, i.e., ∼8.7% of molecules flip-flop between MB and LMB redox at the peak where oxidation and reduction AC–CV envelopes nominally align. The width of a SEED peak spans 0.5 s (Figures e, and b). Thus, the total number of oxidation/reduction flip-flops measured every 0.02 s in a single cycle is p = 0.5/0.02 ∼ 25. To first order, if we assume the shape of the SEED peak to be a triangle (SI, Section S6, Figure S4), then the total fraction reacted during the SEED peak under an up- or down-DC -ramp is [(1/2)­p­(δf)_max] ∼ 1.1, which is reasonably close to 1. Therefore, the assumption is that for a fixed ϕ, (δf)_max remains constant for given scan conditions, i.e., DC scan rate and ω. We note in passing that at high coverage (δf)_max is expected to depend on ϕ due to interchain interactions. Importantly, because ϕ­(δf)_max is amount of electron transfer at the maximum turnover condition, it is proportional to the reaction rate constant, k. Therefore, R_max directly measures the relative reaction rate.

Lastly, the quantitative limit of detection (LOQ) of SEED is estimated. The lowest coverage measurable by CV was simply the smallest identifiable, robust peak (i.e., repeatable over 3–4 scans) above the capacitance background to objectively quantify Q. At the lowest measurable Q recorded by CV, corresponding to ϕ = 4.46 × 10¹⁰ molecules/cm² the SEED signal is strong (Figure a). Thus, the lowest amount of total redox measured by CV at 1 V/s on a 16 mm² electrode area is 7.13 × 10⁹ molecules. This limit of CV for macroscale electrodes, limited by the capacitance background in this study, can certainly be overcome on a ∼ 10 nm UME to measure reactions from ∼7 × 10³ molecules at a comparable scan speed to SEED at 1000 V/s.

5.

SEED measurements at the lowest points. a) CV and SEED response at the lowest recorded coverage measured by CV. b) The lowest recorded ϕ measured by SEED at three different spots on the electrode. The raw data with the error halo are shown in the SI, Section S7, Figure S5.

A central aspect of the study is that the lowest SEED (i.e., LOQ) recorded at R _max of ∼6 × 10^–6 (see Figure b) is limited by the shot-noise of the photon count: The number of photons from a 2 mW laser in a measurement time of 0.02 s (computed above) incident on the detector is ∼1.3 × 10¹⁴ amounting to a shot noise of ∼10⁷ photons. Thus, the R_max limit in the range of 10^–6, ∼10-fold above the inevitable shot noise (background) observed in Figure b is reasonable. Although the corresponding raw data have a significant error halo (see SI, Section S7, Figure S5), the signals at various spots on the electrode are reasonably reproducible (see Figure b). More importantly, the up- and down-ramp peaks of SEED are well aligned, indicating a kinetically limited reaction, which is consistent with SEED for larger R_max (see Figures e, and S3 (SI)). Below this coverage, the scan-to-scan variation is too high for a reasonable peak at a fixed position of R_max.

From eq , at the R_max limit of 6 × 10^–6, the corresponding ϕ is 1.71 × 10¹⁰ molecules/cm². With a laser spot diameter of 20 μm, the area probed by SEED is 3.14 × 10^–6 cm² corresponding to a coverage of ∼5.4 × 10⁴ molecules or ∼90 zeptomoles. At R_max, redox from only a (δf)_max ∼ 0.087 fraction of molecules are recorded; therefore, the peak of SEED corresponds to reaction from 4.7 × 10³ molecules, or 7.72 zeptomoles. Thus, although the measurability of SEED (at 500 V/s) is comparable to that of CV on a ∼10 nm-diameter UME (at 1000 V/s) of ∼7 × 10³ molecules, SEED is not (noise and background) limited by electrode size.

Juxtaposed with the shot-noise limit estimated by photon counting, the electron transport count during the reaction at the SEED limit is estimated. From the charge transport of 2 × 4.7 × 10³ electrons (where the factor of 2 is due to two electrons exchanged per reaction; Figure b) at R_max in the data sampling time τ = 0.02 s, the equivalent peak current is ∼0.075 pA. This is comparable to single-molecule electrochemistry. This count of ∼9,375 electrons at R_max is roughly 4.5-fold higher than the practical electrochemical measurement limit of 2,100 electrons, which is 2-fold above the shot noise including a 10-fold background.

Conclusions

In summary, using a dsDNA monolayer we designed a system to position a redox dye, methylene blue, close to the Au electrode with a DNA scaffold so that (i) on ramping the CV potential around the formal redox potential of MB, all the molecules undergo MB/LMB redox; and (ii) the reaction is kinetically controlled. Because the CV has an insignificant capacitive background and all MB molecules undergo redox, the total coverage ϕ is directly measured. The negligible diffusion effect is consistent with alignment of the oxidation and reduction peaks in CV and AC–CV, and the enhancement ratio of peak current in the latter is linearly proportional to the ramp-rate ratio. An opto-electrochemical method, SEED, was evaluated as a quantitative tool by performing concomitant CV and optical measurements on the same electrode. Three features are highlighted: (i) By performing well over 50 separate experiments with more than six measurements at different spots, the SEED signal is linearly proportional to local current. (ii) By comparing to the total MB/LMB redox per CV cycle, SEED is 10⁵ smaller than the CV. (iii) The peak of SEED corresponds to redox from ∼8 zeptomoles of molecules and an equivalent redox current of 0.074 pA. A simple optical analysis with experimental measurements shows, as confirmed for consistency, that the peak in SEED, R_max is linearly proportional to reaction kinetics. Technologically, by scanning the laser beam over the electrode, SEED can perform combinatorial electrochemistry on a monolithic electrode.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmeasuresciau.6c00156.

• The SEED setup and sample chamber; SEED raw data; SEED signal calculation; detailed Figure plots; calculation for (δf)_max, raw data for the lowest Rmax; and materials (PDF)

The manuscript was written with contributions from all authors. All authors have given approval to the final version of the manuscript. All authors contributed to the planning of the research and the writing of the manuscript.

This work was supported by the National Institutes of Health (R01GM157477). The funding body had no role in study design, data collection and analysis, the decision to publish, or preparation of the manuscript.

The authors declare no competing financial interest.