Single‐Response Duplexing of Electrochemical Label‐Free Biosensor from the Same Tag

Abstract

Multiplexing is a valuable strategy to boost throughput and improve clinical accuracy. Exploiting the vertical, meshed design of reproducible and low‐cost ultra‐dense electrochemical chips, the unprecedented single‐response multiplexing of typical label‐free biosensors is reported. Using a cheap, handheld one‐channel workstation and a single redox probe, that is, ferro/ferricyanide, the recognition events taking place on two spatially resolved locations of the same working electrode can be tracked along a single voltammetry scan by collecting the electrochemical signatures of the probe in relation to different quasi‐reference electrodes, Au (0 V) and Ag/AgCl ink (+0.2 V). This spatial isolation prevents crosstalk between the redox tags and interferences over functionalization and binding steps, representing an advantage over the existing non‐spatially resolved single‐response multiplex strategies. As proof of concept, peptide‐tethered immunosensors are demonstrated to provide the duplex detection of COVID‐19 antibodies, thereby doubling the throughput while achieving 100% accuracy in serum samples. The approach is envisioned to enable broad applications in high‐throughput and multi‐analyte platforms, as it can be tailored to other biosensing devices and formats.

Meshed ultra‐dense chips provide the single‐response duplexing of label‐free electrochemical biosensors. Distinguishable signatures are achieved by interrogating the same redox probe in spatially resolved areas of the same working electrode relative to distinct quasi‐reference electrodes, Au and Ag/AgCl. This spatial isolation prevents crosstalk between probes and interferences over functionalization/binding steps, representing an advantage over existing non‐spatially resolved single‐response multiplex strategies.

1. Introduction

A challenge that has gathered special attention in the biosensing community is the deployment of devices with the ability to provide massive, reliable, and accurate information at the point of care (POC), for example, patient bedside or low‐income settings.^{[ 1} , ² , ^{3 ]} Based on the REASSURED requirements,^{[ 4 ]} an ideal POC platform should afford i) real‐time connectivity and ii) ease of specimen collection. Besides, the device should be iii) affordable, iv) sensitive, v) specific, vi) user‐friendly, vii) rapid and robust, viii) equipment‐free (or relying on handheld equipment), and ix) deliverable to end‐users. The end goal for these POC tests is to contribute to lifesaving procedures by providing patients and healthcare practitioners with fast clinical information, thus assisting personalized medicine, enabling early health treatments, and improving patient prognosis.^{[ 5} , ^{6 ]} It is also worth noting that the decentralization of massive diagnostics is crucial for timely managing infectious disease outbreaks, as witnessed during the coronavirus disease 2019 (COVID‐19) pandemic.^{[ 6} , ⁷ , ^{8 ]}

To fulfill the REASSURED criteria, the diagnostic devices must also offer a reasonable trade‐off between i) the performance of laboratory‐based standard techniques (i.e., polymerase chain reaction and enzyme‐linked immunosorbent assay) and ii) the ease of operation, high throughput, and portability of rapid testing (e.g., lateral flow assays). Although many electrochemical nano‐biosensors have been devised as an effective solution to cope with POC analysis‐related issues,^{[ 9} , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ^{19 ]} this platform has not yet addressed the market needs or reached wide utilization in clinical practice.^{[ 6} , ²⁰ , ^{21 ]} Key challenges facing the development of diagnostic devices across all nine technology readiness levels (TRLs) are tightly tied to the device lifespan, prototyping, throughput (the ability to analyze hundreds of samples in a short period), and bodily fluid properties that hinder accuracy by implying fouling and cross‐reactivity issues.^{[ 6 ]}

While low‐cost approaches, such as screen/stencil and inkjet printing, offer an easy strategy to prototype electrochemical sensors, the resulting devices are amenable to poor batch‐to‐batch reproducibility and pattern‐transfer resolution.^{[ 22 ]} Accordingly, although microfabrication approaches based on photolithography and vapor phase deposition involve costly and sophisticated processes and equipment, they still stand out as the most convenient alternatives for the reproducible and large‐scale production of chips bearing from macro to microelectrodes.^{[ 23} , ^{24 ]} These methods can also deliver a reasonable final unit cost because of their high scaleup compatibility.^{[ 25 ]} Regarding throughput, platforms with the ability to provide high‐frequency monitoring are of pivotal relevance not only to increase the testing capacity toward massive diagnostics,^{[ 26} , ²⁷ , ²⁸ , ^{29 ]} but also to ensure early and accurate diagnosis and assist in the diagnosis of diseases with similar symptoms (e.g., Zika and Dengue virus) as they can be used in the rapid detection of multiple biomarkers.^{[ 30 ]}

In this scenario, multiplex systems are powerful alternatives to boost throughput by providing the analysis of multiple targets from one sample and/or the same analyte from different samples simultaneously or over a single response.^{[ 31 ]} The multiplex electrochemical biosensors are based on i) multi‐ or ii) one‐electrode strategies,^{[ 32} , ^{33 ]} which can be classified as spatially resolved and multiple‐label tests, respectively, in the context of multi‐analyte detection.^{[ 34 ]} In the multi‐electrode case, multiple working electrodes (WEs) with spatially separated sensing areas have been used to detect various biomarkers with the aid of multichannel potentiostats.^{[ 31} , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ^{47 ]} In contrast, the one‐electrode platforms have enabled single‐response electrochemical multiplexing (SERM) through cheap, handheld one‐channel workstations. To provide SERM, the distinct targets captured on the same WE are labeled with specific signaling tags (e.g., quantum dots, redox probes, metal ions, and nanoparticles). Then, these elements are interrogated with a single measurement of pulsed voltammetry, creating resolved current peaks (without overlap) tied to each sandwich‐type binding and ultimately rendering the detection of a panel of biomarkers along a single voltammetric scan.^{[ 29} , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ^{52 ]} Distinguishable signatures from one WE have also been reached by reagentless systems.^{[ 53} , ⁵⁴ , ^{55 ]}

Although valuable, the one‐electrode multiplex methods are prone to two limitations, that is, i) the occurrence of crosstalk between the different tags and ii) the absence of spatial isolation, with the modification of the same WE with different bioreceptors.^{[ 26} , ²⁷ , ²⁸ , ^{29 ]} To date, the crosstalk can be detrimental to the resolution of the current peaks over the voltammetric scan, therefore requiring a meticulous selection of the signaling tags to resolve the electrochemical signatures. Conversely, the chemical modification of the same WE with various recognition elements may harm the reproducibility and specificity of the devices due to steric effects and further interferences from other species. Further, the single‐response multiplexing of electrochemical biosensors in the absence of labels remains a challenge.^{[ 33 ]} Based on adding the probe to the solution, label‐free biosensors are an attractive approach for POC diagnostics as they render the detection of binding events without labeling tags.^{[ 11} , ¹⁶ , ³³ , ⁵⁶ , ⁵⁷ , ^{58 ]}

To further advance high‐throughput diagnostics in decentralized settings, we here address scalable, reproducible, and low‐cost ultra‐dense chips that could provide SERM from one WE, but spatially resolved sensing areas that were interrogated with different quasi‐reference electrodes (QREs). In this way, the duplexing of label‐free electrochemical biosensors could be addressed from a single square wave voltammetry (SWV) experiment and only one signaling tag. Specifically, by leveraging the unique design of vertical, meshed two‐electrode cells, the probe ferro/ferricyanide ([Fe(CN)₆]^3−/4−) was analyzed on two isolated regions of the same Au WE finger in relation to distinct QREs, that is, Au and Ag/AgCl (these electrodes were in electric contact with each other). Therefore, the binding events taking place in these areas could be monitored along a single SWV scan in an unprecedented mode by reading out distinguishable SWV current peaks versus Au (0 V) and Ag/AgCl (+0.2 V) QREs.

Duplex analyses from spatially separated sensing surfaces of the same WE lead to advantages over the existing one‐electrode methods by i) avoiding the risks of crosstalk between the tags (as shown here) and ii) allowing the functionalization of WEs individually, as aforesaid.^{[ 29 ]} The spatial separation of the sensing areas further allowed iii) the inedited single‐response duplexing, to our knowledge, of a traditional label‐free biosensor (with a probe in solution), iv) the concurrent detection of a single biomarker from two samples, which was leveraged to increase twice the throughput for COVID‐19 screening, and v) duplexing with a single tag. As proof of concept, a biosensor with peptide as a recognition element^{[ 59 ]} was engineered to detect anti‐SARS‐CoV‐2 spike (S) protein antibody (IgG^S). We could use only the reversible pair [Fe(CN)₆]^3−/4− as a probe, which is low‐cost^{[ 60 ]} and tends to deliver sensitive blocking strategy‐based label‐free immunoassays.^{[ 33 ]} Using a single tag can also save time and costs, whether throughout the TRLs for technology deployment or practical applications.

One should also emphasize that the adaptation of SERM in vertical ultra‐dense chips prevents two limitations usually noted in the multi‐electrode methods, that is, the difficulty of connecting wire leads to each of all the electrodes on a small‐size device^{[ 33 ]} and the increased consumption of clinical samples.^{[ 61 ]} These issues were avoided as the chips exhibit a reduced number of contact pads to the external equipment and small sensing areas, respectively. In practice, droplet volumes as low as 10 µL were sufficient to ensure stable analyses (without solvent evaporation effect) at room temperature for a few minutes.

2. Results and Discussion

2.1. One‐Tag SERM Principle

We have fabricated scalable, reproducible, and low‐cost devices from 3D film engineering. The sensors showed recessed WEs vertically insulated from QREs by a dielectric layer of the negative photoresist SU‐8. Briefly, the WEs consisted of Au thin‐film fingers on a glass wafer, whereas the QREs relied on Au thin‐film and on‐adhesive Ag/AgCl ink fingers. Three or more QREs were patterned onto the top of SU‐8, creating a mesh‐like layout with the underneath WE fingers and then generating an array of sensors along each WE. For the application to serum samples, we built up a prototype carrying a portable one‐channel potentiostat and switches to select droplet (10 µL) sensors with individual or common QREs (for the duplex assays), as later discussed.

To address the duplexing of a label‐free biosensor, spatially separated sensing areas of the same WE were interrogated with a single SWV analysis in relation to a pair of QREs, that is, Au and Ag/AgCl, which were in electric contact with each other. Based on this design, affinity biomolecular sensors were adapted into SERM format from a single redox probe, [Fe(CN)₆]^3−/4− (henceforth referred to as Fe^3+/2+), with two distinguishable peaks being reached as a function of Au and Ag/AgCl QREs (Figure  1 ). The biosensors were modified with peptide (receptor to capture IgG^S) and glycine (blocking agent) through simple drop‐casting steps. From preliminary tests (data not shown), the use of Au, carbon (C), and nickel (Ni) as QREs resulted in SWV peaks for Fe^3+/2+ at similar potentials (≈0 V), whereas this potential was shifted to around +0.2 V when using Ag/AgCl QRE. Thus, sensors with common Au and Ag/AgCl QREs could yield resolved peaks along an SWV scan, allowing us to track the recognition events taking place in two areas of WE by collecting the current changes at 0 (vs Au) and +0.2 V (vs Ag/AgCl). It is proposed that these data are driven by the work functions (Φ) of metals (electronic property that alters the interface electrochemical potential),^{[ 62} , ^{63 ]} as endorsed by the distinct Φ of Au, C, and Ni (5 to 5.4 eV) in relation to Ag (4.2 eV).^{[ 64} , ^{65 ]}

Figure 1.

One‐tag electrochemical duplexing. Schematic representation considering Fe^3+/2+ probe and peptide‐tethered biosensors to detect IgG^S. The spatially resolved binding events between the peptide and IgG^S on WE, which are illustrated in an enlarged view, can be assessed from distinguishable SWV peaks due to the distinct Φ of Au (Φ _Au) and Ag/AgCl (Φ _Ag) QREs. The directions of WEs and QREs are indicated by red and blue arrows, respectively. Orange circles indicate the tag and an image shows the charge‐transfer pathway of Fe^3+/2+, that is, OSP. M means metal. Scale bar: 6 mm.

The current peaks are expected to reduce with the concentration of IgG^S biomolecules bound to the peptide, as these affinity binding events impair the approaching dynamics of the redox probe with the electrode surface,^{[ 5} , ¹⁶ , ^{66 ]} thus slowing down the electron transfer kinetics. While the electron‐transfer reactions with the Fe^3+/2+ tag involve an outer‐sphere pathway (OSP; they occur at the outer Helmholtz plane, being separated from WE by at least a layer of water, see Figure 1),^{[ 67} , ^{68 ]} this redox probe shows a certain inner‐sphere pathway (ISP) character (reactions happen at the inner Helmholtz plane; Figure S1, Supporting Information). Hence, steric hindrance‐based readouts in label‐free biosensors with Fe^3+/2+ are sensitive to the blocking effects of heterogeneous reactions, with current suppressions being reached as a function of the analyte anchoring on the WE surface.^{[ 33} , ^{68 ]}

2.2. Fabrication and Integration Density

Fabrication of the device consisted of three steps of photolithography and two steps of thin film deposition on a glass wafer (75 mm × 35 mm; Figure  2A). Such methods ensure the scaleup of the chips, which remains a challenging milestone in advancing the TRL of diagnostic tools.^{[ 6 ]} The electrical isolation between the bottom WEs and top QREs was provided by a SU‐8 layer (4 µm) that also delimited the detection areas (circular apertures: 4 × 45 or 1 × 800 µm). To afford SERM, adhesives coated with Ag/AgCl ink were fixed on the chips to act as QRE along with the Au fingers. In practice, only a pad was utilized for this pair of QREs that were in electric contact with each other with the aid of an Ag ink. While the fabrication of planar electrodes only allows bidimensional (2D) integration, our 3D integration yielded the engraving of grid‐patterned arrays, with overlapped layer‐by‐layer metal fingers. In practice, each crossing point of the orthogonal electrodes formed a two‐electrode cell. This design significantly reduced the number of electrical contact pads and conductive lines, thus creating ultra‐dense chips by increasing the overall area to attach more sensors. It is also worth noting that a lower number of pads and lines simplifies the connection of cells to equipment, as discussed above, and minimizes the parasitic capacitance,^{[ 69 ]} thus preventing electric crosstalk.

Figure 2.

Vertical meshed devices. A) Major microfabrication steps, covering the fabrication of WE fingers on a wafer, their coating with SU‐8, and the patterning of top QREs. The detection areas and spacing between WE and QRE were photolithographically defined by SU‐8. The layers composing the chip and the cross‐sectional view of a sensor are also illustrated. Apart from the 4 × 45 µm as considered in these images, 800 µm WEs were also tested. B) Chips with 830 (1) and 48 sensors (2), along with the amplified stereoscopy image of a sensor with 4 × 45 µm WEs (3). Scale bars: 10 mm (1) and 50 µm (3). C) Characterization of the plasma‐induced effects on WE and SU‐8 by AFM‐IR. Picture showing chips exposed to plasma (1), image by scanning electron microscopy (SEM) of a sensor with the four areas of WE (yellow) and SU‐8 (white) scrutinized by AFM‐IR being highlighted (2), and average IR spectra on WE (3) and QRE (4) before (red) and after (cyan) plasma exposition. Scale bar: 50 µm (2). Contact angles for water drops after 0, 15, and 30 days exposed to plasma are also shown, along with an SEM image of plasma‐treated SU‐8 (scale bar: 100 nm). D) Chip with 48 sensors and illustration of a generic potentiostat circuit, with the contact pads of Au QRE and WE to trigger the highlighted under‐droplet (10 µL) sensor. Scale bar: 6 mm. E) Preliminary electrochemical CV analyses of 1 mmol L⁻¹ Fe^3+/2+ (1) to assess reproducibility using 4 × 45 µm WEs (2). The colors indicate the anodic current peaks (unit: nA) collected by the 48 sensors (2). The WEs and QREs are indicated by numbers 1–16 and letters a–c, respectively. This representation was used throughout the manuscript.

The high integration density of the meshed vertical electrochemical chips (MECs) is essential to dropping the manufacturing cost as it is reduced with the number of sensors per wafer. The unit final costs were estimated at USD 0.56 and USD 0.04 for 60 and 870 cells per wafer (Figure 2B), respectively. Hence, the prototyping described here merges reproducibility and high resolution (ability to pattern ultramicroelectrodes) with low unit cost, aiding the development of sensors through their TRLs. Importantly, while crossbar electrode fashions have already been addressed in the literature, these arrays have incorporated only WEs to perform redox cycling detection.^{[ 70} , ^{71 ]} Thus, in contrast with the fully‐integrated MECs, these methods require external QRE and CE that inhibit their scale‐up and practicity.

Because of their ability to pattern micro‐scale electrodes, the standard microfabrication processes can deliver the production of high‐density arrays, with thousands of electrodes, even from planar, 2D architectures.^{[ 72} , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ , ^{77 ]} Hence, as a manner to critically investigate the integration capacity of sensor designs, the reduction in the number of conductive lines that the device provides over typical 2D devices (one line for each electrode) is proposed here as an assessment yardstick. This parameter boosts the integration density by increasing the wafer active area, as aforesaid. Regarding the vertical two‐electrode chips, this reduction was roughly 80%, 84%, and 96% for the chips with 48, 60, and 870 sensors, respectively (Figure S2, Supporting Information). Naturally, thousands of these stacked cells (but with micro‐scale patterns) could be prototyped on a single wafer to ensure high spatial‐resolution assays. Utilizing high‐density 2D WEs, with either external reference and counter electrodes^{[ 72 ]} or fully‐integrated cells,^{[ 73 ]} and a customized instrumentation, this type of analysis has been used for the electrochemical imaging of a single sample. Here, conversely, 48 and 60 on‐chip sensors were used to deliver the analysis of nearby, isolated 10 µL droplets by simply dropping the solutions. This volume proved to be sufficient to avoid evaporation interferences in tests for a few minutes at room temperature, as aforesaid. Alternatively, microfluidics^{[ 78 ]} could be used to ensure stable small‐volume analyses on an increased number of on‐chip cells. Next, we address surface and electrochemical characterization studies using all‐Au chips (Sections 2.3 to 2.6). Chips with Ag/AgCl QRE are discussed in Sections 2.7 to 2.9.

2.3. Shelf‐Life and Plasma Effects

The as‐fabricated chips were subjected to Argon (Ar) plasma for 5 min to remove photolithographic organic residues (descumming step) from the electrode surfaces and increase the sensor wettability. This treatment proved to be key for the reproducibility of sensors.^{[ 79 ]} Just after exposure to the plasma, the sensor wettability increased pointedly as indicated by the reduction in contact angles from 72° ± 5° (before; n = 3) to 14° ± 1° (after plasma; Figure 2C and Figure S3, Supporting Information). Then, the wettability gradually decreased over the sensor storage, with an average contact angle of 50° ± 8° (n = 18) being reached over 30 days. The SWV peaks for 2 mmol L⁻¹ Fe^3+/2+ employing 4 × 45 µm WEs and Au QREs also increased noticeably after Ar plasma treatment (see Figure S3, Supporting Information), as expected. Despite the loss of wettability, the chips had a shelf life of at least 25 days after plasma exposure. In practice, the peaks remained nearly the same after this period (157.9 ± 1.7 nA), while dropping by 12% after 30 days.

The plasma‐induced descumming of WE and SU‐8 oxygenation processes was confirmed by distinct characterization techniques. In particular, atomic force microscopy (AFM) coupled with infrared nanospectroscopy (AFM‐IR) allowed us to make a punctual chemical characterization of four regions (50 × 50 nm) of SU‐8 and WE. The descumming was confirmed through the decrease in peak intensities at 1732 cm⁻¹ which was attributed to the C═O bond stretching. The ratios between the intensities tied to carbonyl (C═O stretching) and aromatic groups (C═C stretching at 1505 cm⁻¹)^{[ 80} , ⁸¹ , ⁸² , ^{83 ]} decreased by 62% after Ar plasma treatment (see Figure 2C and Figure S4, Supporting Information).

The gain in sensor wettability is likely due to the formation of reactive species on SU‐8 after the action of Ar plasma.^{[ 83 ]} It is expected that these species lead to the generation of oxygenated groups on SU‐8 surfaces when placed in contact with oxygen and water present in ambient air.^{[ 80 ]} AFM‐IR analyses also confirmed such plasma‐promoted oxygenation, which revealed an increase in carbonyl peak intensity. The ratios of C═O/C═C signals increased by 108% (see Figure 2C and Figure S4, Supporting Information). We also observed the incidence of a peak at 1666 cm⁻¹ that is tied to the stretching of the C═C bond in alkenes, along with a drop in peak intensity at 1606 cm⁻¹, associated with the C═C stretch in aromatics.^{[ 83 ]} The AFM‐IR data agreed with Kelvin probe force microscopy (KPFM; Figure S5, Supporting Information) and X‐ray photoelectron spectroscopy (XPS; Figure S6, Supporting Information) experiments.

The Ar plasma also altered the SU‐8 topography (see Figure 2C and Figure S7, Supporting Information), whose root mean square (RMS) roughness boosted from 1.5 ± 0.5 to 2.9 ± 0.2 nm (n = 3). The treated dielectric presented a particulate surface, which was likely caused by the processes of crosslinking (recombination of structures of lower molecular mass) and physical etching.^{[ 80 ]} Conversely, WE and QRE did not undergo significant modifications in RMS after Ar plasma exposure, ranging from 1 ± 1 nm up to 1.5 ± 0.6 nm (WE) and 1.4 ± 0.5 to 1.6 ± 0.2 nm (QRE).

An additional gain provided by the Ar plasma is the regeneration of the WE surface after biofouling. Specifically, the exposure of 4 × 45 µm WE (Au as QRE) to fetal bovine serum (FBS) for 30 min first led to a 52% drop in SWV peaks for 4 mmol L⁻¹ Fe^3+/2+ (Figure S8, Supporting Information). Then, while the plasma treatment for 5 min could recover 93% of the initial peak, this signal was integrally recovered after incubating the passivated WE in plasma for 10 min. This property is important because it allows the reuse of the biosensors, minimizing the environmental burden derived from chip disposal.

2.4. Reproducibility

The MECs presented high reproducibility. For 48 sensors on a glass wafer with 4 × 45 µm WEs and Au QREs (Figure 2D), the average anodic current peak by cyclic voltammetry (CV) for 1 mmol L⁻¹ Fe^3+/2+ was 43 ± 1 nA (n = 192) and the relative standard deviation (RSD) was found out to be 2.4% (intra‐wafer precision; Figure 2E). For five wafers, these parameters were, respectively, 42 ± 3 nA (n = 352) and 4.7% (inter‐wafer precision; Figure S9, Supporting Information). Sensors bearing 800 µm WEs also had good reproducibility, showing average anodic current peak and RDS of 6.1 ± 0.7 µA (n = 125; 5 sensors of five wafers) and 2.4% (inter‐wafer precision). Such low deviations agree with the high reproducibility that was reached throughout this work. Specifically, more than 2000 sensors were utilized by distinct users in two potentiostats over the last 20 months, confirming the reliability of the microfabrication method.

Importantly, because no passivation layer was used to delimit the exposed region of the QREs (see Figure 1), their areas depended on the contact angles of the droplets added to the sensors (10 µL). In this regard, these areas change over time due to the alterations in SU‐8 wettability, as discussed before (see Figure S3, Supporting Information). Nonetheless, this variation did not harm the reproducibility of the experiments, which was satisfactory as aforesaid. Alternatively, perforated adhesives, 3D‐printed wells, or microfluidics could be fixed on the chip to delimit the QRE area.

It is also worthwhile noting that, as verified in two‐electrode cells,^{[ 62} , ^{63 ]} the QREs not only were in charge of keeping the potential constant but also maintained the overall system neutrality, which is the function of the counter electrode in three‐electrode sensors. Significantly, the requirement for using two‐electrode cells was met as the potential peaks versus Au QREs remained stable throughout this work. Specifically, this data shows that the faradaic currents generated in the Au WEs did not disturb the Au QRE potential.

2.5. Serial Analyses

Apart from multiplexing, the convenient accomplishment of sequential analyses on a single, fully‐integrated electrochemical device using a one‐channel potentiostat can also provide high throughput.^{[ 84 ]} By exploiting the ultra‐dense array of the MECs, these chips can offer the combination of such strategies to further boost testing capacity, as preliminarily assessed here. Amperometric analyses of [Fe(CN)₆]⁴⁻ using 4 × 45 WEs confirmed the absence of crosstalk when dropping solutions onto different WEs along the same QRE. Conversely, interferences from the droplets on other QREs were avoided by grounding these electrodes (Figure S10, Supporting Information), as expected since the QREs assumed the function of applying input potential to the WEs, as aforesaid.^{[ 25 ]} In this way, serial tests of samples added to all the cells could be made simply by measuring the solutions onto a specific QRE while grounding the other QREs.

2.6. Mass Transport, Redox Reactions, and Reversibility

CV analyses of 1 mmol L⁻¹ Fe^3+/2+ were made at different scan rates (υ; 5 to 1000 mV s⁻¹) to scrutinize the diffusion regimes and their reversibility (Figure  3A). The measurements were first performed using 4 × 45 µm WEs as these microelectrodes are supposed to undergo a stronger impact of the SU‐8 passivation layer over the macroelectrodes (800 µm WEs), thus allowing us to get a more comprehensive understanding of the recessed sensors. The semi‐infinite linear diffusion was prevalent, with the faradaic current peaks increasing linearly with υ ^1/2, as predicted by the Randle–Ševčík equation.^{[ 67 ]} Conversely, slow analyses (υ < 25 mV s⁻¹) led to an expansion of the depletion layer and, then of the diffusion layer thickness (δ) that appeared to generate a quasi‐stationary regime (δ > WE radius). In this case, we noted the occurrence of radial diffusion as supported by the maintenance of a stationary limiting current.^{[ 85} , ⁸⁶ , ^{87 ]}

Figure 3.

Electrochemical analyses to investigate mass transport on the vertical chips and their reversibility. A) CV analyses of 1 mmol L⁻¹ Fe^3+/2+ using 4 × 45 µm WEs and varying υ to assess the diffusion regimes. Generic illustrations of the semi‐infinite linear diffusion (1) and quasi‐stationary regimes (2), voltammograms at different υ as highlighted (3), and plots of anodic (I _a) and cathodic (I _c) current peaks versus υ ^1/2 (4). Inset shows a stereoscopy image of the sensor (3); scale bar: 50 µm. The currents (4) are related to linear (I _al and I _cl) and quasi‐stationary regimes (I _aq and I _cq). B) Nyquist plots to 1 mmol L⁻¹ Fe^3+/2+ using 4 × 45 µm WEs. The semicircles are hypothesized to represent redox reactions on Au WE (*) and SU‐8 (**). Z″ and Z′ mean imaginary and real impedances, respectively. In the equivalent circuit, CPE 1 and CPE 2 are constant phase elements that express the electric double layer capacitances on Au and SU‐8, respectively; R _Ω means uncompensated resistance. C) SECM‐based current mapping of 1 mmol L⁻¹ FcMeOH over the sensor with 4 × 45 µm WEs (1) and approach curves on Au, glass, and SU‐8 (2). The color bar means the currents in nA (1), whereas the current values (i _i) in (2) were normalized by the stationary state current (i _ss). Insets illustrate the phenomena noted in each case (2). D) SICM plots obtained in bulk and SU‐8 surface to 10 mmol L⁻¹ KCl using 4 × 45 µm WEs. Inset illustrates the ensuing interface phenomena for negatively charged SU‐8, with red and blue circles indicating K⁺ and C⁻ ions, respectively. E) CV analyses of 1 mmol L⁻¹ Fe^3+/2+ utilizing 800 µm WEs at specific υ values as highlighted. Inset shows a stereoscopy image of the sensor, with the bars stressed by arrows signaling the SU‐8 ring width. The dimension bar means 150 µm. F) Nyquist plots to 1 mmol L⁻¹ Fe^3+/2+ using 800 µm WEs. The same circuit in (B) was applied here.

Charge‐transfer reactions were also revealed to occur on SU‐8 as signaled by electrochemical impedance spectroscopy (EIS) and scanning electrochemical microscopy (SECM) analyses. Nyquist plots recorded by EIS to 1 mmol L⁻¹ Fe^3+/2+ exhibited two semicircles (Figure 3B), suggesting the incidence of redox reactions onto two surfaces. The semicircle at high frequencies, with low charge‐transfer resistance (R _ct), was attributed to reactions on Au WEs, while the semicircle at low frequencies, with meaningfully high R _ct, was hypothesized to be tied to reactions on the resistive SU‐8 surface, as confirmed by SECM analyses. Specifically, we mapped the faradaic currents yielded by 1 mmol L⁻¹ ferrocenemethanol (FcMeOH) on 4 × 45 µm WE, SU‐8, and QRE (Figure 3C) from SECM. SU‐8 produced the lowest charge‐transfer rates, as expected, but such a parameter was not negligible. In fact, from approach curves, SU‐8 showed a transient behavior between Au (conductor) and glass (insulator), signaling the incidence of redox reactions on this dielectric. The incidence of these electrochemical reactions is likely due to the Ar plasma‐induced breakage of SU‐8 bonds, generating occupied electronic energy levels (i.e., HOMO molecular orbitals) more negative than the redox reaction potential (0 V). In this scenario, as proposed by Liu and Bard,^{[ 88 ]} there is the formation of available electrons on the polymer that can spontaneously be transferred to reducible species in solution. In line with our hypothesis, analyses of scanning ion‐conductance microscopy (SICM) disclosed the existence of an overall negative net charge on SU‐8, as discussed next. One should also emphasize that the redox reactions on SU‐8 are not expected to affect the SWV current signals because the available charge in these polymer surfaces is small, as indicated by EIS (i.e., high R _ct; see Figure 3B).

The negative surface charge of SU‐8 was confirmed through ionic current analyses of 10 mmol L⁻¹ KCl using SICM. From the attained approach curves (Figure 3D), the ratio between the negative and positive currents was 1.29, indicating that SU‐8 carries a negative charge (further discussion is presented in the Supporting Information).^{[ 89 ]} In our meshed sensors, the existence of an electrically charged dielectric can be detrimental to the charge‐transfer kinetics as it is amenable to repelling or attracting the probes, especially to its surface (top of the sensor), inhibiting their transport toward the recessed surface of WE. Indeed, the peak‐to‐peak separations (ΔE _p) in the CV scans (see Figure 3A) were ≈170 mV, uncovering an unexpectedly poor reversibility of the system (Fe^3+/2+ on Au WEs) that is supposed to yield empiric ΔE _p values of ≈60 mV.^{[ 87 ]}

CV and EIS analyses of 1 mmol L⁻¹ Fe^3+/2+ were then made using 800 µm WEs. As expected, only the semi‐infinite linear diffusion regime was noted throughout the window of υ (Figure 3E), and the effect of SU‐8 on the system reversibility was reduced since the WE/SU‐8 size ratio increased; in practice, ΔE _p dropped to 83 mV. The same prior equivalent circuit fitted the Nyquist plots well (Figure 3F), showing a low but still‐existing influence of SU‐8 on the reversibility of the 800 µm WEs.

2.7. One‐Tag SERM: Preliminary Analyses

Next, we sought to evaluate the SERM format through SWV duplex assays of 2 mmol L⁻¹ Fe^3+/2+ with common QREs and 800 µm WEs, as these macroelectrodes provided redox kinetics higher than the 4 × 45 µm WEs. Adhesive coated with Ag/AgCl ink and Au fingers acted as QREs along a single SWV scan for ensuring one‐tag duplexing (Figure  4A). These analyses led to similar signals compared with the current peaks from individual assays (with only a single droplet on Au or Ag/AgCl QRE; see Figure 4A). Using a single WE to detect spatially resolved redox activities of the same probe was also assessed by the finite‐element method (FEM). In line with the experimental results, these simulations indicated the absence of crosstalk between the charge‐transfer events around the Fe^3+/2+ current peaks (vs Au, Ag/AgCl; Figure 4B), meaning that the faradaic contributions from each sensing area can be assessed individually after a single, duplex measurement.

Figure 4.

SERM analyses with 800 µm WEs and QREs consisting of Au and Ag/AgCl. A) Chip with the same probe, that is, 2 mmol L⁻¹ Fe^3+/2+, dropped on WEs versus Au (orange) and Ag/AgCl QRE (cyan asterisk) to afford SERM (1) and duplex SWV scan (red line) (2). The colored peaks mean the individual analyses using either Au or Ag/AgCl QREs, as stressed. Scale bar: 5 mm (1). B) FEM simulations for 2 mmol L⁻¹ Fe^3+/2+ and SERM format (1) and current densities at around the peaks of Fe^3+/2+ versus Au (2) and Ag (3). C) CV analyses and analytical curve for Fe^3+/2+. CV scans attained by duplex assay (red line) and individual analyses (colored scans) versus Au or Ag/AgCl QRE, as indicated (1), along with duplex SWV scans to different contents (unit: mmol L⁻¹) of probe (2). Inset shows the analytical curves for current densities (j: mA cm²) versus Au QRE (orange) and Ag/AgCl QRE (cyan) (2).

Interestingly, as noted empirically and via FEM simulations (Figure S11, Supporting Information), the chip failed to render resolved peaks when interrogating a single droplet on both QREs. In this case, only the response versus Ag/AgCl QRE was collected, disclosing that spatial resolution is mandatory for the one‐tag duplexing. While more studies are required for a comprehensive understanding of this phenomenon, it is proposed to be caused by the higher conductivity of Ag over Au. In this sense, as the contact pad of the pair of QREs is a type of junction (or node) connecting two conductive lines, the input currents in this pad are supposed to be completely drained into the Ag/AgCl finger.

We then interrogated the electrodes with individual CV tests to investigate the influence of QREs over reversibility (Figure 4C). The use of Au QRE led to a more reversible system for 2 mmol L⁻¹ Fe^3+/2+ over Ag/AgCl, with ΔE _p data of 84.5 (Au) and 133.1 mV (Ag/AgCl). This sluggish redox kinetics when using Ag/AgCl is probably because of the Ag oxidation along the cathodic scans, limiting the currents generated on WEs. This oxidation is expected to happen as the QREs also apply input potentials to WEs, as previously discussed.^{[ 90 ]} Duplex experiments were next conducted. Consistently with the prior SWV data (see Figure 4A) and FEM simulations (see Figure 4B), these duplex analyses led to the same ΔE _p as the values achieved by the prior individual analyses (see Figure 4C).

Because of the Ag oxidation, the assays with Ag/AgCl QRE are supposed to generate a sensitivity lower than the Au‐based analyses, as it was indeed verified for duplex SWV analyses of Fe^3+/2 at increasing concentrations (0.8 to 4 mmol L⁻¹; n = 4). This parameter diminished from 1 (Au) to 0.8 mA cm² mmol⁻¹ mL (Ag/AgCl; see Figure 4C), whereas the limits‐of‐detection (LODs) were 4.7 (Au) and 5.2 µmol⁻¹ mL (Ag/AgCl). As expected, individual measurements yielded similar results (Figure S12, Supporting Information). EIS analyses also could confirm the sluggish kinetics provided by Ag/AgCl QRE, which led to an R _ct (4 kΩ) higher than the value when utilizing Au QRE (1.4 kΩ) to individually assess 2 mmol L⁻¹ Fe^3+/2+ (Figure S13, Supporting Information; n = 4). The same prior equivalent circuit fitted well with these Nyquist plots (see Figure 3). For the duplex tests, R _ct (7.1 kΩ) was only 1.3‐fold higher than the linear combination of the R _ct values attained by the individual measurements (5.4 kΩ).

As abovementioned, the Au thin film provided stable QREs in the two‐electrode cells. This stability was confirmed via open circuit chronopotentiometric analyses of 2 mmol L⁻¹ Fe^3+/2+ (100 µL) for 1 h. In this case, we used the Au QRE as a WE, a commercial reference electrode of Ag/AgCl, and platinum wire as a counter electrode. Compared with its initial data, the open circuit potentials (E _OCP) changed by only 1 ± 0.9% (n = 3) for assays after 1, 2, and 60 min (Figure S14, Supporting Information). Conversely, as expected because of the Ag oxidation, the SWV peaks versus Ag/AgCl QRE suffered from a certain alteration along this work; experiments after 1, 2, and 60 min revealed an overall relative change in E _OCP of 7 ± 3% (n = 3). Despite these alterations in potentials versus Ag/AgCl QREs, this phenomenon appeared not to damage either the linearity for the SWV analysis of Fe^3+/2+ solutions (see Figure 4C and Figure S12, Supporting Information) or the inter‐wafer precision. To date, the average intra‐ and inter‐precision values were 5.6% (n = 10) and 10.8% (1 sensor from 7 wafers; n = 49). In summary, the above data support using the MECs with spatially resolved measurements over the same WE versus Au and Ag/AgCl QREs for duplexing label‐free biosensors from a single response and tag.

Finally, we investigated the possibility of using Ag thin films as QRE. These electrodes are an attractive fit for the scaling‐up of MECs, as modern vacuum‐assisted machines allow the deposition of distinct films (in our case, Au and Ag to act as QREs) on numerous chips in a serial but fast way. First, while the E _OCP for Ag varied by as much as 26.2% after 60 min, this variation was only 3.5 and 4.2% for assays after 1 and 2 min (see Figure S14, Supporting Information). Such results show that this film may yield reasonably stable potentials within fast tests. Indeed, duplex SWV analyses of Fe^3+/2 (0.8 to 4 mmol L⁻¹; n = 4) versus Au and Ag QREs led to similar sensitivities compared with the prior data with Ag/AgCl QRE (Figure S15, Supporting Information). However, one should emphasize that a challenge facing the use of Ag films as QREs is their on‐chip adhesion. Here, for example, the cracking of Ag QREs usually happens after incubating the sensor in the succeeding media to prepare the biosensor. As a solution, a new photolithographic mask should be designed in the future to avoid the contact of these solutions with Ag during biosensor preparation.

2.8. One‐Tag SERM: Bioassays

Label‐free biosensors were first developed for detecting anti‐SARS‐CoV‐2 S protein antibodies, IgG^S, through individual analyses. We relied on simple drop‐casting procedures, with the succeeding physisorption of the peptide and glycine on WEs before adding the probe for SWV analyses. From optimization studies, the concentrations were adopted as 200 ng mL⁻¹ (peptide) and 1 µmol L⁻¹ (glycine; Figure S16, Supporting Information). As described in the literature,^{[ 59 ]} this peptide shows 15 amino acids and is placed within the receptor binding domain (RBD) in the S1 subunit of the SARS‐CoV‐2 spike (S) protein. Such a region is targeted by 90% of the circulating neutralizing antibodies.^{[ 91 ]} In particular, the peptide presented SARS‐CoV‐2 IgG‐against reactivity that was revealed to be positively tied to the neutralization titers.^{[ 59 ]} Here, immunosensors bearing this peptide as a recognition element displayed high specificity, as later discussed. To date, the functionalization of the Au WE with peptide, glycine, and IgG^S was confirmed by XPS (Figure S17, Supporting Information).

Biosensors with 4 × 45 µm WEs could provide the detection of IgG^S by SWV. In comparison with the current peaks of these WEs functionalized with peptide and glycine (blank), a drop of 31% (n = 4) in response was reached after binding with 10 µg mL⁻¹ IgG^S (3 mmol L⁻¹ Fe^3+/2+ as probe). This biosensor also proved to be specific when challenged to a non‐targeted antibody at 10 µg mL⁻¹ (IgG⁻). The signal variations (n = 4) after analyses of IgG^S were ≈83% higher than the data for IgG⁻ (Figure S18, Supporting Information). Next, the use of 800 µm WEs revealed to improve the sensitivity over the 4 × 45 µm WEs. These macroelectrodes led to a peak variation of 59% for 10 µg mL⁻¹ IgG^S (Figure S19, Supporting Information). This result is likely due to the increase in accumulation (i.e., the maximum number of targets bound at the biosensor surface per time)^{[ 92} , ⁹³ , ^{94 ]} and the lower effect of the SU‐8 charge on the 800 µm WE signals than on the 4 × 45 µm WE peaks, as discussed above. Also, the use of SWV plays a relevant role by inhibiting a major increase in the capacitive current with the enlargement of WE.^{[ 67 ]} The next assays on one‐tag duplex format proceeded with 800 µm working macroelectrodes because they yielded the most sensitive analyses.

Duplex biomolecular tests were conducted in the presence of IgG⁻ (10 µg mL⁻¹) and IgG^S (1 to 40 µg mL⁻¹; n = 4). Compared with the response of bare 800 µm WEs, the modifications of the WEs with only glycine and then with glycine and peptide provided a low decrease in current peaks (<4%; Figure S20, Supporting Information). Hence, the analytical signals were not calculated over the blank (WE/glycine/peptide), but by subtracting the final current peaks from the signals of bare WEs (ΔI). This procedure makes the biosensor preparation more convenient by dispensing SWV tests during its steps. To date, ΔI could easily be obtained from the absolute height of the peaks after subtraction from capacitive current. For four wafers, the average ΔI signal (in modulus) versus Au QRE and resulting RSD were calculated as 12.6 ± 0.9% (n = 12) and 11.5% (inter‐wafer precision), respectively. Similar average parameters were obtained for the same biosensor (n = 4: intra‐wafer precision), that is, 12 ± 2% (n = 12) and 10%, respectively.

Remarkably, the biosensors could gauge two samples over an SWV scan utilizing the same tag, with current suppressions being achieved in relation to Au and Ag/AgCl QREs (Figure  5A). Once again, the biosensors did not respond to IgG⁻, and ΔI decreased with the IgG^S concentration, as predicted by the Langmuir isotherm binding curve (Figure 5B), revealing a process driven by single‐site binding events.^{[ 95 ]} A linear profile was attained with the logarithm of these concentrations for the signals versus Au QRE (0.983 R²; Figure S21, Supporting Information), resulting in a sensitivity and LOD of 6 µA µg⁻¹ mL and 8 ng mL⁻¹, respectively. The analytical curve with ΔI data as a function of Ag/AgCl QRE, conversely, showed just a reasonable linear fitting (0.868 R²), with the prior parameters being reached as 3.8 µA µg⁻¹ mL (sensitivity) and 83.1 ng mL⁻¹ (LOD). Following our hypothesis on the Ag oxidation triggering sluggish redox kinetics, the experiments with Ag/AgCl QRE led to a sensitivity lower than the Au QRE‐based assays.

Figure 5.

Bioassays to standard samples and serum samples for COVID‐19 screening. A) Multiplexed detection of 10 µg mL⁻¹ IgG⁻ and IgG^S using biosensors (Bios.) as stressed by red and cyan asterisks, respectively. B) Analytical curves for IgG^S standards utilizing Au and Ag/AgCl QREs. C) In‐house‐built handheld box (1) carrying potentiostat (2), chip connector, and manual selectors (3) of WEs (right) and QREs (left). The potentiostat can be operated via a smartphone as illustrated. Scale bar: 3 cm (1). D) Currents for bare WEs in relation to both QREs (as represented by Au and Ag/AgCl) and after incubation in positive (Pi; vs Au QRE) and negative sera (Ni; vs Ag/AgCl QRE). E) Currents for bare WEs (vs Au and Ag/AgCl QREs) and after exposure to Pi and Ni samples in two cycles of duplex analyses, that is, Pi on Au QRE and Ni on Ag/AgCl QRE (1) followed by the opposite situation (2). (F) Resulting ΔI for positive (PS) and negative sera (NS) using Au (1) and Ag/AgCl (2) QREs.

The limited sensitivity and dynamic range of the biosensors can be univocally attributed to the absence of signal amplification strategies (i.e., nanostructured microelectrodes) and the use of a small molecule as a recognition element. For instance, WEs electrochemically coated with Au nanostructured flowers could quantify IgG^S (15 ng mL⁻¹ LOD) under a wide linear window, ranging from 20 to 1000 ng mL⁻¹.^{[ 37 ]} Further, biosensors loading S protein as bioreceptors are expected to imply more sensitive signals by showing a higher number of epitopes,^{[ 7 ]} increasing the IgG^S‐anti reactivity and, therefore the accumulation.^{[ 94 ]} Nonetheless, one should emphasize that the linear range of our devices is within the clinical range of COVID‐19 IgG^S (0 to 40 µg mL⁻¹).^{[ 38} , ^{96 ]} Further, using peptide as a recognition element typically adds advantages over big‐sized biomolecules in terms of cost and stability, which play a crucial role in the commercial translation of diagnostic technologies.^{[ 6} , ⁷ , ^{16 ]} In practice, the peptide could be chemically synthesized, and it is not amenable to denaturation. This feature is supposed to extend the shelf‐life of the immunosensor.^{[ 16 ]}

2.9. Boosting the Throughput

After confirming the capability of the chips based on Au and Ag/AgCl QREs of ensuring the duplexing of IgG^S over a single analysis of Fe^3+/2+ solution, we sought to challenge them to a set of serum samples from SARS‐CoV‐2 positive (9) and negative (9) human subjects to investigate their feasibility in boosting throughput. As aforesaid, this ability stems from the spatial isolation of the redox reactions occurring on the same WE. To provide decentralized and mobile diagnostics using our devices, we built up a prototype integrating a portable one‐channel potentiostat and switches for cell selection (Figure 5C). The switches were manually triggered to contact a single or a pair of QREs, hence devising one‐sample or duplex analyses. The potentiostat could be operated via a smartphone that can oversee acquiring, storing, processing, and transmitting the data for real‐time and remote monitoring. This sample‐to‐answer platform can afford POC assays by untrained users and telemedicine.^{[ 97} , ^{98 ]}

In practice, positive (9) and negative (9) serum samples were first interrogated with WEs versus Au and Ag/AgCl QREs, respectively (Figure 5D and Figure S22, Supporting Information). The testing capacity was duplicated (n = 5; 45 analyses) over traditional one‐sample analyses (90 analyses would be required in this case). Although the positive and negative samples could be recognized from these data (ΔI was roughly −8 and −2 µA for these sera, respectively), the accuracy of biosensors should be evaluated by comparing data obtained with the same QRE, as Au and Ag/AgCl led to distinct sensitivities. Thereby, duplex analyses were next performed on ten samples (five negative and five positive sera) to assess the screening capacity of the approach by establishing specific thresholds for each QRE. These measurements were performed two times to allow the interrogation of both QREs with the two types of sera (Figure 5E and Figure S23, Supporting Information). The samples could be discriminated based on the recorded ΔI values when using Au and Ag/AgCl QREs (Figure 5F). The threshold limits were −6.1 (Au) and −3.3 µA (Ag/AgCl QRE). Student's t‐tests yielded p‐values of 1.5 10⁻⁴ (Au) and 1 10⁻³ (Ag/AgCl QRE; 95% confidence interval), confirming a statistically meaningful difference among the signals for positive and negative sera. Finally, it is noteworthy that the negative sera tested positive for other coronavirus variants (229E, OC43, NL63, and HKU1), as recently revealed by enzyme‐linked immunosorbent assays.^{[ 99 ]} In this way, the results obtained from the serum samples reinforce not only the ability of the chips to enhance throughput but also the selectivity of the peptide‐tethered immunosensors.

3. Conclusion

Here we demonstrate vertically engineered ultra‐dense electrochemical chips and a single‐response but spatially resolved duplexing concept that synergistically provides insights for further advances in clinical diagnostic tools envisioning a rapid, accurate, and large‐scale screening of on‐demand diseases outside of laboratory settings. When it comes to the commercial adaptation of clinical methods, fabrication aspects become challenging facing their advancement through the TRLs.^{[ 6 ]} In this way, the reported prototyping has been devised as a promising option to cope with these issues by affording reproducible, scalable, and fully integrated sensors with high integration density, thus yielding a low unit cost. By leveraging the design of the MECs, blocking strategy‐based label‐free biosensors could be adapted in a one‐tag SERM assembly. Based on the readout of the faradaic currents for Fe^3+/2+ with spatially resolved Au and Ag/AgCl QREs along a single SWV scan, this approach is envisioned to hasten the deployment of multiplex label‐free biosensors. In addition to implying duplex analyses via the use of simple, cheap, handheld, but high‐performance one‐channel potentiostats, one should mention that the one‐tag SERM can be adopted in combination with serial assays on a fully‐integrated device, as rendered by the ultra‐dense chip, to boost testing capacity further. The approach also delivers advantages over the existing SERM approaches such as the absence of crosstalk between the tags (it makes more convenient the choice of such species toward well‐resolved peaks) and interferences derived from the presence of different recognition elements on the same WE.

Intensive surface characterization studies provided a comprehensive understanding of the vertical devices, whereas electrochemical experiments and application with peptide‐tethered biosensors confirmed their ability to track spatially resolved molecular binding events along a single SWV scan. As advantages, the preparation of these immunosensors involves simple drop‐casting steps and the peptide (receptor) holds merits in synthesis and stability. The reached sensitivity was commensurate with a peptide‐based EIS biosensor,^{[ 16 ]} but it is still greatly lower than the data reported by existing nano‐biosensors that entail LODs ranging from ng mL⁻¹ to pg mL⁻¹ for IgG^S.^{[ 37 ]} Powerful alternatives to improve the sensitivity of our vertical chips are the use of metal‐organic frameworks (MOFs),^{[ 100 ]} the electrodeposition of nanostructured microelectrode on recessed WE,^{[ 37} , ¹⁰¹ , ^{102 ]} reagentless molecular pendulum‐based detection,^{[ 10} , ³⁵ , ^{103 ]} the electrochemical nano‐roughening of Au film,^{[ 18 ]} the anisotropic etching to create nanoporous Au films,^{[ 15 ]} and the use of electrocatalytic palladium nanoclusters.^{[ 19 ]} Such methods are particularly promising as they can entail antifouling capacity against untreated biological fluids, which are expected to inhibit the use of the MECs by passivating the recessed WEs, as demonstrated here with FBS.

Future studies should also be focused on the design of SERM biosensors for quantitative assays and the detection of different biomarkers to boost clinical accuracy and assess coexisting diseases.^{[ 30} , ³⁵ , ³⁶ , ^{104 ]} The possibility to make numerous serial assays on a fully integrated device is an additional benefit of the ultra‐dense chips that can also be harnessed in future works to further boost throughput. Regarding the sensor shelf‐life after plasma treatment, it is a key parameter by directly impacting their commercial translation. An efficient approach to provide stability superior to 25 days, as achieved here, maybe utilizing a more appropriate vacuum sealer machine to store the chips (a simple food sealer was used in this work). Alternatively, the adoption of atmospheric pressure or, especially, oxygen plasma may be an attractive solution. From preliminary data, these treatments are also capable of yielding the descumming and SU‐8 oxygenation processes. Besides, more in‐depth studies are demanded. Indeed, tests with distinct QREs acting along the same voltammogram can raise further questions and insights, even in areas other than sensors. Importantly, in addition to delivering high sensitivity even in undiluted biological fluids,^{[ 37} , ¹⁰¹ , ^{102 ]} the use of nanostructured microelectrodes may also avoid the overlap of the current peaks versus Au and Ag/AgCl over a scan, with the resulting complete separation of the duplex responses.

In summary, by combining ultra‐dense electrochemical chips with single‐response but spatially resolved duplexing, we have developed a broadly adaptive biosensing platform. Indeed, the SERM addressed here could be extended to track other on‐demand targets/samples and to various biosensing modes, including reagentless multiplexing.^{[ 48} , ⁴⁹ , ⁵⁰ , ⁵¹ , ^{52 ]} One should also mention that the vertical sensors as reported here could be retrofitted into devices reached by low‐cost methods. For instance, while patterning WEs on a paper substrate utilizing techniques such as screen printing or laser‐assisted pyrolysis,^{[ 22 ]} QRE finger‐coated adhesives with holes delimited by a knife plotter could be fixed onto the paper, producing a vertically engineered ultra‐dense device. Thus, looking forward, we envision that the concepts addressed here will provide the field a venue with opportunities for engineering a new class of high‐throughput biomolecular sensors.

4. Experimental Section

Chemicals

All reagents used were of analytical grade without prior purification. Ultra‐high purity water (Millipore Milli‐Q system; resistivity of 18.2 MΩ cm) was used in the preparation of the solutions. Potassium ferrocyanide (K₄[Fe(CN)₆]), potassium ferricyanide (K₃[Fe(CN)₆]), FcMeOH, potassium nitrate (KNO₃), potassium chloride (KCl), dibasic sodium phosphate (Na₂HPO₄), monobasic sodium phosphate (NaH₂PO₄), phosphate buffer saline tablet (PBS, 10 mm, pH 7.4), and phosphate buffer (PB) were purchased from Sigma‐Aldrich (St Louis, MO). Conductive Ag/AgCl ink (60:40 v/v) was purchased from Gwent Group (Wales, UK).

Biological Material

As a biological biorecognition element, the synthetic peptide P44 was used.^{[ 59 ]} The lyophilized powder was resuspended and diluted in PBS. Antibodies used as positive (IgG^S) and negative (IgG⁻) controls were provided by Biodefense and Emerging Infections Research Resources Repository (BEI Resources) and diluted in PBS. Rio de Janeiro Cell Bank (BCRJ, Brazil) supplied sera from pre‐pandemic individuals and convalescent through an approved Ethics Committee of the Federal University of ABC (Ethical Review Presentation Certificate (CAAE) number: 43 139 921.2.0000.5594). Importantly, written consent was collected from all participants for these patient serum samples. Sera from SARS‐CoV‐2 positive (convalescent) and negative (uncontaminated) donor patients were analyzed by the Medtest SARS‐CoV‐2 IgG/IgM rapid test cassette (COV20030081 batch). These samples were stored in a freezer at −80 °C. Before the electrochemical tests, they were kept at room temperature until complete thawing and, then diluted 1:100 v/v in PBS.

Microfabrication

The device fabrication included photolithography and electron‐beam evaporation (EvoVac 046, Angstrom Engineering) steps. The vertical films were constructed on a glass substrate (75 mm × 35 mm) in the following order: i) patterning pads and fingers of WEs, ii) dielectric (SU‐8) deposition with sensor area definition (4 × 45 µm or 800 µm), and iii) patterning pads and fingers of QREs. WEs and QREs were composed of a 20 nm‐thick Cr film (adhesion layer) and a 100 nm‐thick Au layer. For pattern transfer through standard UV photolithography, the substrates underwent UV exposure (λ = 300 nm) by employing a photoaligner (Karl Suss, Garching) at 9.5 mW cm⁻². Before applying the photoresist or metallization, an O₂ plasma treatment (100 W, 70 mTorr; Plasma Cleaner CY‐P15L‐B300, Zhengzhou CY Scientific Instrument) was performed for 40 s. The final devices were exposed to Ar plasma (250 W, 20 Pa) for 5 min to clean the WEs via the descumming process. To provide SERM, adhesives with holes on the regions of WEs and deposited Ag/AgCl ink were fixed on some top Au thin‐film fingers. This ink acted as QRE along with uncoated top Au fingers. Ag ink was utilized to make an electrical contact between this pair of QREs. More details on the microfabrication process are available in the Supporting Information.

Surface Characterization

The effects of exposure to Ar plasma were investigated through contact angle, SEM (Thermo Fisher Scientific Quanta 650 FEG), AFM (Bruker MultiMode8), KPFM (NX10, ParkSystems), AFM‐IR (nanoIR2s‐s, Anasys) and XPS (Thermo Scientific, K‐alpha) measurements. Contact angles were measured to analyze wettability using a tensiometer (Theta Lite, Attension) with the sessile drop method and a deionized water droplet volume of 10 µL. Imaging of the sensor and analysis of physical changes induced by Ar plasma were conducted using SEM and AFM. The effectiveness of Ar plasma in cleaning the WEs was evaluated both electrically and chemically. KPFM was employed for electrical assessment, while AFM‐IR and XPS provided chemical insights. AFM‐IR measurements were performed at a resolution of around 50 nm, enabling localized chemical analyses. These techniques were also utilized to study chemical changes in the SU‐8 dielectric after Ar plasma exposure. In addition, each step of electrode modification was investigated by XPS.

Electrochemical Analyses

The electrochemical assays were conducted using benchtop (Metrohm Autolab PGSTAT 302N) or handheld potentiostats (PalmSens4 and Sensit BT). The supporting electrolyte consisted of a solution containing 1 mol L⁻¹ KNO₃ in 12 mmol L⁻¹ PB. EIS experiments were performed using an amplitude of 10 mV and a frequency range from 1 × 10⁻¹ up to 1 × 10⁴ Hz. SWV tests were conducted within a potential window ranging from −0.8 to +0.6 mV, a modulation amplitude of 35 mV, a step potential of 3.5 mV, and a frequency of 15 Hz. For the electrochemical tests on fully integrated chips, a volume of 10 µL was dropped on the cells. Exceptionally, 100 µL were added on QREs to make the open circuit chronopotentiometry assays. In this case, a commercial reference electrode of Ag/AgCl (saturated KCl, Metrohm) and platinum wire as a counter electrode were used.

The precision assessments were undertaken both within sensors on the same wafer (intra‐) and between sensors on different wafers (inter‐wafer) using CV for 1 mmol L⁻¹ Fe^3+/2+. For intra‐wafer precision, 48 sensors with 4 scans each (n = 192) were studied. For inter‐wafer precision, ten sensors were scrutinized from four different substrates (n = 352). All measurements were conducted with a υ of 50 mV s⁻¹. To investigate the shelf life of the bare sensor after Ar plasma exposure, SWV assays were made to assess current stability over 30 days. Three different sensors were tested for each day, resulting in a total of 27 sensors (n = 135). These chips were stored in closed plastic bags using a simple food sealer machine (Minivac, Cetro).

The diffusion regime was explored by CV measurements of 1 mmol L⁻¹ Fe^3+/2+ at different υ values, ranging from 5 to 1000 mV s⁻¹. Furthermore, the electrochemical reversibility of the 4 × 45 µm sensor was also investigated through CV, EIS, SECM, and SICM (Versascan scanning electrochemical system Princeton applied research, PAR). SECM measurements utilized a microelectrode (25 µm in diameter) and 1 mmol L⁻¹ FcMeOH probe in 0.1 mol L⁻¹ KCl to map the faradaic currents linked to each sensor component (WE, SU‐8, and QRE) and to get approach curves for Au (positive feedback), glass (negative feedback), and SU‐8. For SICM tests, a nanopipette (100 nm in diameter) was submerged in 10 mmol L⁻¹ KCl to assess the surface charge of the material, which was determined by the ratio between the negative and positive ionic currents (further details are available in the Supporting Information).

Statistics

The LOD values were calculated as 3.3 × standard deviation of the blank ÷ sensitivity.^{[ 105 ]} The RSD was employed as a metric parameter to evaluate the precision of the microfabrication process and bioassays. Further, the variation ranges (±) of the data presented throughout the article were assessed as a function of the confidence interval, which was calculated for α = 0.05.

FEM Simulations

Finite‐element calculations were performed using COMSOL Multiphysics. A 2D model was employed to elucidate the current density distribution along the multiplex devices considering Fe^3+/2+ probe, Au WEs, SU‐8 as dielectric material, and QREs composed of Au and Ag. The material properties used as input parameters for the AC/DC Module (In‐Plane Electric Currents) were electrical conductivity and relative permittivity. The electrode conductivities were set in agreement with the software library, whereas the electrolyte properties were set proportionally to each solution differential resistance (i.e., estimated from the experimental data). As a boundary condition for the simultaneous simulation, a surface potential was forced at the Ag QRE accounting for the experimental peak separation (viz. 0.2 V). For each simulation, 78 864 triangular‐mesh elements with 40 283 mesh points were defined (5058 boundary elements, and 53 vertexes with 0.7881 minimum element quality and a ≈0.00011 area ratio). The resulting current density heat maps were normalized by the respective minimum current density value, providing an arbitrary current density scale for semi‐quantitative analyses.

Biosensor Preparation and Bioassays

To modify the WE surfaces, 6 µL of P44 solution (40, 200, 500, or 750 ng mL⁻¹) were drop‐cast on the clean electrodes and allowed to dry. Afterward, 6 µL of glycine solution (1 or 10 µmol L⁻¹; see Figure S12, Supporting Information) was added for 10 min to promote blocking and avoid non‐specific reactions with the medium. Thus, the functionalized electrodes were placed in contact with 6 µL of the antibody sample until dried. Between all the previous steps, a washing routine was performed involving 20 µL of deionized water. Control samples of IgG^S were diluted from 1 to 40 µg mL⁻¹, whereas the serum samples from convalescent patients or blood plasma from pre‐pandemic patients were diluted 1:100 v/v in PBS. All the experiments were conducted at room temperature.

Conflict of Interest

R.S.L, J.N.Y.C., G.J.C.P., A.L.G., M.H.O.P., and F.M.S. are listed as inventors on a patent filing application describing the microfabrication presented here. The remaining authors declare no conflict of interest.

Supporting information

Supporting Information

Costa J. N. Y., Pimentel G. J. C., Poker J. A., Merces L., Paschoalino W. J., Vieira L. C. S., Castro A. C. H., Alves W. A., Ayres L. B., Kubota L. T., Santhiago M., Garcia C. D., Piazzetta M. H. O., Gobbi A. L., Shimizu F. M., Lima R. S., Single‐Response Duplexing of Electrochemical Label‐Free Biosensor from the Same Tag. Adv. Healthcare Mater. 2024, 13, 2303509. 10.1002/adhm.202303509

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.