Nonenzymatic Glucose Biosensor Based on Selective Reduction of Silver Ions

Abstract

A nonenzymatic conductivity-based glucose biosensor, reported here, designed for potential noninvasive measurement in accessible biofluids such as saliva, sweat, and tears. The sensor exploits a classic Tollens’ silvering process that deposits metallic silver on a sensitized surface via a quantitative, selective reduction of silver ions by glucose. Of the two surface sensitization methods used, the self-assembled bifunctional thiol/silane surfactant yields a more stable and sensitive sensor than tin­(II) chloride. The conductometric sensor consists of two spiral interdigitated silver electrodes fabricated by lithography and wet etching. The sensor impedance changes after selective metallic silver deposition between the electrodes. The sensor exhibits a sigmoidal response in a wide range of glucose concentrations from 200 mg/dL to less than 0.2 mg/dL and an ultralow limit of detection (ULOD) of 0.3 × 10^–5 mg/dL. The ULOD is attributed to the percolation network morphology of the deposited silver, connecting two electrodes (fractal dimension D = 1.5) that function like parallel nanoresistors. The log–log plot of conductivity vs glucose concentration yields a conductivity exponent that increases from 1 to 2, predicted for a 2D to 3D percolation transition in going from ultralow to high glucose concentration.

1. Highlight

A nonenzymatic glucose sensor to selectively detect glucose at ultralow concentration.

Interdigitated, spiral-patterned silver electrodes facilitate charge transfer.

The percolation network formed by silver nanoparticles enables the detection of ultralow glucose concentrations.

2. Introduction

Diabetes is a chronic disease in which uncontrolled high blood glucose levels lead to blindness, kidney failure, and cardiovascular disease. In 2022, the estimated cost of diagnosed diabetes in the U.S. was 412.9 billion dollars. Currently, 529 million people are living with diabetes worldwide. Therefore, the high occurrence and morbidity rates caused by diabetes have attracted attention to the development of glucose biosensors. User-friendly, cost-effective, and accurate detection of blood glucose concentrations is desirable for disease management. The earliest biosensor for quantifying glucose, by Clark and Lyons, relied on glucose oxidase’s redox ability and electron transfer in solution. Recent examples of enzymatic glucose sensors include the FreeStyle Libre by Abbott and Dexcom G7 by Dexcom. Challenges with enzyme-based sensors include high costs, complex fabrication, short shelf life, and limited reproducibility. These challenges have encouraged researchers to explore enzyme-free glucose sensors, such as the ZnO/Co₃O₄/rGO nanocomposite and the AuNi@AC nanoparticle electrode sensor, which directly oxidize glucose at electrodes. A more common challenge is patient compliance with monitoring regular blood glucose concentrations due to the painful blood extraction process. Therefore, other accessible biofluids, such as saliva or sweat where the glucose concentration is proportional to blood glucose, provide better targets; however, the low glucose concentration in these fluids has limited the development of appropriate sensors, a challenge that this work addresses.

In fabricating nonenzymatic biosensors, electrodes modified with nanomaterials such as copper, nickel, platinum, gold, and silver have significant potential as scaffolds for immobilization and rapid electron transfer. Examples of ultrahigh sensitivity metal, semiconductor nanoparticles-based gas sensors are well-known. , Among these materials, silver nanoparticles are more stable, nontoxic, biocompatible, antibacterial, and cost-effective, demonstrating high conductivity and excellent electrocatalytic activity. Recent studies on nonenzymatic silver-based sensors rely on the implantation of various materials and complex detection methods. Optical and electrochemical detection methods are examples of these advancements on a laboratory scale. Unlike optical detection, electrochemical devices can measure glucose in cloudy samples and are more cost-effective. Still, the kinetics of glucose oxidation are too slow to generate a faradaic current. Moreover, the electrode surface is susceptible to poisoning by chloride ions and adsorptive species, to the extent that amperometric signals may disappear during continuous operation. To provide a user-friendly glucose detection method, this paper presents a new type of glucose sensor that utilizes the impedance of silver nanostructures formed by the reduction of Ag ions by glucose, which can even be measured with an inexpensive multimeter.

Here, the sensor fabrication begins with the deposition of a silver by glucose reduction (via an aldehyde group; Tollen’s reaction) of silver ions on the modified glass surface, sensitized through self-assembled thiol surfactants and tin­(II) chloride. Optical lithography and KI/I2 wet etching enable the patterning of interdigitated spiral electrodes from the silver film. Silver deposition from samples with known and unknown glucose concentrations creates short-circuiting pathways between electrodes, acting like parallel resistors. The number of structures connecting the patterned electrodes depends on the concentration of reduced silver ions, which is proportional to glucose concentration. These connections, or the number of clusters connecting electrodes, affect the sensor’s electrical DC conductivity, which can be measured with an inexpensive multimeter. A remarkable morphological change occurs in the interelectrode deposits at low glucose concentrations. Homogeneous thin films of silver break into fractal/diffusion-limited aggregates (DLA) (still able to conduct) at ultralow glucose concentrations, which allows a vast improvement in the limit of detection (LOD) over the current generation of glucose sensors. A recent study reported an LOD as low as 2 nM for nonenzymatic, noninvasive glucose sensors. Our sensor showed an LOD of 0.3 × 10^–5 mg/dL (0.17 nM), about 10 times higher, yet still exhibits an excellent LOD.

3. Experimental Section

3.1. Material and Apparatus

These studies utilized a commercially available silvering kit (HE-300, Peacock Laboratories, Inc., Philadelphia, PA, USA) for the electroless deposition of silver thin films (i.e., mirroring). The stock HE-300 reagents consisted of (A) silver nitrate solution (0.034 M), (B) activator solution (ammonium hydroxide + NaOH), (C) reducing agent (glucose, 800 mg/dL), and (D) sensitizer solution (tin­(II) chloride, 0.76 M in water). Other reagents, 3-mercaptopropyl trimethoxysilane (HS (CH₂)₂Si (OCH₃)₃) (MPS) in anhydrous toluene (99.8%), Glucose, lactose, ascorbic acid, NaCl, KCl, MgCl₂ and KI/I₂ were purchased from Sigma-Aldrich. Photolithographic patterning utilized I-line (365 nm) positive photoresist SPR220 (3.0, Shipley Company, LLC. SPR 220 Positive Photoresist Process Guidelines. Marlborough, MA, USA). The Tamarack 160 photoprinter provided an exposure dose of 200 mJ/cm² from a high-pressure Hg lamp (1 mW/cm² power) to pattern an interdigitated electrode. An ultrasonicator (Chicago Electric Power Tools, 2.5 L capacity) was used to clean the glass substrates. A PC-interfaced Keithley 617 multimeter was used for measuring two-wire electrical resistance. The morphology of electroless-deposited silver structures was investigated using an optical microscope (Carl Zeiss Microscopy GmbH, Germany) and a Zeiss Sigma VP FEG scanning electron microscope (SEM) equipped with an EBSD detector at a beam voltage of 10 kV. The Filmetrics F20 spectrometer, operating in reflectance mode (400–1000 nm), enabled measurements of thin-film thickness. Contact angle measurements were performed using a CCD camera (Kodak 100) connected to a computer, which captured droplet images on the modified surfaces. Contact angles were measured from analysis of images of water droplets placed on sensitized glass using ImageJ software.

3.2. Surface Sensitization/Modification

Surface sensitization is necessary for the homogeneous deposition of silver on a macroscopic scale and is a common practice in mirror fabrication, where a thin, uniform silver film is required.

3.2.1. Sensitization by Hydrated Tin­(II) Oxide

A previous study conducted in our group suggested that the sensitizing step plays a critical role in determining the structure and surface roughness of the deposited silver during the silver mirroring process. The morphological features of SnO _x nanoparticles influenced the controlled deposition of silver on glass [12]. In this study, glass slides were cleaned using the RCA-1 method. Self-assembled monolayers of hydrated SnO _x nanoparticles were deposited from SnCl₂. xHCl solution onto microscopy-grade cover glass. Surface sensitization is due to Sn²⁺ species, which can reduce Ag⁺ ions to Ag⁰ to initiate the heterogeneous nucleation of silver. However, it should be noted that immediately upon sensitization, reduction of Ag⁺ should be carried out to prevent desensitization of Sn²⁺ to Sn⁴⁺ state due to oxidation from atmospheric O₂.

3.2.2. Surface Derivatization Based on Self-Assembled Thiol Surfactant

We used 3-mercaptopropyl trimethoxysilane (MPS) as a sensitizer/binder for Ag on the glass surface. Thiol groups are well-known binders to noble metals through chemisorption and silane groups react with a hydroxylated glass surface to anchor the MPS molecule via Si–O–Si bonds. Cleaned glass slides were immersed in a 50 mM MPS solution prepared in anhydrous toluene (99.99%) and incubated for 24 h inside a nitrogen-purged glovebox. After the reaction, silane-functionalized glasses were rinsed with anhydrous toluene.

3.3. Surface Characterization after Sensitization

Surface characterization involved water contact angle measurement to assess the surface sensitization. The observed water contact angles after silane derivatization and tin­(II) chloride sensitization were 65 ± 3° and 30 ± 2°, respectively, which agreed with previous studies [15] and indicated successful derivatization of these surfaces (S4).

3.4. Uniform Electroless Deposition of Silver on the Sensitized Glass Surface

Metallic silver was deposited on two different sensitized glass surfaces using the Tollens’ process from a commercially available kit (HE-300, Peacock Laboratory) for silver mirroring. A shiny silver film (thickness of 50 nm) resulted from the reduction of silver ions (silvering solution A, 0.0034 M AgNO₃) by (800 mg/dL glucose (aldehyde) solution C) in the presence of a base, NH₄OH (Activator solution B).

3.5. Characterization of Deposited Silver Films

SEM characterization showed variations in the thickness of silver films on two surface-modified samples. The thickness measurements obtained using Filmetrics were consistent with SEM images (Figure A,B). The thickness of the silver layer on the MPS silanized surface was 50 ± 3 nm, whereas on the tin­(II) chloride-sensitized surface, it was 15 ± 2 nm. Additionally, sheet resistance measurements showed lower resistance of the silver layer on the silanized surface compared to the tin­(II) chloride-sensitized surface, at 7 ± 2 Ω and 30 ± 4 Ω, respectively, supporting the presence of a thicker silver nanolayer on the silanized surface. This result suggested that the silver nanolayer on the thiol-silanized surface had a stronger binding affinity for silver than did tin­(II) chloride sensitization.

1.

SEM images (cross section and top–down views) of deposited silver on sensitized glass: (A,B: from blanket silver deposited before lithography), (C,D: on a patterned sensor after the glucose detection). (A) Tin­(II) chloride, (B) MPS sensitized surfaces, (C) on top of the silver electrode (C1: pattern line). D from an Interelectrode gap region where silver is deposited due to Ag⁺ ion reduction by glucose during sensing. EDX data for C and D images appears in Supporting Information Section S6.

3.6. Mask for an Interdigitated Electrode/Sensor

A prior study showed that spiral patterns provide more efficient charge collection than other shapes, such as interdigitated combs or simple parallel lines. Also, a circular footprint is advantageous for receiving hemispherical biofluid droplets. Spiral electrode patterns were CAD-designed and printed on a transparent sheet to create lithography masks (Figure ). Each spiral electrode had a length of 15 mm and a line width of 0.5 mm. The interelectrode spacing was 0.5 mm. These geometrical parameters could be further optimized to reduce the electrode resistance $\mathrm{R}=\frac{\rho \mathrm{L}}{\mathrm{A}}$ (ρresistivity, L: electrode length, and A: electrode cross-sectional area = electrode thickness × line width). The spiral/circular shape provided a longer path within a compact area, optimal for detecting glucose from a small volume of biofluid compared to straight lines or comb-like structures. As discussed later, interelectrode separation was also a crucial parameter for achieving high sensitivity, which depended on the number of interconnecting bridges between the electrodes. The electrode separation should be comparable to the length of the average silver aggregate structure deposited during glucose detection through silver deposition, allowing the electrode lines to be connected. If the gap is too wide, fewer clusters connect the electrodes, resulting in higher resistance and, consequently, higher impedance, which can be challenging to measure accurately using an inexpensive multimeter that typically measures up to 20 MΩ. If the gap is too narrow, the resistance will be too low, potentially below the tip contact resistance (<1 Ω), reducing the sensor’s dynamic range.

2.

Electrode mask prepared on transparency surface.

3.7. The Photolithographic Fabrication of the Conductometric Sensor

Briefly, the photolithography process involved (Figure ):

• (1)Coating the photoresist (SPR220) on the glass surface by spinning at 3000 rpm to spread out the resist,
• (2)Prebaking (95 °C for 60 s) to remove any remaining solvent in the resist film,
• (3)365 nm UV light exposure, 200 s at 1mw/sec source,
• (4)Postexposure heat treatment at 100 °C for 95 s,
• (5)Development in 0.1 M KOH 30 s and
• (6)Wet etching, and
• (7)Finally, an acetone wash to remove the unexposed resist.

3.

Schematic of the photolithography procedure. (A) Photoresist coating on the surface, (B) light exposure, (C) development, and (D) wet chemical etching by KI/I₂ mixture. The sensitized layer is deactivated for the SnCl₂-based system during photolithography due to oxidation of Sn²⁺ to Sn⁴⁺. Therefore, a second sensitization is needed before glucose detection. However, the thiol layer remains active for silane (MPS)-based sensitization even after photolithographic steps.

Wet-chemical etching with potassium iodide/iodine (KI/I₂) solution effectively dissolved the exposed silver, forming soluble silver–iodide complexes while preserving the underlying silver in the unexposed resist. The subsequent dissolution of the remaining resist by acetone completed the mask pattern transfer, thereby completing sensor fabrication (S1, S2).

3.8. Silver Deposition at Different Glucose Concentrations

Measurement of the sensor response curve employed 27 glucose concentrations varying from 0.3 × 10^–5 to 200 mg/dL. The silver deposition for glucose detection used A, B and C solutions from the Peacock laboratory. However, the concentration of glucose in solution C was varied as needed for calibration or detection of an unknown glucose concentration. Solutions A, B, and C were mixed in a microtube, and a 20 μL aliquot was applied to the electrode surface, allowing it to dry completely. Optical images of silver deposition at high and low glucose concentrations revealed distinct aggregate structures (Figure ).

4.

Optical images of silver deposition on the spiral electrode patterns, where the dark regions correspond to the thick silver film areas and the bright regions correspond to the semitransparent interelectrode region. Note at different densities of aggregates connecting electrodes as glucose concentration varies from A: 0­(bare electrode), B: 0.003, C: 2.6, D: 20, E: 100, F: 200 mg/dL.

3.9. Fractal Dimensionality of Ag Clusters at Low Glucose Concentrations

A standard method for determining the fractal dimension of aggregates is image analysis using the box-counting method. The number of particles in aggregate, N­(ε), in a square box of size ε is measured as a function of box size. Then the Fractal dimension D is evaluated. as

$$D=\frac{d(\log \mathrm{N}(\epsilon ))}{d(\log (1/\epsilon ))}$$ 1

We performed box-counting on optical images using three scale-down factors, which define how much the image is subdivided at each step (i.e., the ratio of the original image size to the box size). Specifically, the chosen scale-down factors were 1, 2, and 4. At each scale, we counted the number of boxes covering the structure: 4, 11, and 29 (Figure ).

5.

Box-counting analysis at different scale-down factors and Plot of Log N versus Log ε. Scale-down factors: (A) 1, (B) 2, (C) 4, (D) optical images of silver deposition at a glucose concentration of 0.00075 mg/dL, (E) the slope of line is the fractal dimension on tin chloride sensitize surface­(red), silanized surface­(black).

The logarithms of these counts (0.602, 1.041, 1.462) were plotted against the logarithm of the inverse scale, and the fractal dimension D was calculated from the slope of this log–log plot (Figure E) yielding D = 1.5 as shown in Table .

1. Fractal Dimensions of Silver Clusters on the tin­(II) Chloride and MPS Sensitized Surfaces.

modified surfaces	D
tin(II) chloride	1.51 ± 0.01
silanized	1.49 ± 0.03

The morphology of aggregation is directly linked to the ultralow detection limit by enabling efficient conductive pathways through the aggregated fractal structure.

4. Results and Discussion

4.1. Calibration

We designed experiments on two different sensitized surfaces across a wide range of glucose concentration from very low (0.3 × 10^–5 mg/dL) to high (200 mg/dl), covering the full range of glucose concentration in body fluids. To estimate errors, conductance measurements utilized three identically prepared sensors for each glucose concentration. Additionally, on a tin­(II) chloride-sensitized surface, a second sensitization step before silver deposition, resulting from glucose reduction, was necessary, as Sn²⁺ on the surface was prone to atmospheric O₂ oxidation, vide supra.

The conductometric sensor was activated by silver deposition between the electrodes using silver mirroring chemistry (Tollen’s reaction). The typical time required for solution evaporation time on the electrode is about 15 min with a 177 mm² active area and the current sample volume of 20 μL. Detection time depends on both electrode area and applied volume. For a fixed volume, a larger area shortens drying time, while reducing the area without reducing volume increases it. To decrease detection time, both electrode geometry and sample volume must be optimized. Reducing volume and adjusting the area-to-line-spacing ratio created a thinner film, increasing the effective surface-to-volume ratio. This approach would improve detection speed and better suit it for point-of-care applications.

The reproducibility and stability of the sensor were characterized by measuring the conductance response of five separately fabricated sensors to a 20 mg/dL glucose solution. The result indicated a low relative standard deviation (RSD) of 5.5%, confirming the sensor’s reproducibility (S9). For stability studies, conductance measurements were performed with 12.8 mg/dL glucose on sensors stored in air at room temperature every 5 days over a month. Results showed 88.2% conductance retention over a month, suggesting adequate stability of the silver interdigitated electrode (S8).

The amount of silver deposited was a function of glucose concentration since a single glucose molecule can reduce two silver ions (Ag⁺) to metallic silver (Ag⁰). The deposited metallic silver formed an electrically conductive path between the two electrodes, reducing the near-infinite DC resistance to a finite value that decreased with glucose concentration. This irreversible reduction reaction limited the use of this type of sensor to a single use. Simple multimeter two-wire resistance measurements were transformed into conductance vs Glucose concentration plots, as shown in Figure A,B, corresponding to MPS- or SnO _x -sensitized glass surfaces, respectively. A novel feature of these measurements became apparent on a log–log plot (Figure C), where the ultralow limit of detection and the high sensitivity of the sensor became evident (vide infra). For reference, glucose concentrations (1–25 mg/dL) in tears, sweat, and saliva are well within the sensor’s detectable dynamic range.

6.

Conductance at different glucose concentrations on: (A) silanized surface, (B) tin­(II)­chloride sensitized surface. Solid lines represent the best fit to the eq (C) logarithmic scale of conductance on: C1: silanized, C2: tin­(II)­chloride-twice, C3: tin­(II)­chloride. Red and green straight lines schematically illustrate a 2D-3D transition for the conductivity percolation model proposed.

To model the sensor response curve, the following empirical sigmoidal functional form was used to fit data in Figure A,B (eq )­

$$\sigma (C)=\frac{a}{{(1+{\exp }^{-bC})}^c}$$ 2

In eq , C represents the glucose concentration, σ­(C) is the conductance, and the coefficients a, b, and c are empirical fitting parameters. The best-fit values of the parameters obtained by least-squares fitting for the data shown in Figure A,B appear in Table .

2. Fitted Coefficients (a,b,c), Correlation Coefficients (R ²) and Limit of Detection (LOD) for Two Samples A: MPS Silanized Surface, B: Sensitized with Tin­(II) Chloride.

sensitization	a	b	c	R 2	LOD (mg/dL)
A: MPS	0.20 ± 0.06	0.013 ± 0.003	7.4 ± 0.8	0.96	0.3 × 10–5
B: SnO x	0.006 ± 0.001	0.016 ± 0.005	7.0 ± 2	0.92	0.2

In this model, coefficient a is related to the overall quality of the silver film (thickness/packing, etc.) deposited, which is higher for MPS vide supra than for SnO _x -sensitized glass. Interestingly, the b and c coefficient values for both surfaces are comparable within the fitting uncertainty, suggesting some general morphological or material features.

4.2. Morphology of Silver Clusters

Silver morphology changes from a fractal/dendritic 2D structure made of silver nanoparticles in x–y plane at very low glucose concentrations to an undulating thick film at high concentrations. At low concentrations, the aggregate patterns resembled the pattern of cluster–cluster aggregation (Figure b,c), transitioning to diffusion-limited cluster aggregation at higher silver concentrations (Figure d–f). The fractal dimensionality as determined from image C is consistent with a fractal exponent of 1.5 (Figure and Table ). It is this cluster–cluster aggregation observed at low silver concentrations that provides the conductive channel, hence the observed ultralow glucose concentration detection. These features are also apparent in Figure c.

The z-directional topology of the deposits resembles a shallow mountain/droplet shape. Such a formation of silver islands by nucleation and surface growth mechanisms is a characteristic of the Volmer–Weber model, which results from the higher surface energy of heterogeneously nucleating silver (>1000 mN/m) than the low surface energies of SnOx and MPS (<150 mN/m).

4.3. Potential Mechanism of Ultralow LOD of the Conductivity Sensor

Conductance of the sensor is related to the conductivity and morphology of deposited silver. At high glucose concentrations, a uniform thin film of silver is deposited, as shown in Figure A,B. In this case, conductance is related to conductivity and geometrical dimensions of the interelectrode deposited film $\sigma ={\sigma }^{\prime }.\frac{A}{l}$ where σ′ is conductivity, l is the spacing between electrodes, and A is the cross-sectional area through which current flows between the electrodes (=T*L, where L is the length of the electrode and T is the thickness of the deposited film). Since L and l are fixed, while film thickness depends on the amount of silver deposited, which is expected to vary linearly with glucose concentration, the observed conductance data in Figure are inconsistent with this hypothesis. Note, however, that the thin-film model grossly simplifies the deposited film’s structure to a uniform slab, which is not observed at low glucose concentrations (Figure ).

To understand the very low LOD and high sensitivity of the sensors, a percolation model may be invoked to correlate the electric conductivity of these connected dendritic/fractal structures of silver (Figure ) deposited on the sensor surface (S5). At ultralow glucose concentrations, deposited silver forms nanoparticles that self-assemble into 2D clusters. If the mean cluster size exceeds the electrode separation, a pathway for electron flow (current) is formed. As glucose concentration increases, the number of nanoparticles and nanoparticle clusters that bridge the two electrodes increases, thereby increasing conductivity. It is well-known that the conductivity above the percolation threshold varies as

$${\sigma }^{\prime }(C)\propto {(C-C_{\mathrm{P}})}^t$$ 3

In eq , Cp is the percolation threshold concentration of silver particles, and t is the conductivity exponent, whose value is approximately 1–1.4 in two dimensions­(2D). At higher glucose concentrations, the deposited structures become three-dimensional (3D) objects, and the same equation can still describe them, but with a higher predicted conductivity exponent, t ≈ 2. Experimentally, the slopes extracted from the log–log conductance plot (Figure C) were t _2D ≈ 0.6and t _3D ≈ 1.5, showing the expected increase in t with dimensional transition. A significant increase in the slope of the curve occurs at higher glucose concentrations, where the transition from a 2D (red line) to a 3D (green line) structure is expected. Plausible approach is found using Fuchs-Sondheimer model wherein the conductivity dependence on the film thickness is predicted assuming the electron scattering length to be greater than the cluster thickness. The model predicts

$$\frac{\sigma 0^{\prime }}{{\sigma }^{\prime }}=\frac{4}{3}+\frac{\lambda }{T\log \left(\frac{\lambda }{T}\right)}$$ 4

where T, λ = 52 nm, σ₀ ^′ = 6.25 × 10⁷ S/m) are the film thickness, the electron scattering length (dominated by phonon scattering for silver), and the bulk conductivity of silver. Calculation of T requires an ensemble average over electrode-connecting clusters, which is expected to depend on cluster size. Therefore, the exponent t in eq is expected to be different than the theoretically predicted exponents for 2D and 3D percolation model. Further discussion is beyond the scope of this paper.

4.4. Sensor Sensitivity

The observed nonlinear response of the sensor, as shown in Figure , implies that the sensitivity depends on the glucose concentration, as sensitivity is defined as the first derivative of the response curve (eq )­

$$\mathrm{s}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}=\frac{\partial \sigma }{\partial C}$$ 5

From the empirical eq , sensitivity can be calculated empirically and analytically. The results show that the sensor’s sensitivity on a silanized surface is higher than that of a tin­(II) chloride-sensitized surface over a wide range of glucose concentrations (S3). Note that in the ultralow concentration region, the sensitivity is lower, and at the 2D-3D transition in the deposited layer, the sensor sensitivity increases. The crossover point at 1 mg/dL is a typical lower limit of glucose concentration in saliva in nondiabetic patients.

4.5. Dynamic Range

The conductivity of trapped water limits the sensor’s dynamic range in the low-glucose concentration region in the evaporated film. In contrast, the upper detection limit is controlled by the electrical contact resistance. The lower limit can be extended by reducing the water content of the deposited silver film (1), waiting longer for evaporation (2), or evaporating at modestly elevated temperature; all of which increase the detection time. The primary reason for choosing silver as the electrode material for the sensor was to minimize contact resistance. The current dynamic range of the sensor is 0.3 × 10^–5 mg/dL–200 mg/dL.

4.6. Effect of Inferring Species

Body fluids contain inorganic species such as Na⁺, K⁺, Ca⁺, Cl^–, Mg⁺ and phosphate, as well as organic compounds like sugars (glucose, fructose, sucrose) cholesterol, creatinine, and proteins. Among these species, glucose concentration is highest among aldehyde-functional group-bearing molecules, which can reduce Ag ions. Preliminary results showed that KCl, MgCl₂, NaCl, and sucrose did not react with Tollen’s reagents, resulting in high resistance on the electrode surface once the deposited solution dries. However, high protein concentrations in body fluids, including mucins, enzymes, and food debris, may contain aldehyde groups that artificially increase the reported glucose concentration. Specific calibrations using typical biofluids are therefore necessary. To some extent, this interference can be addressed by developing permselective protective sensor coatings or filtration that exclude macromolecules from the surface and will be addressed in future work.

5. Conclusion

This study introduces an innovative, noninvasive, enzyme-free biosensor based on conductive and cost-effective materials for measuring low glucose concentrations. A novel spiral interdigitated electrode design achieves high sensitivity, making it suitable for easily accessible biofluids such as saliva, sweat, and tears, and offering a potential replacement for blood-based methods. The sensor exploits the controlled modulation of interelectrode resistance by connecting electrodes within situ-synthesized silver nanoparticles, whose concentration is a function of glucose concentration. Current limitations are a single-use application and a bit slower detection time, both of which ongoing studies address.

Glossary

Abbreviations

ZnO/Co₃O₄/rGO

zinc oxide–cobalt­(III)­oxide–reduced graphene oxide

AuNi@AC

activated carbon supported gold–nickel

MPS

3-mercaptopropyl trimethoxysilane (HS (C ${{\mathrm{H}}_2)}_2$ Si ( ${\mathrm{O}\mathrm{C}{\mathrm{H}}_3)}_3$ )

DI

deionized water

SnOx

tin­(II) oxide

RCA-1

radio corporation of America

SnCl₂.xHCl

tin­(II) chloride solution stabilized with hydrochloric acid

KI/I₂

potassium iodide/iodine

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c08894.

• Overview of the sensor fabrication and glucose detection (S1); lithographic fabrication process of interdigitated spiral electrodes (S2); sensitivity performance on SnOx and MPS-modified surfaces (S3); water contact angle measurements (S4); optical images of silver deposition on spiral electrodes (S5); EDX analysis of silver deposited on modified surfaces (S6); FTIR spectra of sensitized surfaces (S7); sensor stability under ambient storage conditions (S8); Sensor Reproducibility­(S9); UV–visible spectra of silver films as a function of glucose concentration (S10); absorbance behavior in the Beer–Lambert concentration range (S11); fractal dimension analysis (S12); and tabulated fractal dimension values of silver clusters on modified surfaces (Table 1) (PDF)

The authors declare no competing financial interest.