Gold-Dotted Titanium Nanotubes Deliver Enzyme-Free Glucose Sensing With Record Low Detection Limit

A glucose sensor that swaps fragile biological enzymes for specks of gold has been tuned to a striking level of performance by researchers who learned to control exactly how many nanoparticles settle onto a bed of titanium dioxide nanotubes. The team, working across institutions in Tunisia, Spain, Morocco, Saudi Arabia and Australia, grew the nanotubes directly on titanium foil, decorated them with gold nanoparticles by electrochemical deposition, and then varied the number of deposition cycles from 10 to 60. At the highest loading, the electrode achieved a sensitivity of 7.96 microamperes per millimolar per square centimeter and a detection limit of 15.1 micromolar over a linear range of 0.5 to 4 millimolar, figures that outperform several comparable gold- and titanium-based sensors reported previously.

The device belongs to a class of technologies known as non-enzymatic glucose sensors. Conventional glucose monitors rely on enzymes such as glucose oxidase, which are exquisitely selective but inherently unstable: they degrade with temperature and pH swings, lose activity during storage, and suffer mass-transfer bottlenecks once immobilized on an electrode. A non-enzymatic electrode instead catalyzes glucose oxidation directly at an active surface, eliminating the biological degradation problem, at the cost of needing careful engineering to keep other electroactive molecules from interfering.

Titanium dioxide nanotubes were chosen as the scaffold because their vertically ordered architecture offers a very large surface area for catalyst attachment, and their walls provide a direct electron pathway to the underlying substrate. Titanium dioxide is also chemically stable across a wide pH range, biocompatible, and far cheaper than noble metal electrodes. The team made the nanotubes by anodic oxidation of commercially pure titanium foil in an ethylene glycol electrolyte containing ammonium fluoride and water. In a two-step process — a first anodization at 60 volts for 45 minutes followed by ultrasonic removal of the initial layer, then a second anodization under identical conditions for 120 minutes — they grew well-ordered nanotubes roughly 100 nanometers in diameter and up to 15 micrometers long. Annealing in air at 400 degrees Celsius for three hours crystallized the amorphous material into the anatase phase.

Pristine titanium dioxide, however, makes a poor glucose sensor on its own. Its wide bandgap of about 3.2 electronvolts renders it electrically insulating in the dark, and its surface lacks the catalytic sites needed to oxidize glucose at useful rates. Decorating the nanotubes with gold nanoparticles solves both problems at once: the gold supplies catalytic active sites while the metal–semiconductor interface forms a Schottky barrier that assists electron transfer. Electrodeposition was the method of choice because, unlike chemical reduction, photodeposition or sputtering, it allows precise, real-time control over particle loading, size and dispersion simply by adjusting the deposition potential, time and number of voltammetric cycles.

The deposition itself was carried out in a three-electrode cell, with the nanotube electrode as the working electrode, a silver/silver chloride reference, and a platinum counter electrode. Cyclic voltammetry swept from −1.25 to −0.7 volts versus the reference at 50 millivolts per second in a potassium chloride solution containing tetrachloroauric acid. Scanning and transmission electron microscopy then revealed what happens as the cycles accumulate. Nucleation begins along the inner nanotube walls, but the elongated channels impose severe diffusion constraints on the gold precursor. Ions deposit preferentially near the nanotube entrances, where the diffusion path is shortest, so particle number and size grow with each additional cycle and eventually begin to choke the pore openings.

Particle statistics tracked the cycle count closely. Mean particle sizes were 27 nanometers at 10 cycles, 33 at 20, 39 at 40 and 33 at 60, with individual particles ranging from roughly 15 to 60 nanometers. Energy-dispersive X-ray analysis showed gold content peaking at 13.38 atomic percent at 40 cycles before easing back to 9.15 percent at 60 cycles, a dip the authors attribute to saturation or changes in deposition dynamics. By 60 cycles, transmission electron microscopy showed the nanotube surface covered with well-dispersed spherical particles inside and outside the tubes, and lateral coalescence had begun to stitch neighboring particles into an interconnected network about 33 nanometers across on average.

X-ray diffraction confirmed the anatase phase in every sample and revealed a subtle but meaningful structural story. Pure nanotubes had an average crystallite size of 9.13 nanometers. Early gold incorporation disrupted the lattice: at 10 cycles the crystallite size dropped to 7.86 nanometers while microstrain rose to 1.343 × 10−3 percent. As deposition continued, the structure recovered and then improved — at 60 cycles the crystallites grew to 12.42 nanometers, the largest in the series, while microstrain fell to its minimum of 0.85 × 10−3 percent and peak widths narrowed correspondingly. Prolonged deposition, in other words, does not just pile on gold; it refines the underlying crystal structure and reduces lattice defects that would otherwise impede charge transport.

Optically, the gold decoration transformed the material in a non-monotonic way. Pure nanotubes showed a bandgap of 3.36 electronvolts, consistent with anatase. After 10 cycles the bandgap narrowed to 3.21 electronvolts, and at 20 cycles it reached its deepest redshift at 2.94 electronvolts, which the authors attribute to new electronic states introduced by the gold, strong interfacial coupling at the metal–semiconductor junction, and near-field enhancement from localized surface plasmon resonance — the collective oscillation of conduction electrons in the nanoparticles. Then the trend reversed: at 40 cycles the bandgap widened to 3.47 electronvolts, above even the pristine value, apparently because the particles had coalesced into larger aggregates that damp the discrete plasmon resonance and reduce the interfacial area. At 60 cycles it fell again to 3.01 electronvolts, signaling the emergence of a percolated conductive network with different dominant absorption behavior. Photoluminescence measurements echoed this evolution: the 10-cycle sample showed the lowest emission intensity, consistent with a Schottky junction that ferries photogenerated electrons from the semiconductor into the metal and suppresses recombination, while the 60-cycle sample showed the highest intensity, consistent with a shift toward ohmic-like contact behavior as the particles agglomerate.

The electrochemical payoff came at 60 cycles. In cyclic voltammetry, a well-defined anodic peak near 334 millivolts, tied to the formation of an incipient hydrous gold oxide layer, grew steadily with deposition cycles and reached its maximum for the 60-cycle electrode. When glucose was added, the peak shifted to about 350 millivolts and its current rose linearly with glucose concentration from 0.1 to 4 millimolar, reflecting the catalytic oxidation of glucose to gluconic acid. In alkaline media the gold nanoparticles mediate this oxidation through active gold hydroxide species, while the nanotube scaffold conducts electrons along the tube walls and prevents particle aggregation. Calibration gave a sensitivity of 7.96 microamperes per millimolar per square centimeter for the 60-cycle electrode — the best among bare and decorated samples, which ranged from as low as 0.68 for the 10-cycle electrode up to 6.96 at 20 cycles — and a detection limit of 15.1 micromolar calculated by the 3-sigma method with a blank standard deviation of 0.04 microamperes per square centimeter.

Two further results explain why the best performance arrived at the very loading where the nanotubes are most clogged. First, selectivity: against dopamine, ascorbic acid, lactose and fructose, the selectivity coefficients all stayed at or below 0.055, corresponding to less than 6 percent cross-reactivity, and the glucose signal recovered to roughly 62 percent of its original 81.2-microampere value after an interferent cocktail, indicating robust anti-fouling behavior. Second, and more intriguingly, the partial pore choking at 60 cycles turns out to be offset by a mechanistic transition. Although the roughly 95-nanometer surface clusters restrict diffusion into the deep channels, their lateral coalescence creates a continuous, highly conductive metallic network across the electrode interface, cutting charge-transfer resistance and enabling rapid electron extraction. The team describes this as a shift from porous, diffusion-limited kinetics to electronic percolation-dominated charge collection — the interconnected gold topology simply wins out over the lost pore accessibility.

The authors are careful to frame the limits of the claim. The reported electrochemical measurements were performed in concentrated alkaline media rather than under physiological conditions, and they note that further testing in such conditions would be required to confirm clinical potential. Still, the low detection limit falls within the 10-to-200-micromolar glucose range found in biofluids such as tears and sweat, and the sensor retained reproducible operation over 40 operational cycles. Beyond glucose monitoring, the same deposition-controlled tuning of bandgap, crystallite quality and metal–semiconductor interface bears directly on photocatalysis, anticorrosion coatings and photoelectrochemical energy conversion, where the gold–titanium dioxide junction is already under active study. What the work adds is a demonstration that a single fabrication dial — the number of voltammetric cycles — can steer the optical, structural and electrochemical character of the whole system, and that the sweet spot for sensing may lie precisely where discrete nanoparticles stop being isolated islands and start becoming a wire.

Subject of Research: Non-enzymatic electrochemical glucose detection using gold nanoparticle-decorated TiO2 nanotube electrodes

Article Title: Insight on gold nanoparticle-decorated TiO2 nanotube electrochemical sensor for non-enzymatic glucose detection

Article References: Benghanoum, H., Sassi, S., Zaghouani, R. B., Bouich, A., Arbaoui, A., El Hawary, M., Dhiflaoui, H., Khezami, L., Karrech, A., Soucase, B. M., & Hajjaji, A. (2026). Insight on gold nanoparticle-decorated TiO2 nanotube electrochemical sensor for non-enzymatic glucose detection. Frontiers in Materials, 13, Article 1938438. https://doi.org/10.3389/fmats.2026.1938438

Image Credits: AI Generated

DOI: 10.3389/fmats.2026.1938438

Keywords: gold nanoparticles, TiO2 nanotubes, non-enzymatic glucose sensor, electrodeposition, electrochemical biosensor, localized surface plasmon resonance, Schottky junction, anodization, detection limit, sensitivity, selectivity, bandgap tuning