Molybdenum–titanium nanopillar sensor detects ammonia at 2 ppm

A heterostructure comprising a molybdenum trioxide film supported by vertically aligned titanium dioxide nanopillars produced a strong and selective electrical response to ammonia in laboratory gas-sensing tests. At 300 °C, the sensor changed resistance by approximately 265% when exposed to 10 parts per million (ppm) of ammonia, compared with responses of about 40% to carbon monoxide, 20% to nitric oxide and less than 10% to sulfur dioxide at the same concentration. The device also produced a measurable response at 2 ppm ammonia, the lowest concentration tested, although its response and recovery times at that level were longer than the response time measured at some higher concentrations.

The sensing material was fabricated using glancing angle deposition (GLAD) combined with electron-beam evaporation. GLAD created a forest of upright TiO2 nanopillars on a silicon substrate before a continuous MoO3 layer was deposited over the pillars. The nanopillars had a nominal thickness of 250 nm, while the MoO3 film was deposited to a nominal thickness of 200 nm. For the working device, both materials were deposited only over the exposed sensing area of a silicon dioxide/silicon substrate carrying titanium/platinum interdigitated electrodes. The electrodes contained 100 fingers, each 5 µm wide with 5 µm spacing, across an active area of approximately 1.1 × 1.0 mm. The researchers stated that this GLAD-fabricated MoO3 film/TiO2 nanopillar structure had not previously been reported for chemiresistive ammonia sensing.

Structural measurements confirmed the presence of both intended oxide phases. X-ray diffraction showed peaks assigned to anatase TiO2 at 2θ values of 47.81° and 51.90°, together with peaks from orthorhombic α-MoO3 at 27.57° and 56.15°. Slight broadening of the MoO3 diffraction peaks suggested the presence of nanocrystalline domains, while characteristic peaks associated with metastable β-MoO3 or h-MoO3 phases were absent. Raman spectroscopy independently identified anatase TiO2 modes and several orthorhombic MoO3 vibrations. The MoO3 signals were more prominent in the Raman spectrum because the MoO3 film lay above the TiO2 nanopillars. Together, the diffraction and spectroscopic results indicated that the deposited material contained polycrystalline anatase TiO2 and orthorhombic α-MoO3 without detectable secondary phases.

Electron microscopy showed dense, vertically aligned TiO2 columns that retained their form after the MoO3 coating was added. Cross-sectional high-angle annular dark-field scanning transmission electron microscopy placed the deposited MoO3 layer at about 170.4 nm and the TiO2 nanopillar layer at about 207.3 nm. Elemental mapping placed molybdenum mainly in the upper film, titanium in the nanopillar region and silicon in the substrate, with oxygen distributed through both oxide layers. No visible cracks or delamination appeared in the cross-sectional images, and the interface showed little apparent interdiffusion. Atomic force microscopy found a continuous granular surface without visible pinholes, cracks or large agglomerates over a 10 × 10 µm² scan area. The measured average surface roughness was 29.5 nm and the root-mean-square roughness was 36.5 nm.

The architecture was intended to expose more sensing material while retaining a continuous route for charge transport. Gaps between neighboring TiO2 nanopillars create open pathways through which ammonia molecules can move beyond the upper surface and reach deeper parts of the sensing layer. The close MoO3/TiO2 interface provides a pathway for charge transfer, while the MoO3 surface supplies sites for oxygen adsorption and reactions with ammonia. Electrical measurements showed nearly linear and symmetric current-voltage behavior at 300 °C, with current ranging from approximately −13 mA to +13 mA over a −10 to +10 V sweep. At 27 °C, the current was instead in the microampere range. The researchers associated the higher current at 300 °C with thermally activated charge transport and selected that temperature for the ammonia measurements because molecular adsorption, desorption and surface reactions depend on available thermal energy.

Gas tests were conducted in a custom stainless-steel chamber using dry synthetic air as the carrier and reference atmosphere. A mass-flow controller supplied air at 40 mL min−1 until the sensor reached a stable baseline resistance. A known concentration of the test gas was then introduced, and synthetic air was restored after the signal stabilized. Response was defined as the absolute percentage difference between resistance in synthetic air, designated Rsa, and resistance after gas exposure, designated Rg, using |((Rsa/Rg) − 1) × 100|. This convention produced positive values because the device’s resistance decreased during ammonia exposure. The sensor was tested with ammonia, carbon monoxide, nitric oxide and sulfur dioxide, each at 10 ppm, to compare its response to a target gas with its responses to the other gases under the same operating conditions.

Ammonia response increased systematically as concentration rose from 2 to 40 ppm. The reported response was about 140% at 2 ppm, 225% at 5 ppm, 265% at 10 ppm, 320% at 20 ppm and 375% at 40 ppm. The sensor returned to its baseline after each exposure, and repeated response-recovery cycles remained consistent at every tested concentration. A power-law fit gave the relationship R = 132.92C^0.288, where R is response and C is ammonia concentration in ppm. The sub-linear exponent was interpreted primarily as evidence that surface adsorption influences the measured response, although the researchers noted that diffusion-limited transport, heterogeneous adsorption sites and charge transport through the heterostructure could also contribute. The experiments therefore demonstrated detection down to 2 ppm under the reported laboratory conditions, rather than establishing that the same threshold would hold in outdoor air or an industrial environment.

Response and recovery timing varied with ammonia concentration. Defined as the time required to reach 90% of the maximum response, response time fell from approximately 500 seconds at 2 ppm to around 200 seconds at 20 ppm, before increasing slightly to approximately 260 seconds at 40 ppm. The authors associated this non-monotonic pattern with faster surface reaction kinetics at higher concentrations, followed by a possible effect of active-site saturation at the highest concentration. Recovery time, defined as the interval needed for resistance to return to within 10% of its initial baseline after ammonia flow stopped, was approximately 700–750 seconds at 2–5 ppm, increased to about 800 seconds at 10–20 ppm and reached nearly 960 seconds at 40 ppm. Recovery therefore became longer as ammonia concentration increased, which the researchers related to more intense adsorption and slower desorption. The minutes-long recovery periods show that the 300 °C device offered a clear concentration-dependent signal but was not characterized by rapid reset times.

The proposed mechanism combines surface chemistry with heterointerface-controlled charge transport. In synthetic air, adsorbed oxygen extracts electrons from the TiO2 nanopillars, producing ionized oxygen species, an electron-depletion region and a relatively high-resistance state. When ammonia is introduced, the researchers proposed that it reacts with adsorbed oxygen, producing nitrogen and water while releasing electrons. Those electrons narrow the depletion region and lower the potential barrier at the MoO3/TiO2 interface, making electron transport through the TiO2 nanopillars easier and reducing the measured resistance. When ammonia flow stops, oxygen is re-adsorbed and the resistance returns toward its original value. The authors did not directly investigate defect states in the new device. As a result, the roles of vacancies and other microscopic adsorption sites in the observed behavior were inferred from the wider literature on semiconductor heterostructures rather than established by direct measurements in this experiment.

Overall, the study presents a GLAD-fabricated one-dimensional/two-dimensional MoO3 film/TiO2 nanopillar architecture as a promising laboratory platform for ammonia detection. Its principal demonstrated features include preferential ammonia sensing, a measurable response at low ppm concentrations in a controlled chamber, repeatable reversible cycles and consistent operation across 2–40 ppm. The authors attributed the performance to the combined effects of MoO3 surface reactions, the TiO2 conduction pathway, the large exposed area of the nanopillar forest, open diffusion paths and the heterointerface. The results remain laboratory findings rather than proof of a deployable environmental-monitoring instrument: the sensor required a 300 °C operating temperature, exhibited comparatively long recovery times, and was evaluated against a limited set of interfering gases in one dry-air test environment. Performance in humid air, complex gas mixtures and field conditions would require further validation.

Subject of Research: Chemiresistive ammonia sensing with a MoO3 thin-film and TiO2 nanopillar heterostructure

Article Title: Enhanced NH3 sensing using a MoO3 thin film/TiO2 nanopillars heterostructure‑based chemiresistive sensor

Article References: Rasool, S. K., Singh, S. J., Lynrah, S. A., Kar, A. K., Selliyan, S., Elayaperumal, A., & Chinnamuthu, P. (2026). Enhanced NH3 sensing using a MoO3 thin film/TiO2 nanopillars heterostructure‑based chemiresistive sensor. Journal of Materials Science, 61(40), 30797-30811. https://doi.org/10.1007/s10853-026-13611-0

Image Credits: AI Generated

DOI: 10.1007/s10853-026-13611-0

Keywords: ammonia sensing, molybdenum trioxide, titanium dioxide, nanopillar heterostructure, chemiresistive sensor, glancing-angle deposition, gas selectivity, heterojunction, response recovery, low-ppm detection, Enhanced, sensing