Multi-Mode Sensor for Hydrogen Detection via Colorimetric and Electrical Signals Based on WO₃-Pd Films

2.1 Sensor fabrication

2.1.3 Pd deposition

Pd, which serves as a catalytic layer that dissociates hydrogen molecules into ionic forms, was deposited on top of the WO₃ thin film using an e-beam evaporator (Sci & Tech, Korea) with a high-purity Pd target (99.995% purity, VTM, Korea). The deposition process was performed under a pressure of 3 μTorr with a deposition rate of 0.1 nm/s. To investigate the effect of the Pd layer thickness on the sensing characteristics, we varied the Pd layer thickness to 2, 4, 6, and 8 nm.

The transmittance change (ΔT) and surface roughness (RMS) were measured for each sample to evaluate the influence of the deposition conditions of the Pd catalytic layer on the optical and physical properties of the WO₃ film. For experimental consistency, the WO₃ layer was fixed at a sputtering time of 70 min (corresponding to sample 4 in Table 1).

Table 1.

Deposition time of WO3.

Table 2.

Deposition thickness of Pd

Fig. 1.

3D schematic structure image of WO3-Pd sensor.

Fig. 2.

Schematic cross-sectional image of WO3-Pd Sensor.

Fig. 3.

WO3-Pd hydrogen sensor chip fabrication up to PCB wire-bonding.

Fig. 4.

Illustration of the sensor packaging process and resistance-based hydrogen detection in a sealed module.

To analyze the electrical response characteristics quantitatively, we deposited an additional metal electrode on top of the Pd catalytic layer. Electrode deposition was performed using an e-beam evaporator (Sci & Tech, Republic of Korea) using high-purity Cr (99.995% purity, VTM, Korea) and Au (99.999% purity, VTM, Korea).

A Cr layer was deposited as an adhesion layer, and an Au layer with a thickness of 3 nm was subsequently deposited to ensure electrical conductivity with a thickness of 50 nm. The electrodes were fabricated with an active area of 1 mm × 1 mm corresponding to each sensing region, enabling the stable and accurate measurement of resistance variations during hydrogen exposure.

2.2 Sensor package design

After the deposition, the multilayer structure was diced into individual sensor chips, each forming a single sensing unit. The fabricated sensor chips were then mounted onto a printed circuit board (PCB), and electrical connections between the sensor electrodes and the PCB circuitry were established through a wire bonding process. This process resulted in a packaged sensor module, enabling the electrical response of the sensor to be transmitted to external measurement equipment.

Based on the fabricated hydrogen sensing chip, a modular multi-mode hydrogen-sensing system was designed and developed considering practical operating environments. This system was not limited to a sensing function but was also equipped with additional features to enhance real-time responses and user convenience.

To effectively capture hydrogen gas and improve sensing sensitivity, we installed a miniature gas intake unit on the upper part of the module. This intake unit directs the gas flow into the sensor chamber, and its internal structure was optimized through a flow-path design to ensure that hydrogen is rapidly and selectively delivered to the sensing region. Consequently, the system achieved a shorter response time and enabled rapid detection, even at trace hydrogen concentrations.

In addition, when the detected hydrogen concentration exceeds a predefined threshold, the system automatically identifies the conditions and provides immediate warning signals to the user. The alarm module integrates both visual LED blinking and auditory buzzer alerts, which are activated according to pre-set criteria through automatic data analysis of the sensor output. These multi-alert functions enable intuitive and rapid confirmation of hydrogen leakage in industrial or enclosed environments, ensuring both the safety and practical applicability of the developed system.

Fig. 5.

Electrical signal measurement using MFC and I-V probe station.

2.3 Electrical signal measurement

To evaluate the electrical performance of the fabricated WO₃-Pd-based hydrogen sensor, we employed a mass flow controller (MFC, Atovac, Korea), a sensor probe station (Nextron, Korea), and an I–V source meter (Keithley 2400, USA). This measurement setup enables precise control of the gas flow rate and environmental conditions, such as temperature and humidity, making it suitable for quantitative analysis under simulated operating environments. Measurements were conducted under stabilized conditions of relative humidity (RH) 15% and 50^oC, with N₂ used as the background gas to control the hydrogen concentration. The sensor chip was mounted inside the probe station and its electrical response before and after hydrogen exposure was monitored in real time. Hydrogen gas was precisely introduced according to the set flow rate, and the change in the resistance of the sensor was used as a key indicator of the hydrogen-sensing behavior. From these data, major sensing parameters such as response, response time, and recovery time were quantitatively evaluated.

2.4 Colorimetric signal measurement

Ultraviolet–visible (UV–Vis) spectroscopy was used to evaluate the colorimetric characteristics of the fabricated sensors. This analysis enables the quantitative assessment of the optical changes in the sensor, and the color difference (ΔE*) value was introduced to quantify the degree of color variation. The ΔE* was calculated using the following equation:.

Fig. 6.

Schematic of UV–Vis equipment for colorimetric detection.

3.1 Variation in sensing characteristics with WO₃ thickness

Fig. 7 shows the variation in WO₃ film thickness as a function of deposition time, as determined by atomic force microscopy (AFM). As the deposition time increased, the WO₃ film thickness increased linearly, from 100.8 nm for Sample #1 (WO₃-10m) to 868.6 nm for Sample #5 (WO₃-90m), as summarized in Table 1.

Fig. 7.

Schematic illustration of WO3 thickness increase with sputtering time.

Fig. 8.

Time-dependent changes in the surface roughness of WO3.

The AFM analysis revealed that the RMS roughness increased from 1.642 to 3.346 nm with increasing film thickness, confirming that the surface roughness varied depending on the deposition conditions.

Fig. 9 presents the transmission electron microscopy–energy dispersive spectroscopy (TEM-EDS) cross-sectional image of the WO₃ film deposited by sputtering for 70 min. The analysis indicated that, during the initial deposition stage, a dense and compact film was formed, whereas when the thickness exceeded 150 nm, the morphology transitioned into a columnar and porous structure.

Fig. 9.

Cross-sectional TEM image and corresponding EDS line-scan analysis of the Pd/WO3 interface

Although the Pd catalytic layer was deposited under the same e-beam evaporation conditions, the measured Pd thickness varied depending on the WO₃ film thickness. For the thicker WO₃ layers, the Pd thickness was 4.93 nm, whereas that for the thinner WO₃ layers was approximately 3.97 nm. This difference suggests that the variation in surface roughness affected the uniformity of Pd deposition, leading to a dispersed distribution of Pd particles rather than a continuous film. In particular, the Pd nanoparticles tended to remain well-dispersed rather than agglomerate on surfaces with higher roughness.

Furthermore, as the thickness of the WO₃ film increased, the surface roughness gradually increased. Such morphological variations induced scattering and non-uniform distribution of Pd particles, thereby affecting the deposition characteristics. Consequently, we confirmed that the surface morphology of WO₃ had a critical influence on the growth behavior and microstructural formation of the Pd layer.

Fig. 10 shows the correlation between the surface roughness and electrical response characteristics of WO₃ thin films with varying thicknesses. A comparison between Sample #4 (WO₃–70 m) and Sample #1 (WO₃–10 m) revealed distinct differences in the microstructural features and corresponding electrical responses resulting from changes in the film thickness.

Fig. 10.

Effect of WO3 thickness on electrical signal behavior.

For Sample #4, the WO₃ film with a thickness of approximately 800 nm exhibited a dense morphology within the initial 200 nm region, followed by a transition to a columnar porous structure as the deposition proceeded. This morphological transition facilitated the penetration of hydrogen atoms (H), thereby promoting the formation of conductive pathways within the WO₃ matrix. The measured surface roughness of this sample was 2.775 nm, indicating that the deposited Pd nanoparticles were distributed non-uniformly and in a scattered form. Thus, hydrogen atoms dissociated by the Pd catalyst combined with oxygen species within the WO₃ matrix to form hydroxyl (OH) groups, leading to a reduction of WO₃ to WO_3-x. This reduction process increased the number of free W 5d electrons, thereby enhancing electrical conductivity and resulting in a negative resistance change.

In contrast, Sample #1, which consisted of a 100 nm-thick WO₃ film, maintained a compact morphology throughout the entire layer. The surface roughness was relatively low (1.642 nm), and the Pd particles formed a continuous interconnected structure. In this configuration, hydrogen adsorbed on the Pd surface formed PdHₓ, which interrupted the electron flow within the metallic layer, causing a positive resistance change.

Overall, as the WO₃ film thickness increases, changes in surface roughness and microstructure directly affect Pd deposition and the electrical response during hydrogen sensing. Therefore, controlling the WO₃ microstructure and optimizing the Pd nanoparticle distribution are critical factors for enhancing sensing performance and achieving optimal sensor design.

3.2 Optimization of transmittance by adjusting Pd thickness

Fig. 11 shows the surface roughness characteristics of the Pd films with varying thicknesses, as analyzed using AFM. When the Pd thickness increased from 2 to 8 nm, clear differences in the surface morphology and roughness were observed.

Fig. 11.

AFM Surface morphology and corresponding RMS roughness variation of Pd films with different thicknesses. (a) Surface roughness differences observed using AFM at varying Pd thicknesses. (b) Surface morphology of Pd films with varying thickness

• a) At a thickness of 2 nm, the surface roughness was 1.632 nm, indicating a uniform and smooth surface and suggesting that the Pd nanoparticles were evenly distributed during the initial deposition stage.
• b) When the thickness increased to 4 nm, the surface roughness increased to 2.816 nm, and the surface exhibited a highly non-uniform morphology, which can be attributed to the growth and agglomeration of Pd particles during the deposition process.
• c) At 6 nm, the surface roughness decreased slightly to 2.476 nm and the particle boundaries became more ordered, implying that localized coalescence among the Pd particles had begun.
• d) At 8 nm, the roughness further decreased to 1.917 nm, indicating a densely packed surface structure, which suggests that enhanced coalescence among the Pd particles resulted in a locally ordered surface morphology.

This trend indicates that the roughness of the Pd film varied nonlinearly with thickness, initially increasing owing to particle growth and then decreasing as the film became denser.

Larger surface roughness resulted in a greater exposed surface area of Pd, thereby enhancing its reactivity with hydrogen and influencing the amount of hydrogen diffusing into the WO₃ layer. From this perspective, the 4 nm-thick Pd film provided the most favorable structure for hydrogen dissociation and interaction between Pd and WO₃.

Fig. 12 shows the ΔT values of Pd films with different thicknesses. The ΔT values were normalized with respect to the highest response observed at a Pd thickness of 4 nm.

Fig. 12.

Normalized optical transmittance as a function of Pd thickness.

When the Pd thickness was 2 nm, the normalized ΔT value was 0.41 owing to the limited reactive surface area of the thin film. In contrast, at 4 nm, the ΔT reached its maximum, indicating that the active surface area of Pd was maximized, resulting in highly efficient hydrogen dissociation and adsorption. However, when the Pd thickness increased to 6 and 8 nm, the ΔT values decreased sharply to 0.17 and 0.14, respectively. This degradation can be explained by particle agglomeration and a reduced effective reaction area, which hinder hydrogen diffusion and deteriorate the sensing performance. In particular, excessive Pd thickness reduces the optical transmittance of the Pd layer itself, suppressing the colorimetric variation originating from the underlying WO₃.

Therefore, the 4-nm Pd film exhibited the highest surface roughness and strongest transmittance response, confirming that it provides the optimal thickness for hydrogen sensing. These results emphasize the importance of precisely controlling the Pd deposition thickness to enhance sensor sensitivity.

Fig. 13.

Response to hydrogen exposure in the range of 1%–10%.

3.3 Electrical signal measurement results

The WO₃-Pd hydrogen sensor was exposed stepwise to hydrogen concentrations ranging from 1% to 10%, followed by nitrogen purging to remove residual gas, while monitoring the time-dependent variation in electrical resistance.

The concentration sensitivity and recovery characteristics of the sensor were quantitatively evaluated using this process. Upon the introduction of hydrogen, the sensor exhibited an increase in resistance, which originated from the metal–hydride phase transition occurring in the Pd layer. When hydrogen molecules dissociate and are adsorbed onto the Pd surface, they diffuse into the lattice, causing a transition from the pure metallic phase (α-Pd) to a low-hydrogen solid solution phase (α-PdH_x). As the hydrogen concentration increases, this further transforms into the β-PdH_x phase. During this phase transition, the Pd lattice expands by approximately 3%–4%, reducing the mean free path of electrons and increasing the scattering rate, thereby leading to an overall increase in electrical resistance.

Conversely, when the hydrogen flow is stopped and nitrogen is introduced, the hydrogen atoms trapped within the Pd lattice desorb, resulting in the decomposition of the PdH_x phase and the restoration of the metallic α-Pd phase. As this process proceeds, the conduction pathway recovers and the sensor resistance gradually returns to its initial state.

Repeated hydrogen–nitrogen cycling tests demonstrated that the sensor exhibited consistent resistance changes with excellent reversibility and stability. Notably, as the hydrogen concentration increased from 1% to 10%, the proportion of the β-PdH_x phase formation increased correspondingly, leading to a progressive rise in resistance variation. This result confirms that the fabricated sensor exhibits a quantitative and concentration-dependent response to hydrogen.

To further investigate the low-concentration response, we introduced hydrogen stepwise from 0.15% to 2.50% (ppm), and the resistance variation over time was measured.

Upon hydrogen exposure, the resistance increased, and then decreased again upon nitrogen purging, confirming a typical reversible behavior that demonstrates stable sensor operation. As shown in Fig. 14, the resistance increased incrementally with increasing hydrogen concentration, indicating that the surface reaction on WO₃ proceeds proportionally to hydrogen concentration even at ppm levels. In particular, under 0.15% hydrogen exposure, the resistance changed from 2720 to 2740 Ω, confirming that the sensor can detect very low hydrogen concentrations sensitively. As the hydrogen concentration increased to 0.31% and 0.62%, the resistance variation became more pronounced, showing a clear and consistent electrical response even for small differences in the hydrogen concentration.

Fig. 14.

Response to hydrogen exposure in the range of ppm levels.

At relatively higher hydrogen concentrations of 1.25% and 2.50%, the resistance further increased to 2760 Ω, exhibiting a nearly linear correlation between resistance change and hydrogen concentration.

Consequently, the fabricated sensor maintained a high sensitivity even at ppm hydrogen concentrations, exhibiting concentration-dependent and quantitative response characteristics, thereby demonstrating its potential as a reliable hydrogen-sensing device.

Fig. 15 shows the experimental results of the fabricated sensor under repeated hydrogen exposure. Repeatability and reliability were evaluated over 50 consecutive exposure-purge cycles using 4% hydrogen gas. Through all 50 cycles, the resistance variation pattern remained highly consistent, demonstrating excellent repeatability and reversible hydrogen characteristics. During each hydrogen exposure and removal phase, the resistance values were clearly distinguishable, showing stable and reproducible changes within the same range across all cycles. Moreover, the sensor exhibited negligible variations in sensitivity, response time, and recovery time, even after repeated hydrogen exposure. These results confirm that the sensor maintains a reliable and stable sensing performance, ensuring long-term operational durability for practical hydrogen monitoring applications.

Fig. 15.

Electrical stability and repeatability characteristics of the WO3-Pd hydrogen sensor. (a) Electrical response of the sensor under 50 repeated exposures to 4% hydrogen (b) Enlarged view of repeated resistance increase in a specific section.

3.4 Colorimetric signal measurement results

Fig. 16 shows the quantitative analysis of the transmittance variation for the WO₃-Pd-based hydrogen sensor upon exposure to 1% hydrogen gas, as measured using UV–Vis spectroscopy. The sensor exhibited a sharp decrease in transmittance when exposed to hydrogen, followed by clear reversible colorimetric recovery once the gas was purged. In particular, at a wavelength of 650 nm, the transmittance decreased by 34% immediately after hydrogen exposure and recovered by 31% after gas removal, demonstrating the excellent optical response of the sensor.

Fig. 16.

Transmittance response to 1% hydrogen exposure

Fig. 17.

Change in sensor color difference upon hydrogen exposure

The measured ΔE* calculated using Equation (1) in the CIE Lab color space was 53.64, which corresponds to a distinct color change readily distinguishable by the naked eye. This result indicates that the colorimetric response of the sensor upon hydrogen exposure was visually pronounced, demonstrating its practical potential as a color-based hydrogen detection system.