Dataset and publication "WO3/CuWO4 nanocomposite thin films for humidity resilient acetone gas sensing"

Full Length Article WO 3 /CuWO 4 nanocomposite thin films for humidity resilient acetone gas sensing Nirmal Kumar a,* , Akash Kumar a , Jiˇ rí ˇ Capek a , Elisabetta Comini b , Stanislav Haviar a a Department of Physics and NTIS – European Centre of Excellence, University of West Bohemia in Pilsen, Univerzitní 8, Pilsen, 301 00, Czech Republic b SENSOR Laboratory, Department of Information Engineering (DII), University of Brescia, Via Branze 38, Brescia, 25133, Italy ARTICLE INFO Keywords: CuWO 4 Copper tungstate Sputter deposition MGA Acetone sensing ABSTRACT We demonstrate a high-performing and selective acetone gas sensor based on WO 3 /CuWO 4 nanocomposites produced by sequentially sputter-deposited WO 3 thin films and CuO nanoparticles, engineered to reduce the humidity interference. By optimizing deposition order and layer thicknesses, we harnessed synergistic p-n/n-n heterojunctions and the formation of a catalytic CuWO₄ ternary phase. The best performing configuration (20 nm of tungsten oxide film on top of nanoparticles) exhibits a high response (S =23) to 10 ppm acetone at 300 ◦C, fast response in dry/humid conditions (38 s/58 s), and low detection limit (0.6 ppm). More importantly, the sensor exhibited >95 % retention of its response in 90 % relative humidity compared to a loss of >50 % for pristine WO 3 . The reduced humidity interference is assigned to heterojunction formation at the WO 3 /CuWO 4 interface, Lewis acid sites that allow for acetone selective adsorption, and bulk-dominated conduction. This noble-metalfree acetone sensor overcomes a known shortcoming of metal oxide-based sensors, enabling accurate acetone detection in humid environments for breath-based disease diagnosis (e.g., diabetes) and industrial safety monitoring. 1. Introduction Volatile organic compounds (VOCs) such as acetone are widespread in industrial emissions, household products, and human breath, and therefore, their detection remains very crucial for environmental monitoring, occupational safety, and clinical diagnosis [1,2]. Among these, acetone is particularly significant due to its application as a biomarker for diabetes and its prevalence in industries [3,4]. In working environments, prolonged exposure to acetone concentrations greater than 170 ppm can lead to severe central nervous system depression, respiratory irritation, and eye damage, while chronic exposure leads to pharyngitis, bronchitis, and dermatitis. The Occupational Safety and Health Administration (OSHA) has set an 8-hour threshold limit value (TLV) of 750 ppm, although unexpectedly lower concentrations of acetone (50–100 ppm) have resulted in headaches, dizziness, and nausea [1,4]. In addition to occupational risks, acetone is an essential biomarker for non-invasive medical diagnostics. Diabetic patients have elevated breath acetone levels (>1.8 ppm), much higher than those of healthy individuals (0.3–0.9 ppm) and are therefore suitable for early disease diagnosis and metabolic monitoring [1,5,6]. For this reason, it is essential to create sensors that can accurately measure acetone at sub-ppm concentrations for environmental safety, industrial hygiene, and point-of-care diagnostics. The construction of sensitive, selective, and stable sensors for the detection of acetone continues to be a critical issue in the area of chemical sensing. Metal oxide semiconductors (MOS), tungsten trioxide (WO 3 ) among them, have attracted researchers for their application in gas sensing due to their high sensitivity, thermal stability, and tunable electronic properties [3,7–9]. On the other hand, pure WO 3 sensors have limitations such as poor selectivity, slow response and recovery rates, and significant performance loss upon exposure to humid conditions [1,10–12]. These limitations render the pure WO 3 unsuitable for real-world applications where humidity and interfering gases dominate. Additionally, for the clinical diagnosis purpose, it is important to the influence of humidity because human breath contains >90 % of humidity. To overcome these challenges, the combining of WO₃ with other MOS is often employed to build heterostructures that enhance the sensing performance based on synergistic effects. Particularly, mixing ptype copper oxide (CuO) with n-type WO₃ to form p–n heterojunctions for facilitating charge separation as well as enhancing surface reactivity * Corresponding author. E-mail address: [email protected] (N. Kumar). Contents lists available at ScienceDirect Applied Surface Science Advances journal homepage: www.sciencedirect.com/journal/applied-surface-science-advances https://doi.org/10.1016/j.apsadv.2025.100894 Received 10 September 2025; Received in revised form 15 October 2025; Accepted 29 October 2025 Applied Surface Science Advances 30 (2025) 100894 Available online 6 November 2025 2666-5239/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). [13,14]. In addition, ternary phase synthesis, such as copper tungstate (CuWO 4 ) has been reported to improve electrical conductivity and catalytic activity and improve sensor response and stability [15–18]. In this work, we present a comprehensive study on nanostructured CuO–WO 3 composite films synthesized using sequential deposition [16, 19,20]. We combined CuO nanoparticles (NPs) prepared by Magnetron Gas Aggregation Chamber (MGA) with thin solid films of WO 3 synthesized by conventional reactive magnetron sputtering. Detailed investigation of morphology and structure allowed us to change the nanocomposite nanostructure by tuning the synthesis parameters, such as the order of deposition of the mentioned materials or the thickness of the solid WO 3 film. The optimized structures were assembled as a conductometric gas sensor for acetone gas, which exhibits improved sensitivity, response time, and excellent tolerance to humidity. Besides confirming the efficacy of CuO–WO 3 composites in the selective detection of acetone, this work highlights the mechanisms responsible for their remarkably enhanced performance. In this work, we investigated a sensing material that fundamentally overcomes the challenge of effect of ambient humidity. The presented composite based sensor exhibits a negligible performance variation in sensing response (>95 % retention) when tested in dry and 90 % humid environment, a level of humidity resilience that marks a significant step forward for practical acetone detection. This exceptional stability strengthened its potential for practical, realworld applications. The findings advance heterojunction engineering and metal oxidebased gas sensors’ phase transformation, paving the way towards the realization of the next-generation sensing systems that are more selective, stable, and environment-resilient. 2. Experimental 2.1. Synthesis of materials The CuO nanoparticles (NPs) were synthesized using a custom-built system. This system featured a magnetron gas aggregation (MGA) source mounted in a DN 200 ISO-K six-way cross vacuum chamber equipped with a rotating substrate holder and a load-lock system. The main chamber was evacuated by a turbomolecular pump backed by a scroll pump. The base pressure in the aggregation chamber was 9 ×10 –4 Pa, and the working pressure was 130 Pa. The copper metal target was sputtered using DC power supply (GEN 600–1.3, TDK-Lambda, USA) operated at a constant discharge power of 25 W, more details can be found in Refs. [19,21]. WO 3 films were deposited in a LH Z400 (Leybold-Heraeus, Germany) cylindrical chamber (25 L volume) using a tungsten 72-mm metal target at a power of 60 W [15]. The chamber, pumped by a turbomolecular pump with scroll pump backing, reached a base pressure of at least 1 mPa prior to each run. An argon/oxygen gas mixture was used. The sputtering deposition started by setting the argon pressure at 0.7 Pa, followed by the addition of oxygen to achieve an O 2 : Ar ratio of 1:3. The substrate-target distance was fixed at 70 mm. Substrates were held at a temperature of 400 ◦C. CuO NPs and a WO₃ thin film were combined to form a composite system. The materials were always deposited in a sequential way when the deposition of a sparse layer of NPs was followed by the deposition of WO 3 thin film or vice versa. The thickness of the WO 3 film was altered while the density of particles was fixed following the recipe from our previous paper [19]. Samples for structural analyses were deposited onto 10 ×10 mm 2 Si (100) wafers, the gas sensing samples were deposited on 9 ×9 mm 2 quartz glass substrates (acetone sensitivity results), and 3 ×5 mm 2 alumina substrates (selectivity measurements presented in the Supplementary data). 2.2. Structural analysis XRD analysis was performed using a diffractometer (X’Pert PRO, PANalytical, UK) configured in the Bragg–Brentano setup with a Cu K α radiation source. Raman spectroscopy (LABRAM HR Evolution, Horiba Jobin Yvon, France) employing a 532-nm laser was used to understand the crystalline phases to support the XRD results. Morphological imaging of the samples was performed using a scanning electron microscope (SEM) (SU-70, Horiba Ltd., Japan) using primary electron energy of 10 keV. TEM micrographs were acquired using a JEOL JEM-2200FS microscope operating at an accelerating voltage of 200 kV in bright field mode. 2.3. Sensing response measurement The sensing response was measured on quartz glass samples equipped with four 2 ×2 mm 2 platinum (Pt) electrodes in each corner. A custom-built measurement system was used. The setup consists of four stainless-steel clamps coated with chromium and platinum, housed in a brass chamber with a total inner volume of 3.7 cm 3 . The specimen is heated to working temperature by a ceramic hot plate (Bach Resistor Ceramic GmbH, Germany). Resistance measurements were taken at a fixed current value (PCS 6220, Keithley, USA) using a four-point configuration (Voltmeters 6514, Keithley, USA) at various temperatures and concentrations of target gas. Synthetic air (N₂/O₂, 79:21) was generated using two mass flow controllers (Alicat Scientific Ltd., USA). The humidified air was generated by bubbling a fraction of the synthetic air supply through a water bath maintained at temperature 24 ◦C. The resulting humid air stream was mixed with the remaining dry synthetic air. The resulting relative humidity was monitored by an in-tube-placed humidity sensor (SHT 85, Sensirion, Switzerland). By adjusting the ratio of the flow between the humidified air and dry air stream, the desired relative humidity was achieved. For instance, supplying 100 sccm of humid air and 0 sccm of dry air resulted in 90 % relative humidity, while 57 sccm of humid air and 43 sccm of dry air corresponded to 40 % relative humidity. The target gas, acetone, was introduced close to the measurement chamber through a separate flow controller and the final mixture is introduced into the chamber where sample is placed on a hot plate. In order to regulate the acetone concentration, 5 µL of liquid acetone (99.95 % purity) was vaporized in a 5-L glass flask, which was connected to a controlled flow of N 2 gas. The acetone vapors-N 2 mixture was further diluted by a regulated flow of synthetic air. The acetone concentration was calculated by setting the values of the flows of the synthetic air and N 2 in the flask, as illustrated in Fig. 1. For the purposes of this paper, the Sensing Response for n-type materials is defined as: S=Ra Rg ,(1) where S is the sensing response, and R a and R g represent resistances in the presence of synthetic air and a mixture of target gas and air, respectively [11,22]. Eq. (1) was used for calculating sensing response towards reducing gas, for the oxidising gas, inverted fraction was used, i. e., R g /R a . Prior to sensing response measurements, all the sensors were heated at 400 ◦C in the presence of synthetic air for at least 2 h within the same sensitivity measurement system. Such treated samples are then denoted as “stabilized” samples in contrast to “as-deposited” samples. The samples, which are further heated up to 600 ◦C, are denoted as “annealed” samples. For the sake of clarity, the naming of the samples is simplified in a format such 20 WO ‒ NPs. This indicates that a 20-nm-thin film of tungsten oxide was deposited over cupric oxide nanoparticles, which are denoted as NPs. Single-layer 40 nm WO 3 films are denoted as 40 WO, etc. The cross-sensitivity measurement was carried out by employing a N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 2 custom-built gas sensing system, with the capability to simultaneously test up to ten devices within a 1-L chamber. This test chamber was housed inside a thermostatically controlled unit (Angelantoni, Italy, model MTC 120), maintained at 20 ◦C. Various gases—including hydrogen (H₂, 99.999 % purity), ethanol (C₂H₅OH, 99.98 % purity), acetone (C₃H₆O, 99.95 % purity), nitrogen dioxide (NO₂, 99.98 %), and ammonia (NH₃, 99.98 % purity)—were introduced into the system at a flow rate of 200 sccm, under conditions of 40 % relative humidity (RH), to assess the sensor’s selective performance [23,24]. 3. Results and discussion In this section, we provide an overview of the structure and surface characteristics of the synthesized nanostructured materials. Subsequently, we analyse their sensing behavior and highlight the differences among the prepared films. Towards the conclusion, we explore the impact of structural differences on sensing efficiency. While numerous combinations of films and nanoparticles were prepared, we focus our discussion on those that exhibit noteworthy structural features and/or demonstrate significant sensing response. Hence, all the films showcased in this paper are constructed on a foundation of 40or 20-nm-thin films of WO 3 and CuO NPs with a mean diameter of 10 nm. Furthermore, the samples are heated up to 400 ◦C for stabilization and subsequently measured for the sensing performance for various gases. 3.1. Microstructure The structure of the prepared composites was studied using XRD and Raman spectroscopy. Fig. 2a, and Fig. 2b show the XRD pattern and Fig. 1. Scheme of four-point resistance measuring system. Fig. 2. XRD spectrograms (a) and Raman spectra (b) of the as deposited (faded) and measured (solid) films. The orange curve in Raman spectra is corresponding to Si substrate. N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 3 Raman spectra of the prepared nanostructured composites, respectively. The XRD patterns of the composites primarily exhibit a monoclinic WO 3 phase, with faded curves representing the as-deposited composites and solid lines indicating the stabilized composites. The as-deposited WO 3 thin film initially shows a multiphase structure, comprising both monoclinic (PDF Card No 04–005–4272) and tetragonal (PDF Card No 04–007–2426) phases. However, after stabilizing at 400 ◦C, the films’ structure changes to a predominantly monoclinic phase (PDF Card No 04–005–4272). The Raman spectra support the XRD findings. The peaks observed near 263, 705, and 802 cm -1 are attributed to the monoclinic WO 3 phase [22,25,26]. In XRD patterns, it can be seen that the (002) reflection at position 23.09◦is more pronounced than others in the pure WO 3 film and NPs on the film. However, the evolution of plane (200) is observed at position 24.36◦in the samples where the film was over the Cu NPs. Importantly, the stabilization process also leads to the diffusion of some nanoparticles (NPs) into the WO 3 matrix, resulting in the formation of copper tungstate, CuWO 4 , as evidenced by the XRD and Raman spectra in the film on NPs composite and previously reported by us in detail in [16,27]. This phase matches well with the standard triclinic/anorthic CuWO 4 reference (PDF Card No 04–009–6293), and the peak positioned around 900 cm -1 corresponds to anorthic CuWO 4 [11,15]. The CuWO 4 content in the samples is insufficient to be clearly visible in the XRD pattern due to the peak overlap with WO 3 . In composite samples NPs– 40 WO and 20 WO–NPs, the CuWO 4 corresponding Raman peaks are also hard to detect. However, they can be found in the Raman spectrum when the concentration is high enough, as in sample 40 WO ‒ NPs. Although, in our previous work, where the higher amount of the CuO and WO 3 mixed to form CuWO 4 /WO 3 composites for H 2 sensing, the amount of CuWO 4 is sufficient to detect in XRD and Raman [11,15,27]. It was also observed when the amount of CuO NPs increased in the CuO/WO 3 nanocomposites, a metastable phase γ-CuWO 4 is formed at low annealing temperatures and change to stable anorthic CuWO 4 at higher annealing temperature [16,27]. The formation of CuWO 4 is evident in films that we further annealed at 600 ◦C for 3 h. After annealing, the CuO NPs and WO 3 film have mixed together and transformed almost entirely into tungstate. This can be seen in the XRD patterns and Raman spectra shown in Fig. 2 by the grey curves named as “annealed”. These findings underscore the impact of stabilization on the crystallographic properties of composite films. Finally, the temperature value 400 ◦C is selected to stabilize the composites because of the presence of multiphase, expecting to form heterojunctions, which help to enhance the sensing response [11,19,28]. 3.2. Morphology Fig. 3 represents the SEM micrographs of the composites. The top views of as-deposited and stabilized WO 3 films are shown in Fig. 3a and Fig. 3e, respectively. A comparative analysis reveals that the apparent grains in the stabilized WO 3 films are more prominent and show welldefined boundaries. This is in accordance with XRD and Raman spectroscopy findings, which indicated better crystallinity. The improved crystallinity suggests that the thermal treatment facilitates structural reorganization, resulting in more structured film morphology. In Fig. 3b, Cu NPs are observed on the WO 3 film surface, appearing with slightly higher contrast. Higher magnification of the composite is also shown in the insets. Micrographs in Fig. 3c, d display larger “particles”, which is caused by the simple over-layering of NPs with WO 3 films. The apparent particles are bigger when the film thickness is 40 nm (Fig. 3c) and considerably smaller for the 20 WO–NPs composite. Fig. 3. SEM micrographs of as deposited a) 40 nm WO 3 , b) Cu NPs on 40 nm WO 3 film, c) 40 nm WO 3 film on CuO NPs and d) 20 nm WO 3 film on CuO NPs and stabilized e) 40 nm WO 3 , f) Cu NPs on 40 nm WO 3 film, g) 40 nm WO 3 film on CuO NPs and h) 20 nm WO 3 film on CuO NPs. The insets are magnified micrographs of the respective surfaces. The scales are the same for all SEM micrographs and insets, indicated in the top right micrograph (d). N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 4 The formation of the tungstate phase is apparent for both configurations 40 WO–NPs, and NPs– 40 WO. From the morphological point of view, when NPs are on top of the layer, the restructuring respects their original deployment less. Particles may migrate towards each other, or growing oxides may fill the space between them more easily. In contrast, the texture of NPs stranded under the tungsten oxide over deposit is visible even after stabilization. More importantly, since the WO 3 deposition is performed at elevated temperature, the growth of the tungstate phase takes place right during the deposition. On the other hand, when the NPs are deposited on top of a pre-deposited thin film, they arrive at the surface with relatively small kinetic energy, and the formation of an intermixed phase is initiated during the consequent stabilization. These morphological differences are important during the sensing response discussion, as the NPs under the film are not agglomerating and promote a uniform intermixing with surrounding WO 3 , resulting in densely packed CuWO 4 nano-islands (Fig. 3g and Fig. 3h) as observed in one of our previous works [15], where these structures were generated in different way. The magnified micrographs of the composites are shown in the insets of the corresponding sample (Fig. 3b-h), where the microstructure of CuWO 4 over WO 3 grains can be seen clearly. The Bright field TEM micrograph and HRTEM micrograph in Fig. 4 show the structure of the 40 WO–NPs sample, where the particles were over-deposited by the tungsten oxide. The darker areas clearly show the particles formed by the original copper oxide particles. The HRTEM data show the nanocrystalline nature of the film between the particles. Unfortunately, it is not possible to distinguish tungstate, which is most probably formed on the edges of the nanoparticles, since, as mentioned in the XRD paragraph, the interplane distances of copper tungstate and tungsten oxide are very similar. 3.3. Sensing response The prepared composites were stabilized in the measuring chamber at 400 ◦C in synthetic air for at least two hours before the sensing response measurement. After stabilizing the fabricated sensors, the response curves were recorded for various temperatures and concentrations of acetone in dry and humid environments. Firstly, we demonstrate how the composites outperformed pristine WO 3 film, and then we explain the enhancement of the response in specific composites and how the influence of humidity is depressed. Fig. 5 illustrates the response-recovery transients, where sensor resistance is plotted against time for 10 ppm acetone at 300 ◦C. The response for pristine 40-nm WO 3 film ( 40 WO), CuO NPs on 40-nm WO 3 film (NPs ‒ 40 WO), 40-nm WO 3 film on CuO NPs ( 40 WO ‒ NPs), and 20-nm WO 3 film on CuO NPs ( 20 WO ‒ NPs) are shown by black, red, green, and blue curves, respectively. The cycle of 10 ppm of acetone is shown by the dashed brown curve. As can be noted, the resistance of all the sensors immensely decreased on exposure to acetone vapors and quickly regained the baseline resistance values when re-exposed to synthetic air. This behaviour is expected for reducing gases such as acetone and n-type semiconductors (WO 3 and CuWO 4 ) [10,29,30]. The CuO NPs are p-type [19] in their as-deposited state, but due to the reforming of a considerable portion of copper oxide to the tungstate, the response of pure CuO does not play a role. The response is improved when pristine WO 3 film is decorated by Cu NPs (NPs ‒ 40 WO), shown by the red curve. For the reversed system ( 40 WO ‒ NPs), the response further increased (green curve). However, the response for 20 WO ‒ NPs (S =23) is significantly higher than that of 40 WO (S =4.5), NPs ‒ 40 WO (S =6.3), and 40 WO ‒ NPs (S =10). The enhancement in the response value for the composites could be mainly attributed to the formation of a CuWO 4 ternary phase and heterojunctions among the nanostructures of CuO, CuWO 4, and WO 3 . Due to the deposition of the WO 3 layer on NPs at higher temperature, the Fig. 4. Bright field (top) and high-resolution TEM (bottom) micrographs of 40 nm WO 3 on NPs. Fig. 5. Response-recovery transients toward 10 ppm acetone in dry air at 300 ◦C. N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 5 formation of CuWO 4 is more significant for layer 40 WO ‒ NPs and 20 WO ‒ NPs as the surface morphology can be seen in SEM micrographs in Fig. 3f, Fig. 3g and Fig. 3h The tungstate phase is more uniformly distributed on the surface of 40 WO ‒ NPs and 20 WO ‒ NPs than NPs ‒ 40 WO. The thinner film 20 WO ‒ NPs has a higher, i.e., advantageous, surface-tovolume ratio, which explains the larger response. Fig. 6a shows the dynamic response of all the sensors at 300 ◦C in a dry environment towards various concentrations of acetone in the range from 0.6 ppm to 10 ppm. The value of the response is significantly increased for higher concentrations of acetone. It is found that specifically 20 WO ‒ NPs responded even to lower concentrations of acetone (0.6 ppm) while other sensors are either unstable or do not respond at all to such a low concentration (<1 ppm). Moreover, the sensing response was also measured at higher concentration up to 3000 ppm during optimizing the concentration of acetone. However, 40 WO – NPs was degraded after 3000 ppm. Considering the degrading limitations, only the best performing sensor ( 20 WO ‒ NPs) was tested at 100 ppm (S =97) and 600 ppm (S =173). The effect of operating temperature can be seen in Fig. 6b The sensors are measured at various temperatures after being stabilized at 400 ◦C. The resistance of the sensors significantly decreases in air with increasing temperature because the concentration of charge carriers in MOS sensing materials increases with rising temperature. For pristine WO 3 film, the sensing response increases with temperature, while for the composites, the value is maximum at an optimised temperature, i.e., 300 ◦C. The baseline resistance is found to be increasing from pristine WO 3 to the 20 WO–NPs as illustrated in Fig. 6b by dashed lines. The baseline values resistances are higher in the composites compared to WO 3 film due to formation of ternary phase CuWO 4 and heterojunctions. The value is significantly increased for 20 WO–NPs which is not only due to formation of heterojunctions but also the electron depletion layer (EDL) takes up the entire thinner film. Fig. 6c and 6d compare the sensing performance of all the samples in 0 % (dry) and 90 % (humid) relative humidity (RH). In Fig. 6c, it can be seen that the sensing response value for pristine WO 3 is reduced by >50 % in the 90 % RH compared to dry conditions. The effect of humidity is suppressed as the WO 3 is altered by forming composites. For example, the change in the sensing response value is <50 % for NPs ‒ 40 WO and further suppressed for 40 WO ‒ NPs and almost vanishes for 20 WO ‒ NPs. For the best performing material, 20 WO ‒ NPs, the sensing response is also measured at 40 % RH, the response value is found to be within the values for dry and 90 % RH conditions. The comparison of response times and baseline resistances in dry and humid conditions for all the materials is shown in Fig. 6d Compared to the pristine WO 3 film, the baseline resistance is elevated for the composites in both dry and humid conditions (shown in Fig. 6d by purple solid and dashed lines, respectively) due to the formation of heterojunctions among CuO, CuWO 4 , and WO 3 . The reduction of conduction path within the film is logically most pronounced for the 20 WO ‒ NPs, where the film has the highest resistance, mainly caused by the formation of nanoscopic heterojunctions of WO 3 –CuWO 4 effectively filling most of the film’s volume and reducing the conductive path. The absolute values of baseline resistance in dry air are generally higher than the values in humid air, which is mainly due to adsorption of water molecules on the surface sites of the sensors. However, the variation in the baseline resistance for dry and humid conditions for each sensor is getting narrower from pristine 40 WO to 20 WO ‒ NPs, as shown in by the solid and dashed curves in Fig. 6d. The solid bars in Fig. 6d show the response time in dry conditions, while the faded bars with blue outline correspond to response time in Fig. 6. Plot for Sensing Response variation a) with concentrations (solid curves) at 300 ◦C and their linear fit (dashed curves) along with equations, b) with temperature towards 10 ppm of acetone, c) with relative humidity (set at 24 ◦C) towards 10 ppm of acetone at 300 ◦C, and d) Response time along with baseline resistance in dry and humid air at 300 ◦C towards 10 ppm of acetone. N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 6 humid conditions. It is observed that the composites respond faster than the pristine WO 3 film in both conditions, and 20 WO ‒ NPs performs at the quickest response time. The response time for all the sensors is increased in humid conditions compared to dry conditions. However, this change in response time is suppressed in the composites. For example, the response times for standalone WO 3 film are 141 s and 253 s in dry and humid conditions, respectively, while for 20 WO ‒ NPs the values are 38 s and 58 s. A comparison of the sensitivity parameters with other recent work is illustrated in Table 1, where a few WO 3 -based sensing materials with various modifications to improve performance at dry/humid conditions, most of the materials are selective to acetone. Also, the limit of detection (LOD) is compared along with optimum concentrations (OC) of the target gas. The effect of humidity is also compared in the table where the change (reduction) in the value of sensing response is compared in dry and humid conditions. To assess the selectivity of studied materials, some selected material combinations were fabricated on alumina substrates and measured towards gases such as H 2 , ethanol, NH 3 , acetone, and NO 2 . The response of the samples is recorded for 10 ppm of the concentration for all the gases, except H 2 , where a 50-ppm value was used because the samples did not respond to 10 ppm of H 2 at any temperature from the tested range (from 150 ◦C to 400 ◦C). The response values towards all the gases at various temperatures are shown in Figure S1 in the Supplementary data. Response for 40 WO ‒ NPs is found to be better than that of NPs ‒ 40 WO at all the temperatures and mostly all the gases. However, all materials are found to be highly selective for acetone. 3.4. Discussion of sensing mechanism The sensing mechanism of the composite films towards the acetone is primarily governed by electron transfer driven by the surface redox reactions, coupled with the acetone adsorption and desorption processes [29,30,36]. Upon exposure of the composite film to the air, oxygen molecules adsorb onto the film surface. These molecules extract electrons from the conduction band near the surface and form chemisorbed oxygen species (O− 2,O−, and O2−). The reaction processes can be written as: O2(gas)→O2(ads),(2) O2(ads) + e−→O− 2(ads),(3) O− 2(ads) + e−→2O−(ads),(4) O−(ads) + e−→O2−(ads).(5) The electron migration reduces the charge carrier concentration in the films, results in the formation of an electron depletion layer (EDL) near the surface and consequently increases the sensor resistance. For a given material and working temperature, one can expect the preferred way of oxygen adsorption [30,36–38]. For temperatures used in this work, the dominant adsorbed oxygen species is O−. Further, when the surface is exposed to acetone vapor, preadsorbed O−species oxidize the acetone molecules and release the trapped electrons back to the semiconductor. This process then diminishes the EDL and reduces the sensor resistance. The reaction pathways typically involve the oxidation of acetone and the formation of CO 2 and H 2 O as byproducts, as described by the equation: CH3COCH3+8O−→3CO2+3H2O+8e−.(6) In the humid conditions, the adsorption of water molecules leads to the formation of a hydroxyl group (OH−) on the surface that are bound to the metal ion (M+) or with the lattice oxygen (OO) leaving oxygen vacancies (V2+ O) in the material. The mechanism is described by the equation [11,39–41]: H2O+2M+OO→2(M+−OH−) + V2+ O+2e−.(7) Typically, acetone oxidation relies on the adsorbed oxygen species (O−, and O2−) but in humid conditions, the water vapors replace some of these species with (OH−) ions and shifts the reaction to: CH3COCH3+6OH−→CH3COOH +CH3CHO +6e−.(8) Water vapors and acetone both adsorb onto the material’s surface, competing for the same adsorption sites, ultimately resulting in weakening the sensor response signal as absorbed in most of the composite sensors [10]. The change in sensing response of the pristine WO 3 and the composites is already described in both conditions. The improvement in the responses of the composites for the acetone is primarily attributed to the synergetic effects at the heterojunction interface and catalytic transformations. The formation of p-n heterojunctions between p-type CuO Table 1 A comparative analysis of WO 3 -based acetone gas sensors. (T – working temperature, RH – relative humidity, OC – optimum concentration, LOD – limit of detection, τ 90 – response time). Sensing Material Response formula used, S (or S %) T ( ◦C) RH SOC (LOD) [ppm] τ 90 [s] Effect of Humidity 2D WO 3 Nanosheets [5]S=Ra Rg 300 30 % 14.7 50 (0.17 ppm) 8 NA Sea Urchin-like WO 3 [30]S=Ra Rg 200 Dry 27.2 100 (2 ppm) 3 44 % reduction from 0–90 % RH    90 % 15.0    Rh-Loaded WO 3 nanosheets [31]S=Ra Rg 250 20 % 48 20 (0.5 ppm) NA NA Daisy like WO 3 [32]S=Ra Rg 260 30 % 8.4 5 (0.3 ppm) ~30 % reduction from 30–90 % RH    90 % 5.8    Cr doped Urchin WO3 [33]S=Ra Rg 250 25 % 75 100 (0.2 ppm) <100 72 % reduction from 25–90 % RH    90 % 3    WO 3 thin film [7]S=Ra Rg−1300 Dry 2.5 20 (0.2 ppm) <100 NA WO 3 nano fibres [34]S%=|Ra−Rg| Ra×100 150 11 % 90 % 10 (0.5 ppm) 58 20 % reduction from 11–94 % RH    90 % 70 %    WO 3 Nanoneedles [35]S=Ra Rg 300 dry 19.72 10 (2.4 ppb) NA NA WO 3 /CuWO 4 [This work] S=Ra Rg 300 dry 23 10 (0.6 ppm) 38 4.3 % reduction from 0–90 % RH    90 % 22 58  N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 7 and n-type WO 3 widens the electron depletion layer, amplifying resistance changes during acetone exposure (a reducing gas) due to accelerated charge recombination [42,43]. Additionally, as can be seen in SEM micrographs and Raman spectra, stabilizing the composites at 400 ◦C induces partial CuWO 4 formation, facilitating electron transfer and catalytic acetone oxidation [42]. Furthermore, the formation of n-n heterojunction between n-type CuWO 4 and n-type WO 3 leads to further modulation of the depletion layer. The optimal CuWO₄ morphology in 40 WO ‒ NPs and 20 WO ‒ NPs further enhances this effect, explaining their superior response over NPs ‒ 40 WO [11,15,44–46]. The heterojunctions between WO 3 and CuWO 4 further modulate EDL and modify carrier mobility during the sensing mechanism, making the composite a better choice over the individual WO 3 or CuWO 4 . In the case of pristine WO 3 , the sensing mechanism is governed by direct oxidation of acetone, as shown in Eq. (6). However, in the case of WO 3 /CuWO 4 heterojunction, the carbonyl oxygen (C=O) of acetone bonds with Cu 2+ Lewis acid sites, provided by CuWO 4 , polarizes the C=O bond and facilitates the dissociation [44,47]. The reaction pathway for the catalytic diffusion on the surface of CuWO 4 –WO 3 composite takes place in a few sequential steps, including formation of intermediates such as enolate and acetate, followed by the oxidation of the intermediates to the direct byproducts shown by the reactions below: CH3COCH3+Cu2+→CH2C= (O−)CH3+H++Cu++e−(9) CH2C= (O−)CH3+Oo→CH3COO−+HCHO +Vo(10) CH3COO−+HCHO +6O−→3CO2+2H2O+6e−(11) The catalytic adsorption mechanism is faster than the direct oxidation of acetone on the surface of WO 3 and also can take place at lower operating temperatures (RT–250 ◦C) [48]. At 90 % relative humidity, the WO 3 –CuWO 4 composite maintains a stable response to acetone due to several complementary mechanisms. At the interface of CuWO 4 /WO 3 heterojunction, a built-in electric field is created at the junction, which repels polar water molecules via dipole interactions, while simultaneously promoting the diffusion of target gas towards active adsorption sites [11,49]. Furthermore, CuWO 4 coverage passivates the WO 3 surface by blocking hydroxyls (OH−) groups, leading to a reduction of available sites for H 2 O adsorption. The humidity tolerance is further enhanced when the WO 3 layer thickness is reduced to 20 nm. The EDL saturates the entire film at this scale. Consequently, in such thin thicknesses, further adsorption of water molecules induces minimal additional resistance changes, rendering humidity effects negligible. This thin-film configuration minimizes defect density within the WO 3 . Combined with the defect-passivating effect of the CuWO 4 overlayer, this dual strategy significantly suppresses OH−formation by over 40 % [47,50,51]. To wrap this up, humidity interference decreases from pristine WO 3 to the 20 WO ‒ NPs formulation due to a few key factors: (i) CuWO 4 covers OH−groups on WO 3 surface, reducing water’s competitive adsorption, this layer acts as a physical barrier, reducing direct water condensation on active sites of WO 3 while allowing smaller acetone molecules to diffuse through [44,50], (ii) the thinner WO 3 coating in 20 WO ‒ NPs (around 20 nm) exposes relatively more CuWO 4 to be exposed, shifting the material’s conduction behavior from a humidity-sensitive surface regime to a more stable, bulk-dominated mechanism [11,36,43,52], and iii) the Lewis acidity of Cu2+and W6+ion of tungstate phase is borderline acid and hard acid species, respectively [53,54]. According to the theory of Hard and Soft Acids and Bases (HSAB), the affinity towards humidity can be reduced on CuWO 4 surface because of the hard base properties of water molecule [53,55]. Also, the lower thickness of the WO 3 film for composite 20 WO ‒ NPs makes the composite thinner. During the adsorption of oxygen species, the electron depletion layer (EDL) spans almost the entire thickness of the film, let us say complete depletion, which can be concluded by higher baseline resistance (Fig. 5d). As a result, 20 WO ‒ NPs exhibit minimal changes in the magnitude of response to acetone even under high humidity conditions. As was described in Fig. 6d, the response times increase in humid environments for all composites due to competitive adsorption dynamics. Water molecules occupy surface active sites, slowing acetone diffusion and displacement kinetics [1]. For 20 WO ‒ NPs, the unchanged response magnitude but increased response time in humidity stems from kinetic barriers imposed by water vapors, while thermodynamic reaction equilibrium remains unaffected. In humidity, the water molecules may be physiosorbed, which can slow down acetone diffusion to the active sites and compete for chemisorption sites, delaying adsorption/- desorption kinetics and also responsible for lower change in base resistance value. Additionally, the thin WO 3 layer (20 nm) and percolating CuWO 4 nanodomains network maintain bulk-limited conduction, preserving response magnitude by ensuring efficient electron transfer during acetone reactions. However, residual water at grain boundaries or CuO sites creates transient diffusion barriers, slowing acetone access to catalytic interfaces without altering the fundamental redox activity [43,45]. The sensors are selective towards acetone due to the combination of acetone’s unique molecular properties and specific structural, electronic, and surface characteristics, of WO 3 along with its sensing mechanism [5,56]. Moreover, acetone has a carbonyl functional group (C=O), which plays a key role in the selectivity. It adsorbs on the WO 3 surface via the oxygen atom in the carbonyl group, forming a stable intermediate (e.g., acetate or enolate species) before oxidizing to CO 2 and H 2 O [34,57]. The C=O bond in acetone is highly polar, leading to stronger electrostatic interactions with the WO 3 surface oxygen vacancies [7,58], while other VOCs may require different active sites. Additionally, sputter-deposited monoclinic WO 3 exhibits a considerable texture exposing a high-energy plane (002), which is preferred by acetone molecules for adsorption. These facets exhibit higher surface energy (~ 1.56 J/m 2 ) than (200) (1.43 J/m 2 ) or (020) (1.54 J/m 2 ) planes, facilitating stronger carbonyl group (C=O) interactions. WO 3 with oxygen vacancies creates localized electron-rich sites, and these vacancies acquire electrons from acetone to bind with it, confirmed by DFT calculations in Refs. [34,57]. CuWO 4 on the surface provides W 6+ sites in CuWO 4 that act as Lewis acidic sites, attracting acetone carbonyl oxygen [59,60]. Additionally, the higher electrical conductivity of CuWO 4 amplifies resistance changes during the acetone exposure [61]. 4. Conclusion This work demonstrates that sequentially sputter-deposited CuO/ WO 3 nanocomposites, specifically in 20 WO ‒ NPs configuration, are advantageous for effective and selective detection of acetone. By optimizing the deposition order and the mixing portion of CuO and WO 3 , heterostructures were engineered showing improved sensing response (S =23), rapid response, selectivity (even for low concentrations), and most importantly, humidity tolerance (up to 90 % RH) compared to pure WO 3 which make it suitable for breath detection as human breath possess nearly 90 % of relative humidity along with other VOCs. The humidity resilience is uncommon in MOS-based sensors and arises from ternary phase (CuWO 4 ) formation and synergistic p-n and n-n nano heterojunctions that modulate charge transfer. Characterization confirms stable monoclinic phases and shows thermal stabilization improved crystallinity and interfacial interactions. These results stand noble-metal-free CuO‒WO 3 nanocomposite as a promising, low-cost platform for real-world acetone detection. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work the authors used ChatGPT and DeepSeek in order to improve the grammar and style of the draft text. After using this service, the authors reviewed and edited the content as N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 8 needed and take full responsibility for the content of the publication. CRediT authorship contribution statement Nirmal Kumar: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Akash Kumar: Investigation. Jiˇ rí ˇ Capek: Writing – review & editing, Validation. Elisabetta Comini: Writing – review & editing, Resources, Investigation. Stanislav Haviar: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Methodology, Investigation, Funding acquisition, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments This work was supported by the project Quantum materials for applications in sustainable technologies (QM4ST), funded as project No CZ.02.01.01/00/22_008/0004572 by Programme Johannes Amos Comenius, call Excellent Research. The authors sincerely thank Dr. Radomír ˇ Cerstvý, Department of Physics and NTIS, University of West Bohemia in Pilsen for his invaluable assistance in interpreting the XRD spectra. The authors also thank Dr. Rostislav Medlín, New Technologies Research Centre, University of West Bohemia in Pilsen for TEM imaging. Special thanks belong to Dr. Petr Kuberský, Regional Innovation Centre for Electrical Engineering, Faculty of Electrical Engineering, University of West Bohemia in Pilsen, for his insights and technical help. The authors also thank to Dr. Vardan Galstyan, SENSOR Laboratory, Department of Information Engineering (DII), University of Brescia, Brescia, Italy, for measuring cross-sensitivity. Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.apsadv.2025.100894. Data availability Data will be made available on request. References [1] V. Amiri, H. Roshan, A. Mirzaei, G. Neri, A.I. Ayesh, Nanostructured metal oxidebased acetone gas sensors: a review, Sensors (Switzerland) 20 (2020), https://doi. org/10.3390/s20113096. [2] S.R. Sriram, S.R. Parne, N. Pothukanuri, D.R. Edla, Synthesis and characterization of pure and Cu-doped WO3 thin films for high performance of toxic gas sensing applications, App. Surf. Sci. Adv. 15 (2023), https://doi.org/10.1016/j. apsadv.2023.100411. [3] M. Yin, L. Yu, S. Liu, Synthesis of thickness-controlled cuboid WO3nanosheets and their exposed facets-dependent acetone sensing properties, J. Alloys Compd. 696 (2017) 490–497, https://doi.org/10.1016/j.jallcom.2016.11.149. [4] T. Lin, X. Lv, Z. Hu, A. Xu, C. Feng, Semiconductor metal oxides as chemoresistive sensors for detecting volatile organic compounds, Sensors (Switzerland) 19 (2019), https://doi.org/10.3390/s19020233. [5] X. Chang, S. Xu, S. Liu, N. Wang, S. Sun, X. Zhu, J. Li, O. Ola, Y. Zhu, Highly sensitive acetone sensor based on WO3 nanosheets derived from WS2 nanoparticles with inorganic fullerene-like structures, Sens. Actuators B Chem. 343 (2021), https://doi.org/10.1016/j.snb.2021.130135. [6] Tang S., Xu J.-.C., Lu X.-.Q., Chen W.-.J., Chen H.-.W., Du Z., Yu Z.C., Hong B., Wang X.-.Q. Construction of mesoporous Fe 2 O 3 /Cr 2 O 3 n-p heterojunctions for efficient improvement of low-concentration acetone detection and gas-sensing mechanism n.d. doi:10.1007/s12598. [7] S. Americo, E. Pargoletti, R. Soave, F. Cargnoni, M.I. Trioni, G.L. Chiarello, G. Cerrato, G. Cappelletti, Unveiling the acetone sensing mechanism by WO3 chemiresistors through a joint theory-experiment approach, Electrochim. Acta 371 (2021), https://doi.org/10.1016/j.electacta.2020.137611. [8] A. Moumen, G.C.W. Kumarage, E. Comini, P-type metal oxide semiconductor thin films: synthesis and chemical sensor applications, Sensors 22 (2022), https://doi. org/10.3390/s22041359. [9] T.C. Chang, L.H. Lien, H.Y. Lee, C.T. Lee, Investigation performance of NO gas sensors using WO3 nanorod sensing membranes, App. Surf. Sci. Adv. 29 (2025), https://doi.org/10.1016/j.apsadv.2025.100834. [10] B. Jiang, T. Zhou, L. Zhang, J. Yang, W. Han, Y. Sun, F. Liu, P. Sun, H. Zhang, G. Lu, Separated detection of ethanol and acetone based on SnO2-ZnO gas sensor with improved humidity tolerance, Sens. Actuators B Chem 393 (2023), https://doi.org/ 10.1016/j.snb.2023.134257. [11] N. Kumar, S. Haviar, P. Zeman, Three-layer PdO/CuWO4 /CuO system for hydrogen gas sensing with reduced humidity interference, Nanomaterials 11 (2021), https://doi.org/10.3390/nano11123456. [12] W. Gao, X. Chang, O. Ola, Z. Wang, Q. Liao, X. Zhu, J. Li, Y. Jiang, D. Wang, S. Sun, Humidity-tolerant ammonia gas sensors based on PrOx/In2O3/WO3 heterostructure films, Chem. Eng. J. 509 (2025), https://doi.org/10.1016/j. cej.2025.161482. [13] Y. Wang, S. Zhang, D. Xiao, S. Wang, T. Zhang, X. Yang, S. Heng, M. Sun, CuO/ WO3 hollow microsphere P-N heterojunction sensor for continuous cycle detection of H2S gas, Sens. Actuators B Chem. 374 (2023), https://doi.org/10.1016/j. snb.2022.132823. [14] N.P. Hung, N. Van Duy, C.T. Xuan, D.T. Thanh Le, C.M. Hung, H. Jin, N.D Hoa, Enhanced acetone gas-sensing characteristics of Pd-NiO nanorods/SnO2 nanowires sensors, RSC Adv. 14 (2024) 12438–12448, https://doi.org/10.1039/d4ra01265h. [15] N. Kumar, J. ˇ Capek, S. Haviar, Nanostructured CuWO4/WO3-x films prepared by reactive magnetron sputtering for hydrogen sensing, Int. J. Hydrogen Energy 45 (2020) 18066–18074, https://doi.org/10.1016/j.ijhydene.2020.04.203. [16] J. Houska, S. Haviar, J. Capek, R. Cerstvy, K. Shaji, N. Kumar, P. Zeman, New polymorph γ-CuWO4 inspired by γ-CuMoO4: experimental identification and theoretical verification, Scr. Mater 262 (2025), https://doi.org/10.1016/j. scriptamat.2025.116635. [17] F. Zhan, J. Li, W. Li, Y. Liu, R. Xie, Y. Yang, Y. Li, Q. Chen, In situ formation of CuWO4/WO3 heterojunction plates array films with enhanced photoelectrochemical properties, Int. J. Hydrogen Energy 40 (2015) 6512–6520, https://doi.org/10.1016/j.ijhydene.2015.03.131. [18] A.C. Catto, T. Fiorido, ´ E.L.S. Souza, W. Avansi, J. Andres, K. Aguir, E. Longo, L. S. Cavalcante, L.F. da Silva, Improving the ozone gas-sensing properties of CuWO4 nanoparticles, J. Alloys Compd. 748 (2018) 411–417, https://doi.org/10.1016/j. jallcom.2018.03.104. [19] S. Haviar, J. ˇ Capek, ˇ S. Batkov´ a, N. Kumar, F. Dvoˇ r´ ak, T. Duchoˇ n, M. Fialov´ a, P. Zeman, Hydrogen gas sensing properties of WO3 sputter-deposited thin films enhanced by on-top deposited CuO nanoclusters, Int. J. Hydrogen Energy 43 (2018) 22756–22764, https://doi.org/10.1016/j.ijhydene.2018.10.127. [20] S. Haviar, N. Kumar, ˇ S. Batkov´ a, J. ˇ Capek, Nanostructured materials based on thin films and nanoclusters for hydrogen gas sensing, Proc. West Mark Ed. Assoc. Conf. 56 (2021) 38, https://doi.org/10.3390/proceedings2020056038. [21] K. Shaji, S. Haviar, P. Zeman, ˇ S. Kos, R. ˇ Cerstvý, J. ˇ Capek, Controlled sputter deposition of oxide nanoparticles-based composite thin films, Surf. Coat. Technol. 477 (2024), https://doi.org/10.1016/j.surfcoat.2023.130325. [22] N. Kumar, S. Haviar, J. Rezek, P. Baroch, P. Zeman, Tuning stoichiometry and structure of Pd-WO3−x thin films for hydrogen gas sensing by high-power impulse magnetron sputtering, Materials (Basel) 13 (2020) 1–12, https://doi.org/10.3390/ ma13225101. [23] O. Sisman, D. Zappa, E. Bolli, S. Kaciulis, M. Erich, S. Petrovic, E. Comini, Influence of iron and nitrogen ion beam exposure on the gas sensing properties of CuO nanowires, Sens. Actuators B Chem. 321 (2020), https://doi.org/10.1016/j. snb.2020.128579. [24] G.W.C. Kumarage, S. Jayawardena, D. Zappa, M. Shimomura, E. Comini, Unlocking superior acetone sensitivity with co₃o₄/ZnO nanowire innovations, Sens. Actuators B Chem. 431 (2025), https://doi.org/10.1016/j.snb.2025.137450. [25] R.F. Garcia-Sanchez, T. Ahmido, D. Casimir, S. Baliga, P. Misra, Thermal effects associated with the raman spectroscopy of WO3 gas-sensor materials, J. Phy. Chem. A 117 (2013) 13825–13831, https://doi.org/10.1021/jp408303p. [26] M. Boulova, N. Rosman, P. Bouvier, G. Lucazeau, High-pressure Raman study of microcrystalline WO3 tungsten oxide, J. Phys. Condensed Matter 14 (2002) 5849–5863, https://doi.org/10.1088/0953-8984/14/23/314. [27] K. Shaji, S. Haviar, P. Zeman, M. Proch´ azka, R. ˇ Cerstvý, N. Kumar, J. ˇ Capek, Thermally-induced microstructural evolution in nanoparticle-based CuO, WO3 and CuO–WO3 thin films for hydrogen gas sensing, Appl. Surf. Sci. Adv. 28 (2025), https://doi.org/10.1016/j.apsadv.2025.100768. [28] D.R. Miller, S.A. Akbar, P.A. Morris, Nanoscale metal oxide-based heterojunctions for gas sensing: a review, Sens. Actuators B Chem. 204 (2014) 250–272, https:// doi.org/10.1016/j.snb.2014.07.074. [29] B. Liu, D. Hong, L. Li, T. Su, L. Che, X.L. Yang, N. Xu, L. Yue, W. Zhang, Tartaric acid-assisted synthesis and acetone gas sensing properties of hierarchical WO3 phase junctions, Sens. Actuators B Chem. 441 (2025), https://doi.org/10.1016/j. snb.2025.137965. [30] Q. Wang, X. Cheng, Y. Wang, Y. Yang, Q. Su, J. Li, B. An, Y. Luo, Z. Wu, E. Xie, Sea urchins-like WO3 as a material for resistive acetone gas sensors, Sens. Actuators B Chem. 355 (2022), https://doi.org/10.1016/j.snb.2021.131262. [31] Z. Qiu, Z. Hua, Y. Li, M. Wang, D. Huang, C. Tian, C. Zhang, X. Tian, E. Li, Acetone sensing properties and mechanism of Rh-loaded WO3 nanosheets, Front Chem. 6 (2018), https://doi.org/10.3389/fchem.2018.00385. [32] S. Wei, S. Li, R. Wei, S. Liu, W. Du, Different morphologies of WO3 and their exposed facets-dependent acetone sensing properties, Sens. Actuators B Chem. 329 (2021), https://doi.org/10.1016/j.snb.2020.129188. N. Kumar et al. Applied Surface Science Advances 30 (2025) 100894 9