Comparative study of monoclinic and cubic WO₃ nanoplates on NO₂ gas-sensing properties

Abstract

Monoclinic and cubic WO₃ nanoplates were controllably prepared from orthorhombic WO₃·H₂O (o-WO₃·H₂O) nanoplates via a facile calcination method at 200 °C for 2 hours in ambient air using tubular and muffle furnaces, respectively. The o-WO₃·H₂O nanoplates were previously prepared from Na₂WO₄·2H₂O via an acid precipitation method at room temperature. Calcination stimulated the dehydration and phase transformation from hydrated WO₃·H₂O nanoplates to WO₃ nanoplates, and different crystal structures were observed under different air environments. In an open-air environment (tubular furnace), a stable monoclinic WO₃ (m-WO₃) phase was obtained, while in a closed-air environment (muffle furnace), a high-entropy cubic WO₃ (c-WO₃) phase was obtained. The difference in the phase transformation was confirmed using various physicochemical analyses, such as X-ray diffraction, field emission scanning electron microscopy, Brunauer–Emmett–Teller measurement, diffuse reflectance spectroscopy, and Raman scattering spectroscopy. Both m-WO₃ and c-WO₃ exhibited excellent NO₂ gas-sensing performance, with ultra-high sensitivity, exceptional selectivity, and ultra-low theoretical limit of detection, at a mild optimal-working temperature of 150 °C. In particular, chemiresistive sensors based on m-WO₃ and c-WO₃ nanomaterials exhibited responses of 1322 and 780 to 2.5 ppm NO₂ and theoretical limits of detection of 0.10 and 0.05 ppb to NO₂ at 150 °C, respectively. These results imply that the phase transformation of WO₃ nanostructures or even phase junctions could be achieved via a facile calcination process in different controlled environments (in closed or open ambient air) for various designed applications such as gas sensors.

---

1. Introduction

Recently, air pollution has emerged as a significant global concern, especially in developing countries. Air pollution is mainly attributed to the presence of toxic gases, including NH₃, NO₂, H₂S, CO, and volatile organic chemicals, which are mostly emitted as a result of various human activities.^1,2 Among these gases, NO₂ is one of the most active candidates affecting human health. Hence, advancing the research and development of sensors for detecting low concentrations of these gases is of practical importance for monitoring toxic gas emissions and other related applications.

Various metal oxide semiconductor (MOS) nanomaterials, such as SnO₂, ZnO, TiO₂, In₂O₃ and WO₃, have been developed and extensively studied as gas-sensing materials, demonstrating promising applications in gas sensors.^3–7 Among these MOS nanomaterials, tungsten trioxide (WO₃) is one of the most promising candidates for detecting toxic gases owing to its abundance and excellent chemical stability and electrical conductivity.⁸ Tungsten oxide also has high sensitivity to gases including NO₂, H₂, H₂S, NH₃, and CO.^9–13 Notably, WO₃ exhibits outstanding selectivity for NO₂ detection.^14–17 To date, different WO₃ nanostructures, such as nanobelts, nanowires, nanoflakes, nanosheets, and nanofibers, have been fabricated using several methods. Each morphology exhibits good gas-sensing properties for specific target gases.^18–22

The crystal phase of WO₃ nanomaterials is a key factor influencing the gas-sensing performance of the sensor. Different phases of WO₃ nanostructures, such as monoclinic, orthorhombic, hexagonal (h-WO₃), triclinic, and mixed phases, have been reported recently.^23–28 During the synthesis process, various factors, such as the dosing ratio, reaction temperature, reaction duration, and calcination temperature, can influence the crystalline phase, grain size, and crystal morphology, thereby affecting the gas-sensing performance. Liu and coworkers prepared orthorhombic WO₃ (o-WO₃) nanorods and m-WO₃ via a facile hydrothermal synthesis at 180 °C. Results revealed that the m-WO₃-based sensor exhibited a remarkably better response to NO₂ than the o-WO₃-based sensor.²⁹ Chen and coworkers prepared and compared the gas sensing performance of o-WO₃ and m-WO₃ nanosheets, which were synthesized by adjusting the synthesis temperature.³⁰ o-WO₃ nanosheets were formed at temperature below 150 °C, while the transition from orthorhombic to monoclinic occurred at a higher synthesis temperature of 180 °C. The results showed that the m-WO₃ nanosheets exhibited better gas sensitivity than the o-WO₃ nanosheets. L. Zhang and coworkers investigated the gas sensing performance of o-WO₃ and c-/o-WO₃ through a one-step calcination process at temperatures ranging from 500 °C to 1000 °C. The c-/o-WO₃ sample, calcinated at 800 °C, demonstrated a response value of 5.23 to 0.5 ppm of NO₂ and exhibited high selectivity, excellent stability, and reliable repeatability.³¹ S. Wei et al. fabricated h-WO₃ and o-WO₃ via a hydrothermal technique and compared their CO gas-sensing performances. Their findings revealed that variations in the morphology and crystal phase played key roles in the gas-sensing property. The h-WO₃ sensor demonstrated outstanding CO sensing performance at 270 °C and provided a quicker and more enhanced response than that of the o-WO₃ sensor.²⁷

Numerous studies have been published on the gas sensing performances of various crystal phases of WO₃ nanostructures. However, up to now, there have been very few reports on comparative studies of the effects of monoclinic and cubic WO₃ nanostructures on NO₂ gas-sensing applications. In this work, monoclinic and cubic WO₃ nanoplates were selectively synthesized from WO₃·H₂O nanoplates via a facile calcination process in regulated environments using different furnaces. Initially, WO₃·H₂O nanoplates were prepared via a facile and power-saving acid precipitation method at room temperature (RT). The effects of the calcination condition on the morphology and crystal structure of samples were investigated using field emission scanning electron microscopy, X-ray diffraction, Brunauer–Emmett–Teller analysis, and diffuse reflectance spectroscopy. The NO₂ gas sensing performance of the sensors were tested using a chemiresistive configuration. Results revealed that both the monoclinic and cubic WO₃ nanoplates exhibited excellent NO₂ gas-sensing characteristics, in which monoclinic WO₃ nanoplates exhibited a better response than those of cubic WO₃ nanoplates. The gas-sensing mechanism and the difference in gas sensing performance of m-WO₃- and c-WO₃-based gas sensors were also discussed.

2. Experimental

2.1. Synthesis of monoclinic and cubic WO₃ nanoplates

Monoclinic and cubic WO₃ nanoplates were prepared from orthorhombic WO₃·H₂O nanoplates via a facile calcination method at 200 °C for 2 hours, in which WO₃·H₂O nanoplates were previously prepared from Na₂WO₄·2H₂O via an acid precipitation method. The synthesis procedure is briefly presented in Fig. 1. In detail, a total of 8.25 g of Na₂WO₄·2H₂O (Xilong) was first dissolved in 25 mL of distilled water and stirred for 15 minutes to form a clear Na₂WO₄ solution. Subsequently, 20 mL of 65% HNO₃ (Merck) was added, followed by gradual addition of distilled water until the total volume of the precursor suspension reached 80 mL. The suspension was continuously stirred for two hours and then transferred to a Teflon container for acid precipitation at RT. After 24 hours, a yellow slurry was obtained, which was thoroughly rinsed several times with distilled water, and then dried in an oven (Nabertherm oven TR120) at 80 °C for 24 h to produce yellow WO₃·H₂O aggregates, which were further ground into a powder. The obtained yellow powder was calcined separately in a closed muffle furnace (Nabertherm muffle furnaces L 9/11/B410) and an open tubular furnace (Nabertherm tubular furnace R40/250/12) at 200 °C for 2 hours, yielding white yellow samples, labelled as MF200 and TF200, respectively. The obtained samples were subsequently utilized for further investigation and evaluation of their gas-sensing performance.

Fig. 1 Schematic of the synthesis process of cubic and monoclinic WO₃ nanostructures.

2.2. Physicochemical characterizations

The morphological features of the materials were examined using FESEM (Hitachi S4800). The crystal phases of the WO₃ nanostructures were characterized using X-ray diffraction (XRD) using an X'pert Pro (PANalytical) MPD with Cu-Kα radiation, employing a scanning rate of 0.03° per 2 s. The specific surface areas of the samples were assessed using Brunauer–Emmett–Teller (BET) method, employing a MicroActive for Tristar II Plus 2.03 system. BET measurements were performed using nitrogen adsorption–desorption isotherms at 77 K, and the samples were degassed at 200 °C under vacuum for 6 h prior to the BET analysis. The optical properties of the samples were investigated by analyzing the diffuse reflectance spectra obtained with a JASCO V-750 diffuse reflectance spectrometer (DRS), operating at a scan speed of 400 nm min⁻¹, a bandwidth of 2.0 nm, and a data interval of 0.5 nm.

2.3. Fabrication of gas sensors and their gas-sensing measurements

The chemiresistive gas sensors were prepared from calcinated powders via drop-coating method. In detail, 20 mg of white yellow WO₃ powders (TF200 or MF200) was dispersed in 1 mL of dimethylformamide (DMF, Xilong) to obtain a yellow suspension. Subsequently, 0.5 μL of the suspension was drop-coated onto a SiO₂ substrate with interdigitated Pt electrodes on its surface (Fig. S1†). The substrate was then heated at 80 °C in air for 2 minutes. This step was repeated five times. Finally, the sensor was calcined at 200 °C for 2 hours to complete the fabrication process.

The gas-sensing performance of the sensors was assessed through a standard automatic measurement system (Fig. S2†). The resistance of the sensors was measured and recorded with a two-point probe, using a high-precision instrument (Keithley 2602A System SourceMeter) connected to a computer. The concentration of the target gas was controlled by adjusting the volumetric ratio of dried air to the test gas in a standard gas chamber via a mass flow controller system. The sensing response (S) was calculated as R_g/R_a for oxidizing gases and R_a/R_g for reducing gases, where R_a and R_g represent the resistance of the sensor when exposed to air and the target gas, respectively.

3. Results and discussion

3.1. Physicochemical properties of the materials

The XRD patterns of the as-synthesized materials are displayed in Fig. 2. The XRD pattern of the as-prepared WO₃ sample is presented in Fig. 2a, which matches well with the standard ICDD card no. 01-084-0886 of o-WO₃, with the (111) plane being dominant. The XRD analysis of the MF200 sample (Fig. 2b) confirmed that all the diffraction peaks closely align with the standard ICDD card no. 00-041-0905 for c-WO₃, exhibiting a lattice constant of 3.84 Å. Specifically, the peaks observed at diffraction angles of 23.9°, 33.98°, 49.25°, 55.25°, and 61.07° correspond to the (100), (110), (111), (200), and (211) crystal planes of c-WO₃, respectively. All the diffraction peaks of the TF200 sample were consistent with the standard pattern for m-WO₃, as given in the standard ICDD card no. 01-089-4476. The peaks at 2θ values of 23.1°, 23.7°, 24.2°, 34.1°, 49°, 56°, and 62° correspond to the (002), (020), (200), (220), (113), (133), and (202) crystal facets of m-WO₃, respectively (Fig. 2c). Furthermore, the absence of any additional diffraction peaks indicated that the synthesized samples possessed high phase purity. These results suggested a complete phase transformation from o-WO₃·H₂O to m-WO₃ and c-WO₃ through simple calcination in an open tubular furnace and a closed muffle furnace at 200 °C for 2 h, respectively. This difference in phase transformations is attributed to variations in the air environment during calcination, which will be discussed in a later section.

Fig. 2 XRD patterns of the (a) as-prepared WO₃, (b) MF200 and (c) TF200 samples in comparison with standard ICDD cards.

Raman spectroscopy is a valuable technique for analyzing the chemical states and crystal structures at the surface of a material. Fig. 3 shows the Raman spectra of the as-prepared and calcined samples, which further confirm the phase transformation from orthorhombic WO₃·H₂O (as-prepared sample) to cubic WO₃ (MF200) and monoclinic WO₃ (TF200). In detail, the Raman spectrum of the as-prepared material exhibited two characteristic peaks at 643 cm⁻¹ and 947 cm⁻¹, corresponding to the stretching vibrations of the O–W⁶⁺–O and W⁶⁺=O bonds in orthorhombic WO₃·H₂O, respectively.^32–34 In the Raman spectrum of the sample annealed in a muffle furnace, prominent peaks at 266, 676, and 821 cm⁻¹ confirmed the formation of cubic WO₃, and these peaks were assigned to the bending δ(O–W–O) vibration, stretching ν(O–W–O) vibrations in the equator plane and via the axis perpendicular to the equator plane of WO₆ octahedra of the O–W–O bonds in cubic WO₃, respectively (note that WO₆ octahedra is the crystal unit in the WO₃ crystal structure).^35,36 Meanwhile, the Raman spectrum of the sample calcined in a tubular furnace showed distinctive peaks at 273, 717, and 810 cm⁻¹, corresponding to the bending δ(O–W–O), stretching ν(O–W–O) modes of O–W–O vibrations in the equator plane and via the axis perpendicular to the equator plane of WO₆ octahedra in monoclinic WO₃, respectively, confirming the formation of the monoclinic WO₃ phase.^33,37–39 Notably, in the Raman spectra of c-WO₃, a vibration at ∼940 cm⁻¹ was clearly observed, implying the vibration of W⁶⁺=O bond in the MF200 (c-WO₃) sample's surface. Besides, the Raman signals were prominent and distinct, which qualitatively confirmed the high crystallinity of both the as-synthesized and calcined samples.

Fig. 3 Raman spectra of the as-prepared WO₃, MF200, and TF200 samples.

Fig. 4 presents the FESEM images of the samples measured at different magnifications (20 000 and 100 000 times). Results showed that the as-prepared sample has a nanoplate morphology (Fig. 4a and b), which is the preferable morphology of orthorhombic WO₃·H₂O prepared via the acid precipitation method in a highly acidic environment.³² This nanoplate morphology was well preserved after calcination in both the tubular and muffle furnaces, indicating that the calcination process at 200 °C induced a minimal effect on the morphology of samples. In the calcined samples, the nanoplates appeared to be more separated (less aggregated) than in the as-prepared sample, and the calcinated nanoplates were thinner. In detail, in the MF200 sample, the nanoplates exhibited different dimensions, but most of these nanoplates exhibited an average size of about 250 nm × 200 nm × 40 nm (Fig. 4b and c). In contrast, the TF200 sample contained nanoplates with smooth edges and well-separated structures, with the majority of the larger nanoplates exhibiting an average dimension of ∼200 nm × 200 nm × 30 nm (Fig. 4e and f).

Fig. 4 FESEM images of the as-prepared WO₃ (a and b), MF200 (c and d) and TF200 (e and f) samples observed at different magnifications.

Surface area plays a crucial role in gas-sensing performance, as a larger surface area generally enhances the sensing capability. BET analysis was carried out to evaluate the surface area, pore diameter, and pore volume of the samples (Fig. 5). Results implied that all the samples exhibited the characteristics of type IV isotherms, with similar pore-size distribution patterns (insets of Fig. 5). Specifically, the specific surface areas of the as-prepared c-WO₃ and m-WO₃ nanomaterials were 20.8, 25.0, and 40.6 m² g⁻¹, respectively, with corresponding pore volumes of 0.136, 0.146, and 0.178 cm³ g⁻¹ (Fig. 5a–c and Table 1). These results suggested that calcination increased the surface area. Furthermore, the m-WO₃ nanomaterials exhibited a larger surface area and pore volume than the c-WO₃ nanomaterials, which were beneficial for enhanced gas adsorption and gas-sensing property.

Fig. 5 N₂ adsorption/desorption isotherm and pore size distribution (the inset) of the as-prepared WO₃ (a), MF200 (b) and TF200 (c) samples.

Table 1 BET surface area, pore size, and pore volume of the as-prepared and calcined samples

Samples	BET value (m^2 g^−1)	Pore size (nm)	Pore volume (cm^3 g^−1)
As-prepared	20.8	25.0	0.136
MF200	25.0	19.1	0.146
TF200	40.6	13.8	0.178

---

Fig. 6a shows the reflectance spectra of the samples. The as-prepared WO₃ sample effectively reflected visible light in the 550–900 nm range, while MF200 and TF200 exhibited strong reflection over a wider range of 500–900 nm. These results confirmed the phase transformation from o-WO₃ to c-WO₃ and m-WO₃ after calcination. Further analysis using the derived Kubelka–Munk method estimated the optical bandgaps of the as-prepared WO₃, MF200, and TF200 samples to be 2.39 eV, 2.51 eV and 2.69 eV, respectively (Fig. 6b).³⁸ Furthermore, these results aligned well with the XRD results, where the phase transformation corresponded to the change in optical bandgap.⁴⁰

Fig. 6 Comparison of the (a) reflectance spectra and (b) Kubelka–Munk plots of calcined samples with those of the as-prepared sample.

The formation of orthorhombic WO₃·H₂O and the phase transformation from o-WO₃·H₂O to c-WO₃ and m-WO₃ could be explained by the protonation process occurring during the acid precipitation process^32,34 and the dehydration process occurred during the calcination process, respectively.^34,39 In detail, when HNO₃ was introduced into the Na₂WO₄ solution, neutral tungstic acid (H₂WO₄) was first produced, as given in eqn (1):

Na₂WO₄ + 2HNO₃ → 2NaNO₃ + H₂WO₄ (1)

In a highly acidic environment (pH = 1.0, measured by Hanna pH meter model HI2020-02), the high concentration of H⁺ promoted the addition of nucleophilic water molecule (H₂O) to tungstic acid, forming a neutral complex [H₂WO₄(H₂O)₂]⁰ (eqn (2)), which later transformed to a more stable structure of [WO(OH)₄(OH₂)]⁰ (eqn (3)).^32,34

After neutral seed molecules [WO(OH)₄(OH₂)]⁰ were formed, the highly acidic environment further promoted the aggregation of seed molecules through van der Waals forces, leading to the development of a WO₃·H₂O crystalline layer. This layer subsequently transformed into WO₃·H₂O nanoplates via an oxolation process (eqn (4)) during the acid precipitation.

Upon calcination at 200 °C, dehydration was stimulated, which resulted in the phase transition from WO₃·H₂O to cubic WO₃ or monoclinic WO₃ (eqn (5) and (6), respectively).

In the closed furnace (muffle furnace), dehydration was suppressed owing to the closed ambient, which resulted in the formation of high entropy structure of cubic WO₃; the structure is preferably formed at higher temperatures.⁴¹ Alternatively, in an open-air environment, air (especially oxygen) is continuously supplied, endowing the molecules with more freedom to reorganize, resulting in the formation of the stable monoclinic phase.

3.2. Gas-sensing characteristics of the calcined WO₃ nanostructures

The gas-sensing performances were evaluated by monitoring the resistance variation of the sensing material, which depended on the surrounding gas environment and operating temperature, when exposed to the target gases. The response–time curves of the samples were measured in the concentration range of 0.25–2.5 ppm of NO₂ at 50–200 °C, as illustrated in Fig. 7 (the corresponding resistance variation with different NO₂ concentrations at different temperatures are presented in Fig. S3†).

Fig. 7 Gas response changes in the TF200 (a) and MF200 (b) samples towards different concentration of NO₂ gas at 50–200 °C.

The comparative analysis of the sensing response, response–recovery time, and selectivity between the TF200 and MF200 sensors are indicated in Fig. 8. The effect of temperature on the sensing response was evaluated to identify the optimal working temperature. Fig. 8a presents the sensor response to 2.5 ppm NO₂ across the temperature range of 50–200 °C. Results indicated that the sensing response of the TF200 and MF200 sensors initially increased as the working temperature increased from 50–150 °C and then declined when the working temperature was increased further to 200 °C. This result implied that 150 °C is the optimal working temperature for both the sensors. Fig. 8b exhibits the sensor response to different NO₂ concentrations at 150 °C. The TF200 sensor exhibited average response values of 1322, 1061, 366, and 116 for NO₂ concentrations of 2.5, 1.0, 0.5, and 0.25 ppm, respectively. In contrast, the MF200 sensor showed average response values of 780, 511, 183, and 83 for the same NO₂ concentrations. These results implied that the TF200 sensor possessed a higher gas response capability than that of the MF200 sensor under same conditions. This phenomenon can be attributed to the large specific surface area of the TF200 sensor, which increased the number of active sites compared with the MF200 sample.

Fig. 8 (a) Response of sensors to 2.5 ppm NO₂ gas at different temperatures; (b) sensitivity of the sensors to different concentrations of NO₂ at 150 °C; (c) response and recovery times of the sensors to 2.5 ppm NO₂ at 150 °C (c); and (d) selectivity of the sensors to 2.5 ppm NO₂, 100 ppm ethanol, 100 ppm H₂, 100 ppm NH₃ and 100 ppm H₂S at 150 °C (reported values are average values with standard deviation).

The effectiveness and reliability of the sensors were also assessed based on their response/recovery time. The values of the TF200 and MF200 sensors for 2.5 ppm NO₂ at 150 °C are depicted in Fig. 8c. The TF200 sensor exhibited response and recovery times of 239 s and 77 s, whereas the MF200 sensor exhibited response and recovery times of 204 s and 47 s, respectively. Fig. 8d presents the responses of the sensors to 2.5 ppm NO₂ and 100 ppm of different gases, including H₂, ethanol, NH₃ (reducing gas) and H₂S (oxidizing gas) at 150 °C. Results indicated that the response of the sensors to NO₂ is obviously greater than their responses to other gases. This was primarily attributed to the unpaired electrons of NO₂, making it a highly oxidizing and electronegative gas. Accordingly, when it came into contact with WO₃, it readily captured electrons from the surface, leading to a significant increase in the material's resistance. As a result, the sensors demonstrated high selectivity to NO₂ gas.⁴² The selectivity of both m-WO₃ and c-WO₃ to NO₂ gas can be attributed to the acidic W surface sites generated via the dehydration process, which selectivity adsorbed NO₂.^43,44

In addition, the repeatability of the m-WO₃ and c-WO₃ sensors for 1 ppm NO₂ at 150 °C is presented in Fig. 9a and b respectively. The dynamic resistance curves exhibited a consistent behaviour across five consecutive test cycles, demonstrating the excellent repeatability of the WO₃ sensors. This further confirmed their potential as reliable candidates for NO₂ detection.

Fig. 9 Repeatability of the TF200 (a) and MF200 (b) sensors to 1 ppm NO₂ at 150 °C and linear fit curve of the gas response versus gas concentration at 150 °C for the sensors (c).

To evaluate the capability of the sensors, we theoretically estimated its limit of detection using the following equation:⁴⁵

The slope was derived from the first derivative of the linear section of Fig. 9c; noise_rms was the root mean square (rms) deviation calculated from a polynomial fit of 50 data points collected when exposed to the air (Fig. S4†). The estimated theoretical detection limits for NO₂ using the m-WO₃ and c-WO₃ sensors were 0.10 ppb and 0.05 ppb, respectively. These results imply that m-WO₃ and c-WO₃ are highly promising materials as NO₂ gas sensors, as they are capable of monitoring NO₂ gas even at sub-ppb level, which is significantly lower than the threshold limit value of 3 ppm established by the American Industrial Hygiene Association.⁴⁶

The NO₂ gas-sensing mechanism of WO₃ nanostructure-based sensors can be described as follows: in an air environment, oxygen molecules absorb on the surface layer of WO₃ nanostructure, forming oxygen species (such as O₂⁻, O⁻ and O²⁻) at different working temperature.⁴² These species capture free electrons from the material's conduction band, leading to charge depletion.

O_2(gas) → O_2(ads) (8)

O_2(ads) + e⁻ → O_2(ads)⁻ (T < 100 °C) (9)

O_2(ads)⁻ + e⁻ → 2O_(ads)⁻ (100 °C < T < 300 °C) (10)

O_(ads)⁻ + e⁻ → O_(ads)²⁻ (T > 300 °C) (11)

During this process, an electron depletion layer develops on the surface of the WO₃ material, leading to a decrease in the charge carrier concentration and an increase in the resistance. As a strong oxidizing agent, NO₂ not only absorbs on the WO₃ nanostructure's surface but also interacts with the oxygen species, as described by the following equation:⁴⁷

NO_2(gas) + e⁻ → NO_2(ads)⁻ (12)

NO_2(ads)⁻ + 2O_(ads)⁻ + e⁻ → NO_2(ads) + 2O_(ads)²⁻ (13)

Upon exposure to NO₂ gas, more electrons are extracted from the material, further reducing the charge carrier concentration and widening the electron depletion layer, leading to an increased resistance. Conversely, when NO₂ gas is removed, the electrons previously captured by NO₂ molecules return to the conduction band, shrinking the electron depletion layer and decreasing the resistance.

Notably, c-WO₃ has a narrower optical bandgap and more W⁶⁺=O bonding on its surface, which theoretically implies better gas-sensing performance than m-WO₃.^9,44,48 However, the experimental results showed the opposite trend. The better gas-sensing performance of m-WO₃ may be attributed to the dominant contribution of its larger surface area and favorable nanoplate morphology;^46,49 m-WO₃ (TF200) possessed nearly twice the surface area of c-WO₃ (MF200).

Table 2 compares the NO₂ gas-sensing performance of WO₃-based sensors in different phases.^25,50–53 Among the reported NO₂ gas sensors, those based on m-WO₃ and c-WO₃ exhibited excellent sensitivity at 150 °C, demonstrating high response levels and indicating their potential as candidates for NO₂ gas detection.

Table 2 Comparison of NO₂ gas sensing performance of the sensors based on different phases of the WO₃ nanostructure materials reported in recent literature

Materials	Morphology	Working temp. (°C)	NO2 conc. (ppm)	Response (Rg/Ra)	τres/τrec (s)	Ref.
Triclinic	Nanosheets	200	10	228	7/22	25
			5	150	—
Hexagonal	Nanorods	75	10	5.8	80/100	50
Orthorhombic	Nanoplates	100	10	17	50/740	51
Monoclinic	Nanoplates	175	5	60	40/130	52
Hexagonal	Nanoflowers	200	100	2.25	12.9/180	53
Monoclinic	Nanoplates	150	2.5	1322	239/77	This work
Cubic				780	204/47

---

4. Conclusions

Monoclinic and cubic WO₃ nanoplates were controllably prepared from orthorhombic WO₃·H₂O nanoplates via a facile calcination process at 200 °C for 2 hours in a tubular furnace and a muffle furnace, respectively. XRD, BET, Raman, and DRS results showed that the calcination process stimulated dehydration in both the furnaces, which resulted in the phase transformation of hydrated WO₃·H₂O nanoplates into WO₃ nanoplates. However, different atmospheric environments led to distinct crystal structures. Specifically, under open-air condition (tubular furnace), o-WO₃·H₂O was transformed into a stable monoclinic crystal structure, m-WO₃ (TF200), with a surface area of 40.6 m² g⁻¹. In contrast, under closed-air environment (muffle furnace), o-WO₃·H₂O was converted into a high-entropy cubic crystal structure, c-WO₃ (MF200), with a surface area of 25.0 m² g⁻¹. Notably, both m-WO₃ and c-WO₃ demonstrated excellent gas-sensing characteristics, with ultra-high response, exceptional selectivity, and ultra-low theoretical detection limits, at a mild optimal working temperature of 150 °C. At this temperature, the m-WO₃ and c-WO₃ sensors exhibited average responses of 1322 and 780 to 2.5 ppm NO₂ with theoretical detection limits of 0.10 and 0.05 ppb, respectively. The better NO₂ gas-sensing characteristic of m-WO₃ than that of c-WO₃ was mainly attributed to its larger surface area. This work presents a useful method for the controlled synthesis of WO₃ nanostructures with distinct crystal structures, offering promising potential for various sensing applications.

Data availability

The data used to support the findings of this study are available from the corresponding author upon reasonable request.

Author contributions

Dang Trung Do: investigation, methodology, visualization, formal analysis, writing – original draft. Do Y Nhi Nguyen: investigation, formal analysis. Thi Anh Pham: investigation, formal analysis. Cong Tu Nguyen: conceptualization, methodology, formal analysis, writing – review and editing, validation. Van Hieu Nguyen: supervision, validation.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

The authors (Do Y Nhi Nguyen and Cong Tu Nguyen) would like to thank the financial support from the Asahi Glass Foundation and Hanoi University of Science and Technology (HUST) through the project number AGF.2024-04.

References

1. S. W. Lee, H. G. Jung, J. W. Jang, D. Park, D. Lee, I. Kim, Y. Kim, D. Y. Cheong, K. S. Hwang, G. Lee and D. S. Yoon, Sens. Actuators, B, 2021, 345, 130361.
2. Y. Li, M. Dai, J. Bai, Y. Wang, Y. Li, C. Wang, F. Liu, P. Sun, T. Wang and G. Lu, Sens. Actuators, B, 2022, 370, 132398.
3. Y. Ding, B. Du, X. Guo, Y. Dong, M. Zhang, W. Jin, C. Gao, D. Peng and Y. He, Sens. Actuators, B, 2024, 414, 135916.
4. M. Jiao, N. Van Duy, D. D. Trung, N. D. Hoa, N. Van Hieu, K. Hjort and H. Nguyen, J. Electron. Mater., 2018, 47, 785–793.
5. K. Iwata, H. Abe, T. Ma, D. Tadaki, A. Hirano-Iwata, Y. Kimura, S. Suda and M. Niwano, Sens. Actuators, B, 2022, 361, 131732.
6. S. B. Patil, M. A. More, D. Y. Patil, F. I. Ezema and G. E. Patil, Mater. Lett., 2025, 382, 137882.
7. Q. Wang, R. Li, P. Wang, Y. Zhang, Y. Wang, Y. Yang, Z. Wu, B. An, J. Li and E. Xie, Sens. Actuators, B, 2023, 390, 133985.
8. Z. F. Huang, J. Song, L. Pan, X. Zhang, L. Wang and J. J. Zou, Adv. Mater., 2015, 27, 5309–5327.
9. K. K. Devender, B. Palanisamy, A. Jayaram, N. Mani and H. Santhana Krishnan, ACS Appl. Nano Mater., 2024, 8(1), 65–79.
10. R. B. T, P. V. K. Yadav, A. Mondal, K. Ramakrishnan, J. Jarugala, C. Liu and Y. A. K. Reddy, Chemosphere, 2024, 353, 141545.
11. Y. Qin, Y. Zhang, J. Lei and S. Lei, Sens. Actuators, B, 2025, 428, 137239.
12. L. Jian, R. Peng, Y. He, X. Wang and W. Guo, Mater. Lett., 2023, 336, 133897.
13. L. Piliai, T. N. Dinhová, M. Janata, D. Balakin, S. Vallejos, J. Otta, J. Štefková, L. Fišer, P. Fitl, M. Novotný, J. Hubálek, M. Vorochta, I. Matolinová and M. Vrňata, Sens. Actuators, B, 2023, 397, 134682.
14. M. Wang, Y. Wang, X. Li, C. Ge, S. Hussain, G. Liu and G. Qiao, Sens. Actuators, B, 2020, 316, 128050.
15. F. Liu, H. Song, L. Wu, J. Zhao, X. Yao, K. Fu, Z. Jin, J. Liu, F. Wang and Z. Wang, Colloids Surf., A, 2023, 666, 131329.
16. D. Kanchan Kumar, P. Bharathi, J. Archana, M. Navaneethan and S. Harish, Sens. Actuators, B, 2024, 421, 136477.
17. Z. Wan, C. Ge, L. Bai, S. Hussain, G. Liu, G. Qiao and M. Wang, Ceram. Int., 2024, 50, 36900–36907.
18. H. L. Yu, J. Wang, B. Zheng, B. W. Zhang, L. Q. Liu, Y. W. Zhou, C. Zhang and X. L. Xue, Sens. Actuators, A, 2020, 303, 111865.
19. S. Bonam, V. R. Naganaboina, B. Thirupathi and S. G. Singh, J. Mater. Sci.: Mater. Electron., 2024, 35, 1–12.
20. X. Wang, X. Meng and W. Gao, Sens. Actuators, B, 2023, 387, 133790.
21. B. An, Y. Yang, J. Yan, Y. Wang, R. Li, Z. Wu, T. Zhang, R. Han, X. Cheng, Q. Wang and E. Xie, Appl. Surf. Sci., 2025, 692, 162722.
22. S. Cao, C. Zhao, T. Han and L. Peng, Mater. Lett., 2016, 169, 17–20.
23. S. Tripathi, D. Tripathi, R. K. Rawat and P. Chauhan, Mater. Lett., 2023, 335, 133840.
24. W. Zeng, B. Miao, T. Li, H. Zhang, S. Hussain, Y. Li and W. Yu, Thin Solid Films, 2015, 584, 294–299.
25. L. Han, J. Chen, Y. Zhang, Y. Liu, L. Zhang and S. Cao, Mater. Lett., 2018, 210, 8–11.
26. H. H. Afify, S. A. Hassan, M. Obaida, I. Moussa and A. Abouelsayed, Opt. Laser Technol., 2019, 111, 604–611.
27. S. Wei, J. Zhao, B. Hu, K. Wu, W. Du and M. Zhou, Ceram. Int., 2017, 43, 2579–2585.
28. F. Liu, H. Song, L. Wu, J. Zhao, X. Yao, K. Fu, Z. Jin, J. Liu, F. Wang and Z. Wang, Colloids Surf., A, 2023, 666, 131329.
29. Z. Liu, M. Miyauchi, T. Yamazaki and Y. Shen, Sens. Actuators, B, 2009, 140, 514–519.
30. Y. Chen, K. Xue and Z. Wang, Mater. Sci. Semicond. Process., 2025, 188, 109211.
31. L. Zhang, F. Li, Y. Yang, D. Li, H. Yu, X. Dong and T. Wang, Microchem. J., 2024, 205, 111243.
32. N. L. Pham, T. L. A. Luu, H. L. Nguyen and C. T. Nguyen, Mater. Chem. Phys., 2021, 272, 125014.
33. M. F. Daniel, B. Desbat, J. C. Lassegues, B. Gerand and M. Figlarz, J. Solid State Chem., 1987, 67, 235–247.
34. N. L. Pham, T. L. A. Luu, T. T. M. Nguyen, V. T. Pham, H. L. Nguyen and C. T. Nguyen, Commun. Phys., 2022, 32, 307–318.
35. G. N. Kustova, Y. A. Chesalov, L. M. Plyasova, I. Y. Lin and A. I. Nizovskii, Vib. Spectrosc., 2011, 55, 235–240.
36. J. De Ribamar Martins Neto, R. M. Torresi and S. I. Cordoba De Torresi, J. Electroanal. Chem., 2016, 765, 111–117.
37. X. V. Le, T. L. A. Luu, H. L. Nguyen and C. T. Nguyen, Vacuum, 2019, 168, 108861.
38. C. T. Nguyen, T. P. Pham, T. L. A. Luu, X. S. Nguyen, T. T. Nguyen, H. L. Nguyen and D. C. Nguyen, Ceram. Int., 2020, 46, 8711–8718.
39. V. T. Nguyen, H. S. Nguyen, V. T. Pham, T. T. M. Nguyen, T. L. A. Luu, H. L. Nguyen, D. C. Nguyen and C. T. Nguyen, Commun. Phys., 2020, 30, 319–330.
40. M. J. Priya, P. P. Subha, A. Antony, M. K. Jayaraj and K. Rajeev Kumar, Chem. Phys. Lett., 2023, 829, 140749.
41. N. A. Mohd Razali, W. N. Wan Salleh, F. Aziz, L. W. Jye, N. Yusof and A. F. Ismail, J. Cleaner Prod., 2021, 309, 127438.
42. K. Gu, X. Song, Q. Zhang and M. Zhang, Sens. Actuators, B, 2023, 382, 133453.
43. M. Epifani, Sensors, 2022, 22, 2247.
44. R. Li, Q. Wang, Y. Wang, B. An, Y. Yang, Z. Wu, P. Wang, T. Zhang, R. Han and E. Xie, ACS Appl. Mater. Interfaces, 2024, 16, 51738–51747.
45. M. J. Priya, C. S. Saramgi, P. P. Subha, A. Antony, M. K. Jayaraj and K. Rajeev Kumar, Surf. Interfaces, 2024, 48, 104340.
46. Y. Li, Q. Zhou, S. Ding and Z. Wu, Front. Chem., 2021, 9, 1–8.
47. S. S. Sangale, V. V. Jadhav, S. F. Shaikh, P. V. Shinde, B. G. Ghule, S. D. Raut, M. S. Tamboli, A. M. Al-Enizi and R. S. Mane, Mater. Chem. Phys., 2020, 246, 122799.
48. G.-L. Chen, H.-Y. Jiang, Z.-P. Deng, Y.-M. Xu, L.-H. Huo and S. Gao, Sens. Actuators, B, 2024, 415, 135981.
49. M. Epifani, Sensors, 2022, 22(6), 2247.
50. Y. Shen, H. Bi, T. Li, X. Zhong, X. Chen and A. Fan, Appl. Surf. Sci., 2018, 434, 922–931.
51. S. S. Shendage, V. L. Patil, S. A. Vanalakar, S. P. Patil, N. S. Harale, J. L. Bhosale, J. H. Kim and P. S. Patil, Sens. Actuators, B, 2017, 240, 426–433.
52. T. Li, Y. Shen, X. Zhong, S. Zhao, G. Li, B. Cui, D. Wei and K. Wei, J. Alloys Compd., 2020, 818, 152927.
53. G. M. Hingangavkar, Y. H. Navale, T. M. Nimbalkar, R. N. Mulik and V. B. Patil, Sens. Actuators, B, 2022, 371, 132584.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ra01820j

---