In recent years, with the proposed concept of sustainable development, air quality and environmental problem has attracted more and more attention. In general, the air quality can be influenced by various harmful gases (e.g., CO, NO, NO₂, SO₂) and organic volatile gases that are not easy to perceive, such as formaldehyde, acetone, benzene and toluene [1–3]. Benzene and its derivatives volatile organic compounds (VOCs) have been widely recognized as strong carcinogens by the World Health Organization. In addition, short-term exposure can cause headache, allergy and dyspnea, while long-term exposure can lead to all kinds of cancer and leukemia. Therefore, it is extremely important to realize a rapid and convenient detection of the benzene series by using gas sensors based on various sensitive materials [4,5].

By contrast, semiconductor metal oxides (SMOs) gas sensors have unique advantages for VOCs monitoring applications because they can realize highly sensitive and stable gas sensing response with a low manufacturing and application cost, which are suitable for miniaturization and integration with other devices [6,7]. However, the comprehensive property of the existed SMOs gas sensors usually endowed weak selectivity due to lacking of efficient regulating of the microstructures and the gas–solid interfacial property for sensing reaction [8–10]. Numerous studies have demonstrated that porous SMOs could provide high specific surface area and abundant active sites for enhancing sensitivity of gas sensors [11–13]. The gas sensing selectivity is directly relevant with the adsorption, catalytic conversion of target gas molecules, and therefore, considerable efforts have been devoted to modifying SMOs framework via loading noble metal nanoparticles [14–16] and/or doping with heteroatoms [17–19]. In contrast, transition metal element doping is a commonly used method in exploiting high-efficient catalysts, energy materials and sensing materials. Introducing additional transition metal atoms into the metal oxide framework can regulate the energy band structure, electronic structure and chemical microenvironments (e.g., surface acidity, polarity) of materials. In addition, it can also increase defect density, active sites and the content of oxygen vacancy [20–22], which is favorable for many applications such as catalysis, and chemical sensing.

Except for metal element doping, non-metallic element doping was also widely studied to tune the materials' properties via either gas–solid interface reaction or wet chemical synthesis [23–25]. Various guest heteroatoms such as non-metallic elements B [26], C [27], N [28], S [29], P [30], have been employed to dope host metal oxides to regulate their ionic conductivity, reduce the recombination rate of carriers and enhance materials performance [31–35]. Appropriate doping with non-metallic elements can increase the conductivity by adjusting the band gap structure, improve the mobility of carriers, and increase the content of defects and oxygen vacancies [36]. Based on these considerations, it is expected that the rational construction of porous structure and heteroatom doped framework is a promising strategy to develop high-performance sensitive materials for gas sensing applications.

Herein, a multi-component co-assembly method was employed to straightforward synthesize mesoporous tungsten oxides (mWO₃) with in-situ S-doped crystalline framework. The obtained S-doped mWO₃ (denoted as S/mWO₃) possess unique mesoporous structure, high specific surface area (40.9–63.8 m²/g) and uniform pore size (~18 nm). Due to the abundant oxygen vacancies and internal defects generated by S-doping effect, the S/mWO₃ materials showed ultrahigh sensitivity toward low-concentration benzene at 250 ℃ (50 ppm vs. 48), which is about five times higher than that in pure mWO₃. In addition, the sensor has excellent cycle stability, rapid response rate (only 10 s) and excellent selectivity toward benzene series (e.g., benzene and toluene). Interestingly, there was a significant response change even when the benzene concentration was reduced to 500 ppb, which exhibited great application potential in ultra-low concentration monitoring, such as industrial safety production. The enhanced gas sensing performance was attributed to the significantly increased defect concentration and oxygen vacancy content, which was caused by the introduction of sulfur and the loss during calcination. At the same time, the reduction of band gap further promotes the carrier migration in the sensing materials and the reaction at the gas–solid sensing interfaces.

In this study, facile multi-component co-assembly strategy was employed to synthesize mesoporous S-doped WO₃ based on PEO₁₁₄-b-PS₃₁₅ (M_n = 37,729 g/mol, PDI = 1.22) as the structure directing agent. In addition, bifunctional small molecule 4-hydroxyphenylthiophenol as the sulfur source, anhydrous tungsten chloride is used as the W source, and acetylacetone is added into the assembly system as the chelating agent to inhibit the excessive hydrolysis of W ion in the co-assembly behavior (Scheme 1). During the assembly process, these organic–inorganic components can cooperate with each other to form spherical composite micelles via weak interactions such as hydrogen bonds and coordination bonds. As the evaporation of tetrahydrofuran (THF), the formed spherical micelles can accumulate into uniform mesoscopic structure, and the obtained organic–inorganic mesoporous composite is further calcined and crystallized in nitrogen atmosphere. It is worth noting that the 4-hydroxyphenylthiophenol plays an important role in induced assembly and doping engineering. It can connect with the hydrophilic PEO segments through hydrogen bond, and the partially hydrolyzed W⁶⁺ complexed by acetylacetone can coordinate with the hydrophilic segments. The soft template is carbonized into amorphous sp²-hybrid carbon to support the framework structure. Finally, the crystallized S/mWO₃ materials were obtained after removing the residual carbon by calcination in air.

By changing the feeding amount of S: WO₃, a series of S-doped WO₃ samples (denoted as S/mWO₃-X, herein X represents different feed mass ratios S: WO₃ = 0, 2, 5, 10 wt%) were obtained. It can be seen that the obtained mWO₃ and S/mWO₃-5 can maintain well mesoporous structure and high porosity (Figs. 1a–e). In the high-power transmission electron microscope, it can see that the pore size is about 20 nm. In addition, according to the selected area electron diffraction (SAED) diagram (inset of Fig. 1e), it shows that the diffraction diagram is composed of polycrystalline diffraction rings, indicating that the pore wall of the material is mainly composed of polycrystalline nanoparticles, which can be also seen in the high-resolution transmission electron microscope (HRTEM) (Fig. 1f). Especially, the pore wall structure of S/mWO₃-5 is basically formed by the accumulation of WO₃ grains, in which the exposed lattice stripes of 0.384 nm and 0.373 nm are corresponding to the (001) and (020) crystal planes of WO₃, respectively. Scanning transmission electron microscope (STEM) can clearly see the mesoporous structure (Fig. 1g), and the element distribution diagram (Figs. 1h–j) can clearly show that all elements are homogeneously distributed in the pore frameworks. Furthermore, other S/mWO₃-X with various doping ratios also show that the mesoporous structure can be effectively maintained (Fig. S1 in Supporting information).

Fourier transform infrared spectroscopy (FTIR) is used to characterize the component information in the process of material synthesis. Taking S/mWO₃-5 as an example, it can be seen that the absorption vibration peak of organic matter can be clearly seen in the characterization of the sample before calcination (Fig. 2a). The peaks at 2917 and 3028 cm⁻¹ are corresponded to the stretching vibration peaks of C—H and -C═H in the soft template respectively. It is found that these two peaks disappear in the characterization of the sample after calcination under air conditions, implying the removal of organic template [37]. The XRD peaks of all S/mWO₃-x samples showed high crystallinity (Fig. 2b), corresponding to the crystalline orthorhombic WO₃ phase (PDF#20-1324) [38]. However, there was no obvious WS₂ peak, and it was mainly because the doping content of sulfur was not high enough. In addition, in the calcination process, a certain content of S was easy to be lost under high temperature, so only a small amount of S could be preserved, and some of the lost S vacancies would exist in the form of defects. It can be seen from the local enlarged picture of XRD (Fig. 2c) that the peak at about 42° shifts to a high angle to a certain extent, indicating that a certain amount of S element is successfully doped into the WO₃ lattice [39]. Furthermore, it is also proved by Raman spectroscopy (Fig. 2d) that after the introduction of S element, the W—O stretching vibration peak of about 807 cm⁻¹ shifts to a certain extent. Compared with pure mWO₃, the W—O stretching vibration peak of S/mWO₃-5 shifts to a lower angle from 807 cm⁻¹ to 805.5 cm⁻¹, which is due to the lattice distortion caused by the doping of S atoms with different radii, and it is consistent with the results of XRD, indicating that S atoms are successfully doped into the lattice of WO₃ [17,19]. Many researchers have proved that sulfur doping can regulate the band gap width of SMOs, reduce the recombination rate of electron holes and improve the catalytic and sensing performance [40–43]. For the change of the band gap width of the material before and after doping, the analysis of the solid-state UV spectrum shows that compared with the undoped mWO₃ (Figs. 2e and f), the absorption spectra of the S/mWO₃-X materials are red shifted, and the corresponding band gap width is reduced from the initial 2.53 eV to 2.33 eV. It indicates that after the introduction of sulfur element, the band gap width is reduced, which is conducive to the migration of carriers. The narrower band gap of S/mWO₃-X is originated from the decreased energy difference between the maximum valence band and the Fermi level, and combined with the lower conduction band [32,40].

The valence states of elements in the materials were analyzed by XPS spectrum. The peaks at 530.3 eV and 531.0 eV of O 1s are corresponding to lattice oxygen and adsorbed oxygen (O⁻ and O²⁻), respectively (Figs. 3a and d). In contrast, the position of O 1s peak of S/mWO₃-5 shifts to the direction of low binding energy (530.1 and 530.8 eV), which is caused by the change of chemical environment around O atoms caused by S atoms. In addition, the distribution of oxygen species obtained before and after doping is obtained (Table S1 in Supporting information). After the introduction of S element, the adsorbed oxygen content increases from 33.5% to 36.8%. High adsorbed oxygen content will promote the surface gas catalytic reaction. W 4f XPS spectrum (Figs. 3b and e) shows that both W⁵⁺ and W⁶⁺ exist in the materials. After the introduction of sulfur, the content of W⁵⁺ increases significantly [38], indicating that more defects are formed after S doping. Analyze the XPS spectrum of S 2p of the tested samples after N₂ and air calcination respectively, it can be seen that compared with the S 2p peak of the samples after N₂ calcination (Figs. 3c and f), the S 2p peak after air calcination is significantly weakened because the S element is easy to be lost at high temperature. Especially the S element on the pore walls, while XPS mainly tests the component information with a thickness of about 10 nm, so it is difficult to detect the S element. The peak of 162 eV corresponds to the peak of S²⁻, while 168 and 170 eV are assigned to SO_x²⁻, because S element can replace O position and combine with W to form a small amount of WS₂. At the same time, some S atoms combine with O atoms in the form of interstitial doping are oxidized, and the valence state increases. Comparing the samples calcined in N₂ and air, it can be seen that after air calcination, a part of S element in the material will be lost and produce defects. The generated defects are easy to adsorb oxygen molecules and the content of adsorbed oxygen on the surface will increase. For the analysis of the mesoporous structure of mWO₃, it can be seen that the obtained mesoporous S/mWO₃-X has high specific surface area (40.9–63.8 m²/g) (Fig. S2 and Table S2 in Supporting information) and uniform pore size (18 nm). Therefore, it can be seen that S doping will not damage the mesoporous structure.

In general, high specific surface area and connectivity will be conducive to gas molecule transmission and gas–solid interface reaction, increase surface active sites for the adsorption of oxygen molecules and target gas molecules, and improve the utilization rate of internal materials. At the same time, the unique Knudsen diffusion of mesoporous materials can increase the contact frequency between gas molecules and materials, which is benefit to the improvement of gas sensing performance [44–46]. During the sensing test, the working temperature has a great influence on the response and selectivity of gas sensing, so it is necessary to determine the optimal working temperature of gas sensing reaction. The temperature test range is 150–350 ℃, and the response to 50 ppm benzene gas increases first and then decreases. This is mainly because the increase of working temperature can provide more energy for surface adsorption/catalytic reaction between targeted gas molecules and sensitive materials, and in addition it can reduce the activation energy of surface reaction. However, the excessive increase of working temperature will accelerate the desorption process of gas molecules, which is not conducive to the adsorption of gas molecules, so it is not conducive to further optimizing the gas sensing performance. After S doping, the working temperature is reduced, mainly by increasing the content of adsorbed oxygen on the surface and reducing the activation energy of surface reaction, so that the optimal working temperature is reduced from 300 ℃ to 250 ℃ (Fig. 4a). In addition, S/mWO₃-5 shows the best sensing response to benzene (50 ppm vs. 48). For different concentrations, the gas sensing response gradually increases with the increase of concentration (Fig. 4b). In terms of S/mWO₃-5, the corresponding response of benzene concentration increased from 9 (2.5 ppm) to 48 (50 ppm), which is much higher than the response value of mWO₃ (2.5 ppm: 3.5, 50 ppm: 13.7). The improvement of gas sensing performance is due to the increase of internal defects and oxygen vacancy content after the addition of S atoms, which promotes the surface catalytic reaction. The response–recovery curve of S/mWO₃-X (X = 0, 2, 5, 10) also demonstrated that these sensors possess high sensitivity (Fig. 4c and Fig. S3 in Supporting information).

In practical application, the detection of low concentration benzene is very important for the safety forewarning. The prepared S/mWO₃-X materials can still maintain excellent gas sensing performance for low concentration benzene (1.5 vs. 500 ppb) (Fig. S4 in Supporting information). In addition, S/mWO₃-5 shows excellent cycle stability toward 12.5 ppm benzene (Fig. 4d). In nine cycle testing, the change of equilibrium resistance is less than 10%, indicating that the materials have good stability. At the same time, the cyclic test resistance of S/mWO₃-5 basically remained unchanged after one week test, indicating that the device could still maintain a relatively stable response value (Fig. S5 in Supporting information). In order to deal with complex environmental gases and identify specific gases timely and accurately, excellent selectivity is required. As shown in Fig. 4f, it exhibited excellent gas sensitivity to benzene and toluene, which could be significantly different from other interfering gases (such as methanol, formaldehyde, ethanol and CO). The response value of S/mWO₃-5 to 50 ppm benzene and toluene is about 4 times higher than that of interfering gases, indicating that it has excellent selectivity toward benzene series. In addition, the speed of response and recovery is also one of the evaluation criteria of gas sensor. The response–recovery curve of S/mWO₃-5 to 12.5 ppm benzene showed a faster response speed (10 s) (Fig. 4e). It can be seen that the response speed increases with the increase of concentration (Fig. S6 in Supporting information), but a longer response time is required at low concentration, mainly because the low reaction frequency at low concentration leads to slow rate of resistance change. With the increase of concentration, the recovery time increases gradually, which can be attributed to more benzene molecules adsorbed on the material surface, and the desorption takes longer time due to the stronger binding capacity between benzene and sensitive materials. In addition, the produced H₂O molecules can form water film on the surface of sensitive materials to restrict the desorption behavior [47], and thus the recovery time is relative longer.

In the sensing process, O₂ molecules are adsorbed on the surface of the materials, and it can obtain electrons from the interior of the S/mWO₃-X materials to form electron depletion layer (Scheme 2). After reacting with benzene molecules, it can release electrons back to the interior of the S/mWO₃-X materials to reduce the resistance ((1), (2), (3), (4)) [48–51].

(1)

(2)

(3)

(4)

It can be seen from XPS spectra that under the conditions of S doping and high-temperature S losing, the internal defects and surface adsorbed oxygen content will increase significantly, which will promote the catalytic reaction between benzene molecules and S/mWO₃-X materials [52]. The improvement of performance is mainly determined by the reaction between the adsorbed oxygen species and the target gas (Eq. 5):

(5)

The logarithm of both sides of Eq. 5 is plotted simultaneously (Fig. 5), in which a in the equation is related to the materials, target gas and working temperature, and the oxygen species state on the surface is related to b. According to the relevant literature reports, the range of b value is usually between 0.25 and 1. When the parameter is closer to 1, the surface oxygen species mainly exist in the form of O⁻, which can promote the surface catalytic reaction. On the contrary, the existence in the form of O²⁻ will be not conducive to the improvement of response value. After S element doping, the value of b gradually increases and reaches the maximum at S/mWO₃-5. It shows that the active oxygen (O⁻) content on the surface of S/mWO₃-5 is the highest, which helps to improve the sensing response, and it is consistent with the experimental results [53–55]. Therefore, after the introduction of S atom, more highly active O⁻ will be obtained on the material surface to promote the surface catalytic reaction, which can improve the gas sensing performance.

In conclusion, a facile and straightforward multicomponent co-assembly method was demonstrated to conveniently synthesize mesoporous S-doped crystalline WO₃ materials by using PEO-b-PS as the template and 4-hydroxyphenylthiophenol as sulfur source. The as-synthesized S/mWO₃ materials have high specific surface area (40.9–63.8 m²/g) and uniform pore size (~18 nm). After introducing a certain amount of S elements, it can generate more oxygen vacancies and internal defects via local lattice distortion of WO₃ crystal, which can provide more reaction sites for enhancing benzene series sensing performance, especially for the lower working temperature (250 ℃). In addition, the designed S/mWO₃-5 sensor exhibited optimal benzene sensing performance, including ultrahigh sensitivity toward 50 ppm benzene (R_a/R_g = 48), rapid response rate (ca. 10 s), excellent selectivity to benzene series and high long-term stability. Especially, it can detect ultralow benzene concentration (500 ppb), which exhibits extensive potential in multi-application fields. This method provides a new idea for the synthesis of non-metallic elements-doped SMOs sensitive materials for smart gas sensor.

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 National Natural Science Foundation of China (Nos. 22125501, U22A20152, 22105043, 52225204, 52173233), Key Basic Research Program of Science and Technology Commission of Shanghai Municipality (No. 20JC1415300), the state key laboratory of Transducer Technology of China (No. SKT2207), and the Fundamental Research Funds for the Central Universities (No. 20720220010).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.108898.