MoCl₅ was purchased from Sigma-Aldrich (China), absolute ethanol was purchased from Sinopharm (China). All reagents were of analytical grade without further purification, and the deionized water was used in all experiments.

MoO₂ nanospheres were synthesized by microwave-assisted solvothermal method. In a typical synthesis procedure, 0.57 g of MoCl₅ was dissolved in 240 ml absolute ethanol with vigorous stirring for 30 min. The MoCl₅ solution was transferred into autoclaves and heated at 200°C for 3 h in a microwave oven (Multiwave PRO, Anton Paar). After cooled to room temperature, the resulting precipitate was collected and washed by centrifuging in deionized water and absolute ethanol, followed by freeze-drying under vacuum for 2 days. The resultant MoO₂ nanospheres were named as MMOs. MMO-180 and MMO-160 were prepared at 180°C and 160°C for 3 h, respectively. For comparison, MoO₂ nanospheres were also synthesized by conventionally solvothermal method, in which the MoCl₅ solution was transferred into autoclaves and heated at 200°C for 24 h in an oven. The resultant MoO₂ nanospheres were named as CMOs.

A scanning electron microscope (SEM, JEOL JSM-7610F) was used to observe the morphologies of MoO₂. X-ray diffraction (XRD) patterns were obtained on a Bruker D8 Advance Xray diffractometer with a Cu Kα radiation of 0.154 nm at a generator voltage of 40 kV. The chemical compositions of MoO₂ were measured using Thermo Fisher ESCALAB 250 XI X-ray photoelectron spectroscopy (XPS). Thermogravimetric analysis (TGA) was performed in air atmosphere with a heating rate of 10°C/min by using a Shimadzu DTG-60 A thermogravimetric analyzer.

The MoO₂ powder was ground and mixed with terpineol at the mass ratio of 1:1 to form a paste. The paste was uniformly coated on the surface of alumina ceramic tube attached with a pair of gold electrodes, which were connected by Pt wires. A Ni-Cr heating wire was inserted into the tube to heat the gas sensor. Before the tests, the sensors were aged at 100°C for 5 days to improve stability. Gas sensing tests were performed on a commercial CGS-8 Gas Sensing Measurement System (Beijing Elite Tech Company Limited) with a test chamber (500 mL in volume). After the sensors’ resistance was stabilized at the target temperature, a calculated volume of gas was injected into the chamber. All tests were conducted at a room temperature of 25 ± 5°C and at 40 ± 5% relative humidity.

The gas response is defined as (R_air-R_gas)/R_air (R_air and R_gas are the sensors’ resistance in air and target gas, respectively). The response time and recovery time is defined as the time taken for the response to reach 90% of total change after testing atmosphere changed.

Figure 1 shows the morphology of MoO₂ nanospheres prepared from microwave-assisted and conventional solvothermal method. The diameter of MMOs is in the range of 400–1,000 nm and the average diameter is about 740 nm. In contrast, CMOs own broader distribution of diameter and larger particle size, which affects the homogeneity and sensitivity of gas sensors. Besides, the process of microwave-assisted solvothermal method takes much less time than conventionally solvothermal method, because of the rapid microwave heating (Wang B. et al., 2017). The heating temperature is vital for the regular morphology of MoO₂ nanospheres during microwave-assisted solvothermal method. As shown in Supplementary Figure 1, MMO-180 and MMO-160, prepared at lower temperature, exhibit irregular morphology, which may affect their sensing properties (Cai et al., 2015). Therefore, MMO is chosen to do further characterization and gas tests.

The crystal structure and chemical composition of MMOs were inspected by XRD and XPS. As shown in Figure 2A, MMO has distinct diffraction peaks at 2θ = 26.03°, 36.852°, 53.512°, and 66.456°, which could be indexed to (−1 1 1), (1 1 1), (−3 1 2), and (2 0 2) planes of monoclinic MoO₂ phase according to the JCPDS 32-0671 (Kim et al., 2009). This suggests MoO₂ was successfully synthesized by microwave-assisted solvothermal method. On the contrary, MMO-180, MMO-160, and CMO have broader and weaker diffraction peaks, applying to the incomplete crystalline phase, which is consisted with the SEM images. To identify the valence of Mo and the chemisorption of O, we characterized the MMOs by XPS. As shown in Figure 2B, XPS spectra of Mo consists of three peaks: two peaks at 231.7 and 235.6 eV present the Mo 3d_5/2 and Mo 3d_3/2 spin-obit components of Mo⁶⁺, respectively; the peak at 233.1 eV is assigned to Mo 3d_5/2 of Mo⁴⁺ (Choi and Thompson, 1996). The appearance of Mo⁶⁺ indicates the slightly oxidation at the surface of MoO₂ by the exposure to air at room temperature, considering no distinguishing peaks of MoO₃ observed at XRD patterns as shown in Figure 2A. Figure 2C shows the XPS spectra of O 1 s, consisted of two peaks at 531 and 531.9 eV, corresponding to lattice and chemisorbed oxygen, respectively. The appearance of chemisorbed oxygen results from the coordination unsaturation of Mo, implying the presence of oxygen vacancy (Yang et al., 2015). The abundant chemisorbed oxygen is beneficial for the sensitivity of MoO₂, since the resistance change is mainly occurred by the reaction between chemisorbed oxygen and target gas (Jian et al., 2020). TGA curves of MMO (Figure 2D) shows a decrease of mass before 300°C, due to the loss of adsorbed water. During this temperature range, there is no obvious increase of mass, which implies MMOs are relative stable at low temperature. The stability of MMOs at low temperature is crucial for the repeatability of gas sensors. At higher temperature, a slight increase of mass occurred, corresponding to the oxidation of MoO₂.

The response to H₂S depends on the physical and chemical absorption of gas, which is strongly affected by the working temperature (Su et al., 2019). Thus, we investigated the optimal working temperature of MMO gas sensor. As shown in Figure 3A, the response of MMO gas sensors to 10 ppm H₂S increased first and then decreased as the working temperature rising. The optimal working temperature is 100°C, which is much lower than that of other metal oxide gas sensors and beneficial for energy saving (Guo W. et al., 2016; Wang et al., 2019; Nguyen et al., 2020). The low working temperature may come from the abundant chemisorbed oxygen and oxygen vacancy in MMO (Shen et al., 2019). Therefore, further tests of sensing properties are all completed at 100°C.

Figure 3B presents the response of MMO to H₂S at different concentrations (1–240 ppm). It can be seen the response increases significantly with increasing concentration of H₂S, and there is good linear relationship (R² = 0.996) between response and the concentration of H₂S in the whole range. Unlike other sensors’ narrow range of linear relationship, sensors of MMO with good linear relationship in a broad range are suitable for detection of H₂S with large change of concentration (Na et al., 2019; Teng et al., 2020). The response and recovery curve of MMO to 40 ppm H₂S at 100°C is shown in Figure 3C with a response time of ∼6 min and recovery time of ∼1 min. The repeatability presented in Figure 3D is also important for gas sensors and other devices (Kong et al., 2021a, b). The curves of response show negligible difference after repeating five cycles of tests to 40 ppm H₂S, which implies good repeatability and stability of MMO. To investigate the selectivity of MMO sensor, it was exposed to various gases, including CH₄, H₂, and CO. As shown in Figure 4A, the sensor exhibits higher response to H₂S than other gases, which could greatly weaken the interference of non-target gases. The response of MMO, MMO-180, MMO-160, and CMO are shown in Figure 4B, in which MMO has the highest response to H₂S.

Table 1 summarizes the sensing performance of different metal oxide to H₂S. Compared to other metal oxide in early work, MMO sensor exhibits lower working temperature and wider concentration range to detect H₂S. Besides, the good repeatability and selectivity makes MMO sensor suitable for detection of H₂S leakage in chemical petrochemical companies.

As a typical n-type semiconductor, the sensing performance of MMO strongly depends on the free electron density (Figure 5). According to the density functional theory (DFT), the adsorption and dissociation of O₂ on MoO₂ surface could occur rapidly at room temperature, due to the high adsorption energy and low dissociation barrier (Zhang Q. et al., 2017). Therefore, when MMO exposed to air, oxygen molecules adsorb onto the surface of MMO and take free electrons from MMO, forming chemisorbed oxygen (O₂^–) and resistant electron-depletion layer (EDL) as the working temperature below 150°C (Franke et al., 2006). This leads to decreased free electron density and increased resistance (Mirzaei et al., 2018). After H₂S was injected into the chamber, H₂S molecules react with O₂^– to form SO₂ and water vapor. In this process, free electrons trapped by O₂^– come back to the MMO, causing the increased free electron density and decreased resistance (Katoch et al., 2015). After exposed to air again, the oxygen molecules will be re-adsorbed and reconstruct the EDL. During the tests, H₂O also participated in the reaction via reacting with hole (h⁺) to render the radical hydroxyl(•OH), which justifies the optimal working temperature is 100°C.

The whole reaction is described below:

As discussed in XPS characterization before, there is abundant chemisorbed oxygen on the surface of MMO, which could react with a large of H₂S molecules without saturation. This causes the good linear relationship in a broad range of MMO sensors to H₂S.