Zinc nitrate hexahydrate (Zn(NO₃)₂·6H₂O), stannic chloride pentahydrate (SnCl₄·5H₂O), N,N-dimethylformamide (DMF), polyvinylpyrrolidone (PVP, Mw = 1,300,000) and ethanol were of analytical graded and used directly without further purification. All chemicals were purchased from Chongqing Chuandong Chemical Reagent Co., Ltd (Lu et al., 2018).

In the typical synthesis of ZnO-SnO₂ composite nanofibers, 0.7 g of SnCl₄·5H₂O and 0.6 g Zn(NO₃)₂·6H₂O (the molar ratio was 1:1) were dissolved in 5 ml of mixed solvents of ethanol and DMF (The volume ratio was 1:1) and stirred for 2 h. Then 2 g of PVP was added to the mixture and stirred for 24 h to form a viscous and homogeneous solution at room temperature. During electrospinning, the obtained mixture was delivered to a glass syringe. A voltage of 25 kV was applied between the flat aluminum foil and syringe at an electrode distance of 15 cm as shown in Figure 1 and the flow rate is 0.7 ml/h. Finally, the electrospun nanofibers were transferred to a tube furnace and the specimens were annealed at 600°C for 3 h in air for the removal of PVP. For comparison, the pure SnO₂ nanofibers were also synthesized without adding Zn(NO₃)₂·6H₂O. The schematic illustration of producing electrospun ZnO-SnO₂ composite nanofibers was shown in Figure 1 (Bai et al., 2018).

To investigate the structures of electrospun ZnO-SnO₂ nanofibers, X-ray diffraction (XRD, D/Max-1200X, Rigaku, Japan) analysis was carried out at room temperature using a Rigaku D/Max-1200X diffractometry with Cu-Kα radiation (λ = 1.542 Å) over Bragg angles from 20° to 75° and the scanning speed of 2 deg/min. The morphologies of the electrospun nanofibers were investigated with a field emission scanning electronic microscopy (FESEM, JSM-6700F, JEOL, Japan) and transmission electron microscopy (TEM, JEM-2100, JEOL, Japan) operated at 120 kV. The energy dispersive X-ray spectrum analysis (EDS, Oxford INCA 250, JEOL, Japan) and X-ray photoelectron spectroscopy analysis (XPS, KRATOS X SAM800, Kratos, Kingdom) were tested to analyze the elemental compositions of the sample (Lu et al., 2018).

Gas sensors were fabricated with a side heated structure as shown in Figure 2A and a theoretic diagram of the test circuit was showed in Figure 2B. As shown in Figure 2A, there are two gold electrodes connected with platinum wire at both ends of the ceramic tube. Firstly, the as-prepared powder was mixed with appropriate amount of anhydrous ethanol and deionized water to form a homogeneous paste (Xu et al., 2015). Then the obtained paste was coated onto a prefabricated ceramic tube to form the sensing film and dried at room temperature for 2 h. Next a Ni–Cr heating wire was inserted into the ceramic tube to finish the side heated H₂S gas sensor. Finally, the stability of the sensing materials was improved by putting the sensor on the aging instrument of the side heated sensor at 120°C for 10 days. The sensor response was defined as S = R_a/R_g (Zhang Q. Y. et al., 2017), where R_a and R_g were the resistances of the sensing material measured in air and in atmosphere containing target test gas H₂S, respectively (Zhu et al., 2017). Gas sensing properties of the obtained sensors were performed with the CGS-8 TP intelligent gas sensing analysis system (Chemical gas sensor-8, Beijing Elite Tech Co., Ltd., China). The response time was defined as the time required by the sensor to reach 90% of the final stable resistance when target gas in. The recovery time is the time required to return to 90% of its original baseline resistance when the sensor was exposed in air again (Nan et al., 2017). The sensing measurement were tested under laboratory condition with room temperature 25°C and constant humidity (50% relative humidity).

Figure 3 shows the XRD pattern of the electrospun ZnO-SnO₂ composite nanofibers. It can be seen from Figure 3 that the XRD peaks are in accordance with the hexagonal wurtzite ZnO and tetragonal rutile SnO₂, compared with the standard pattern of JCPDS card No. 36-1451 and No. 41-1445, respectively. There is neither apparent peak shift nor any other phase corresponding SnO, ZnSnO₃ and Zn₂SnO₄, confirming that there are only SnO₂ and ZnO co-exist in the prepared material. The result shows a possibility of developing n–n heterojunction at the interface between ZnO and SnO₂ nanomaterial (Bai et al., 2018). The crystallite sizes of nanoparticels were calculated using Scherrer's equation (D = kλ/βcosθ) (Lu et al., 2018), and the average crystallite sizes of nanofibers were calculated by diffraction peaks (100), (101) for ZnO, and (110), (101), (221) for SnO₂. After calculation, the average crystallite size of ZnO is 20.1 nm, while it is 19.8 nm of SnO₂.

Figure 4 presents the FESEM and TEM images of the as-prepared ZnO-SnO₂ nanofibers after annealing. Figure 4a consists of randomly oriented nanofibers with diameters of 80–150 nm and lengths of 0.5–2 μm. Besides, the ZnO-SnO₂ nanofibers consist of nanoparticles and surface of nanofibers is rough, which can be attributed to thermal decomposition of PVP caused by annealing. More structural information of the ZnO-SnO₂ nanofibers is researched by TEM characterization as shown in Figure 4b, the single fiber is composed of grain-like nanoparticles with around 20 nm in size.

Figure 5 demonstrates the EDS spectrum of the electrospun ZnO-SnO₂ nanofibers. It can be seen that the ZnO–SnO₂ composite nanofibers are composed of Zn, Sn, and O elements and the atomic ratio of Zn and Sn is about 19.63:20.76.

To further analyze the compositions and element valences of the ZnO-SnO₂ nanofibers, XPS tests were investigated. Figure 6A shows the full range XPS survey spectra of the sample. It confirms the presence of Zn, O, Sn from prepared nanomaterial and C element which is due to the carbon contamination. The binding energies were calibrated using C 1 s hydrocarbon peak at 284.54 eV. Figures 6B,C demonstrate the high resolution spectra of the Zn 2p and Sn 3d energy state, respectively. The Zn 2p XPS spectrum (Figure 6B) presents the doublet peaks located at binding energies of 1045.6 eV and 1022.6eV, which corresponds to Zn 2p_1/2 and Zn 2p_3/2, respectively (Li W. Q. et al., 2015). The result indicates that the Zn²⁺ is the dominant species in the prepared material and in good agreement with the reported data for ZnO (Zhao et al., 2015). Figure 6C shows the binding energy of Sn 3d_5/2, Sn 3d_3/2 are 487.6eV and 496.1eV respectively, which are assigned to the highest oxidation state of Sn⁴⁺ for SnO₂ (Hamrouni et al., 2014; Chen et al., 2018). It further confirmed that ZnO and SnO₂ coexist in the samples.

The gas sensor responses of pure SnO₂ and ZnO-SnO₂ nanofibers sensors as a function of temperatures in the range of 100–375°C toward 50 ppm of H₂S gas were tested and shown in Figure 7. The sensing responses of all the prepared sensors increase with increasing temperature and attain a maximum value at 300 and 250°C for pure SnO₂ and ZnO-SnO₂ nanofibers sensor, respectively. With further increase in temperature, the sensing responses begin to decrease because desorption of H₂S is dominated and the amount of the adsorbed gas onto the surface decreases (Zhang et al., 2018). The response of gas sensor based on ZnO-SnO₂ nanofibers for 50 ppm H₂S gas at operating temperature of 250°C is 66.23, while it is 10.49 and 300°C for the pure SnO₂ nanofibers based sensor. The results indicate that ZnO-SnO₂ composite nanofibers can obviously improve the response to H₂S at different working temperatures and reduce the optimal operating temperature.

Figure 8 shows the H₂S gas responses of pure SnO₂ and ZnO-SnO₂ nanofibers based sensors to different concentration of H₂S in the range of 0.5–100 ppm at their optimal operating temperatures measured above. The measured results show that gas responses of the as-prepared gas sensors increase in a good linear relationship from 0.5 to 100 ppm. The linear relationship of the response and gas concentration satisfies linear equation y = 1.1003x+6.90664 for electrospun ZnO-SnO₂ nanofibers gas sensor. The higher response of ZnO-SnO₂ nanofibers can be explained by the formation of n-n heterojunctions at the interface between ZnO and SnO₂. Moreover, the sensor detection limit was defined as the target gas concentration value at which the response is above 3. The response of the ZnO-SnO₂ nanofibers sensor to 0.5 ppm H₂S gas can reach up to 3.45, indicating that the detection limit of the sensor for detecting H₂S gas is as low as sub-ppm level.

The dynamic response and recovery curve of the ZnO-SnO₂ nanofibers sensor for 1, 5, 30, and 50 ppm H₂S gas was performed and shown in Figure 9. The obtained sensing response values are about 7.25, 13.37, 40.31, and 66.23, respectively. The ZnO-SnO₂ nanofibers sensor responds rapidly and could recover to its initial value when it was exposed to air again, implying a satisfying stability and reproducibility of the proposed H₂S gas sensor.

Figures 10A,B illustrate the response-recovery curve of electrospun ZnO-SnO₂ nanofibers sensor and pure SnO₂ nanofibers sensor to 50 ppm of H₂S gas at their optimal operating temperature mentioned above. From the curves, it is observed that the response and recovery time of the ZnO-SnO₂ nanofibers sensor is about 18 and 32 s, respectively, whereas for pure SnO₂ nanofibers sensor the corresponding values is 24 and 38 s, respectively.

Finally, the long-term stability of the fabricated ZnO-SnO₂ nanofibers sensor was measured to 10, 30, 50, and 100 ppm H₂S gas at 250°C for 30 days as shown in Figure 11. The measured results show that the response has little change for 30 days and confirm a good stability of the fabricated electrospun ZnO-SnO₂ nanofibers sensor.

Experimental results of ZnO-SnO₂ nanofibers sensor have been compared with the results reported by the other workers on H₂S sensors and presented in Table 1. It can be seen that electrospun ZnO-SnO₂ nanofibers sensor toward H₂S can reach a relatively higher response at lower temperature and the response time is relatively shorter than other sensors reported previously. The obtained results indicate that the electrospun ZnO-SnO₂ nanofibers sensor is promising for H₂S gas sensing.

ZnO and SnO₂ belong to typical n-type semiconductors, characterized by their high free carrier concentration (Hong et al., 2017; Zhou et al., 2018d). The gas sensing mechanism of ZnO-SnO₂ nanofibers is shown in Figure 12. Due to the sensing mechanism of ZnO-SnO₂ sensor follows the surface controlled type, the gas sensing properties are ascribed to the change of the surface resistance, which controlled by the adsorption and desorption of oxygen on the surface of sensing materials (Wei et al., 2014). When ZnO-SnO₂ nanofibers sensor is exposed to air (Figure 12A), the resistance of gas sensor depends on the amount of chemisorbed oxygen species (O⁻, O2-, and O²⁻). The free oxygen molecules are absorbed on the surface and capture electrons from the conduction band of the ZnO-SnO₂ nanofibers, which causes a depletion layer around the surface and the increasing the resistance (Cheng et al., 2014). When ZnO-SnO₂ nanofibers sensing materials are exposed to H₂S (Figure 12B), the target gas reacts with the adsorbed oxygen and then releases the captured electrons into the conduction band of ZnO-SnO₂ nanifibers to reduce the depletion layer and decrease the resistance.

It is well-known that the chemisorbed oxygen depends on the specific surface area of sensing materials and the operating temperature. ZnO-SnO₂ nanofibers show a big surface area as shown in Figure 4. It means that adsorption capability of ZnO-SnO₂ nanofibers was greatly enhanced (Zhang W. et al., 2017). Moreover, O²⁻ and O⁻ species are regarded as the most oxygen adsorption species at 250°C, and the following H₂S sensing reaction can be considered (Kolhe et al., 2017).

In accordance with the definition about gas response (S = R_a/R_g), the increasing of R_a and decreasing of R_g cause that the response of ZnO-SnO₂ nanofibers is significantly enhanced. Additional electron consumption will occur at the boundaries of ZnO and SnO₂, which further enhances the gas response. The contact of ZnO and SnO₂ provides condition for electrons transfer from SnO₂ to ZnO, which has a higher work function of 5.2 eV compared to SnO₂ (4.9 eV). It results in the formation of an additional depletion layer in the vicinity region between the ZnO and SnO₂, eventually generating a potential barrier for electron flow (Choi et al., 2013). The boundary barrier may decrease when the gas sensor is exposed in H₂S. More electrons of oxygen species transfer to the sensing material by reaction of H₂S with oxygen species, which results the resistance of the sensing material decreases and the response of the sensor increases. However, pure SnO₂ does not provide reaction interface, leading to lower gas response. So the ZnO–SnO₂ composite nanofibers exhibit better sensing properties than the SnO₂ nanofibers.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.