Introduction

Metal oxide semiconductor materials, such as WO3, ZnO, SnO2, and In2O3, have attracted considerable attention due to their unique properties and potential application in gas sensors [1–4]. With regard to WO3, the interaction between its crystal structure and the gas molecules is of an intercalating nature, changes the band structure of WO3, increases its conductivity, and can be used for sensing applications [5]. For SnO2-based gas sensors, the sensing mechanism has been studied for decades, and the “oxygen ionosorption” theory is mostly accepted [6, 7]. This theory can be used to interpret the adsorption and the reaction of analyte gases with oxygen species adsorbed on the surface. Therefore, the gas-sensing process can be considered as a phenomenon in which gas diffusion and reaction are coupled together [8]. Considerable efforts are devoted to improving the sensing performance of metal oxide materials by adjusting microstructural properties, including surface area and mesopore size [9–11]. It is well known that larger surface area will enable more adsorption and reaction active sites, resulting in higher gas response [12–14].

The gas diffusion is an infrequently discussed yet important factor for sensing performance, and the defined mesopore size affects diffusion quite substantially [15]. In the mesoporous metal oxide materials whose pore diameter is in the range from 2 to 50 nm, the mean free path of a gas molecule is determined by the pore walls rather than by adjacent molecules, and thus the gas molecules are more likely to collide with the pore wall and then reacts with adsorbed oxygen species [7, 16]. Sakai described the dependence of gas response on mesopore sizes and channel depths successfully [17]. In this sense, for mesoporous gas sensors, the Knudsen diffusion plays a crucial role on the sensing performance. Compared with mesopores, macropores (diameter >50 nm) have less influence on the diffusion due to the fact that big pores allow the target gas to quickly diffuse in by molecular diffusion.

Conventionally, reducing grain size will on the one hand result in higher surface area, which may lead to high gas response by exposing more active adsorption and reaction site to gas molecules; on the other hand, smaller interparticle pores may be disadvantageous, an effect which may overcompensate the positive effect of grain size [7, 15]. This demonstrates that pore diameter and SSA are correlated in the porous metal oxide materials with potentially opposite impacts on the sensing performance. Recently, Tian has investigated the effects of pore diameter on gas-sensing properties of porous SnO2 materials with different calcining temperatures and attributed the higher sensitivity of porous SnO2 calcined at 800 °C to the larger pore diameter and the resultant effective surface area [18]. But another report has proved that the pore volume showed higher consistency with gas response rather than effective surface area [19]. Moreover, the grain growth has a profound effect in changing organized pore arrangement. When the pore diameter increased at elevated temperature, the diffusion rate changed correspondingly, and thus the sensing performance may be deteriorated [20]. Therefore, it remains a challenge to evaluate the effects of the two characteristics, surface area and mesopore diameter, on gas-sensing properties of metal oxide sensors systematically.

In order to investigate the variation of surface area and mesopore size of SnO2, as well as their influence on the gas-sensing properties simultaneously, SnO2 hollow fibers were prepared by sintering with different times from 2 to 24 h at 600 °C. The detailed characterizations showed that the mesopore diameter increased, whereas the SSA gradually decreased with increasing sintering time. Interestingly, we found that firstly the gas response increased with the mesopore diameter when the sintering time increased from 2 to 11 h, and then it decreased drastically when the sintering time kept on increasing. The competitive influence of surface area and mesopore size on gas-sensing properties of mesoporous SnO2 hollow fibers was discussed qualitatively. Throughout the characterizations and the gas-sensing tests, it should be noted that the sensing performance is deemed to be a result of synergistic influence of surface area and mesopore size.

Experimentation

Materials and synthesis

All the chemical reagents were of analytical grade and utilized as purchased without further purification. The materials were prepared as follows. 0.3 g of SnCl2·2H2O was added into 60 ml of pyridine, followed by stirring at room temperature until the SnCl2·2H2O was dissolved completely. This solution was transferred into a 70-ml stainless steel Teflon-lined autoclave, and 0.25 g of absorbent cotton was dipped into the solution. Then the autoclave was tightened and heated up at 120 °C for 24 h in oven. When the autoclave cooled down to room temperature, the cotton–SnO2 hybrid was took out and washed with copious ethanol and water. Then this hybrid was dried in air at 80 °C for 24 h. To remove the cotton and make the SnO2 crystallize, this hybrid was sintered at 600 °C (heating rate was 3 °C min−1) for 2, 6, 11, 17, and 24 h in a resistance-heated furnace, and the obtained products were denoted as Sn2h, Sn6h, Sn11h, Sn17h, and Sn24h, respectively.

Characterization

The X-ray diffraction (XRD) patterns were recorded on a Philips X’Pert diffractometer in the range from 20° to 80° using Cu Kα1 radiation (λ = 1.5406 Å). The nitrogen adsorption–desorption isotherms were measured by Quantachrome AUTOSORB-1 at 77 K, and the Brunauer–Emmett–Teller (BET) and Barrett–Joyner–Halenda (BJH) methods were used here to calculate the SSA and pore size distribution, respectively. Scanning electron microscopy (SEM) observations were conducted using an FEI Sirion 200 microscopy at an acceleration voltage of 10.0 kV. High-resolution transmission electron microscope (HRTEM) characterization was performed with an FEI Tecnai G220 microscope at 200 kV.

Sensor fabrication and measurement of gas-sensing properties

The sensors were fabricated as follows: the sintered SnO2 fibers and a suitable amount of ethanol were mixed to form a paste. Then the paste was coated on the surface of alumina ceramic tubes, which had two Au electrodes and Pt wires at both ends. The coated ceramic tubes were sintered at 500 °C for 2 h. A small heating wire of Ni–Cr alloy was inserted through a ceramic tube when the ceramic tube was fixed on a 6-stitch substrate. In order to improve the long-term stability of the sensors, the sensor devices were aged at 320 °C for 5 days in air.

The gas-sensing properties were tested in a steel-made chamber through which a controlled atmosphere was allowed to flow. The electrical response of the sensor was measured with an automatic test system, controlled by personal computer [1]. The gas response was defined as the resistance ratio R a/R g, where R a and R g stand for the electrical resistance in dry air and in the tested gas, respectively. Our testing formaldehyde gas with different concentrations can be obtained by mixing dry air and formaldehyde specimen gas (120 ppm in N2 without water vapor) which was purchased from Shanghai Weichuang Standard Gas Analytical Technology Co., Ltd., Shanghai, China.

Results and discussion

Structure and morphology

Figure 1 shows XRD patterns of cotton–Sn hybrid and sintered SnO2 fibers. One peak was observed in the pattern of cotton–SnO2 hybrid, which could be indexed to the characteristic (002) plane of the cotton fiber crystal phase [21]. No peak corresponding to tin-related compound was found, indicating that the state of tin was amorphous. The peak of cotton fibers disappeared in the patterns of sintered specimens, and many new diffraction peaks were observed. All these new peaks agreed well with the tetragonal structure in JCPDS 41-1445 (cassiterite), suggesting that the sintered specimens consisted of tetragonal SnO2. This result indicated good crystallization of SnO2 and complete decomposition of organic cotton [22]. Furthermore, the crystalline size of SnO2 was calculated using Scherrer formula [23]:

$$ {\text{D}} = 0.9\lambda /\beta { \cos }\theta, $$
(i)

where θ is the Bragg angle, β the full width at half maxima, and λ = 1.5406 Å. The crystalline sizes of Sn2h, Sn6h, Sn11h, Sn17h, and Sn24h were 5.8, 6.6, 7.4, 9.7, and 11.3 nm, respectively. This result indicated that continuous sintering resulted in an increase in the grain size of SnO2.

Fig. 1
figure 1

XRD patterns of SnO2–cotton hybrid and sintered SnO2 microfibers

Full size image

Figure 2a, b shows the SEM images of Sn2h and Sn24h, respectively. These two specimens were of the similar appearance which was a replica of cotton fiber. The diameter of the sintered fiber was estimated to be 5–10 μm, which was in line with our previous report [3]. As shown in the inset of Fig. 2a, the hole of SnO2 hollow fibers was about 2 μm. The sintered fibers showed a little shrinkage, compared with the diameter (about 20 μm) of absorbent cotton [21]. To further observe the nanostructure of the fibers, HRTEM was carried out. Figure 2c, d, and e shows the HRTEM images of Sn2h, Sn11h, and Sn24h, respectively. These images demonstrated that the hollow fibers were of hierarchical nanostructure assembled by fine nanocrystals. As shown in Fig. 2c, the crystalline size of Sn2h was small and the pores (as denoted by the dashed line) were encapsulated by SnO2 nanocrystals. When the sintering time was increased to 11 h, the crystalline size increased slightly and the pore diameter became larger than Sn2h, as shown in Fig. 2d. Eventually, the sintering time was increased to 24 h, SnO2 crystals grew up, and the pore diameter kept on increasing, as shown in Fig. 2e. Interplanar distances of 0.34 nm corresponding to tetragonal (110) plane could be observed from the inset of Fig. 2d [24]. To investigate the variation of SnO2 nanocrystals, we selected different HRTEM images of three specimens and measured the grain size of SnO2 nanocrystals. Thus the dependence of SnO2 grain size on sintering time is plotted in Fig. 2f. The grain sizes of Sn2h, Sn11h, and Sn24h were 6.9, 7.3, and 9.9 nm, respectively. The change of SnO2 nanocrystals may put significant influence on the gas-sensing properties.

Fig. 2
figure 2

The SEM images of Sn2h (a) and Sn24h (b) and the HRTEM images of Sn2h (c), Sn11h (d), and Sn24h (e), respectively. The inset of a and d shows the typical hollow structure of SnO2 fibers and fine SnO2 grains with (110) lattice plane, respectively. The grain sizes of Sn2h, Sn11h, and Sn24h collected from different HRTEM images are shown in (f)

Full size image

To further investigate the porous nanostructure of the sintered SnO2 fibers, the nitrogen adsorption–desorption characterization was carried out. As shown in Fig. 3, the isotherms of Sn2h, Sn11h, and Sn24h were all characteristic of a type IV with a type H3 hysteresis loop, indicating that the materials had aggregates of platelike particles forming slitlike pores [18, 25]. As calculated by BET method, the SSA of the five specimens was 103.6, 82.8, 70.8, 68.3, and 59.8 m2 g−1. The SSA showed a decreased trend when sintering time increased gradually (as listed in Table 1). This variation could be ascribed to the growth of SnO2 nanocrystals. The pore diameter distribution of the Sn2h, Sn11h, and Sn24h calculated by BJH method is shown in the inset of Fig. 3. The average pore diameters of the five specimens increased with increasing sintering time, as listed in Table 1. The specimens showed large surface area and mesoporous nanostructure, and thus they were supposed to exhibit remarkable gas-sensing properties.

Fig. 3
figure 3

Nitrogen adsorption–desorption isotherms of Sn2h, Sn11h, and Sn24h and their pore size distribution in the inset panel

Full size image
Table 1 The SSA, average pore diameter, and response and recovery time of the mesoporous SnO2 specimens
Full size table

Gas-sensing properties

The gas-sensing tests were carried out to investigate the relationship between the gas-sensing properties of the specimens and their structural characteristics, SSA and pore diameter. Figure 4 shows the gas response of the sensors based on sintered SnO2 hollow microfibers to 300 ppb formaldehyde at operating temperature ranging from 120 to 320 °C. The result indicated that the response of the sensors were strongly dependent on operating temperature. The volcano-shaped temperature dependence of the response has been theoretically explained by surface reaction and gas diffusion kinetics [16, 26]. It was notable that Sn11h exhibited a higher gas response to formaldehyde compared with any other specimens. The response of Sn11h rose gradually with increasing operating temperature and reached the maximum value (16.3–300 ppb formaldehyde) at 240 °C, and then it decreased drastically when the operating temperature kept on increasing. Sn6h, Sn17h, and Sn24h also showed their highest response at 240 °C. Sn2h showed a lower optimal operating temperature which could be ascribed to its higher surface activity due to its smallest grain size [26].

Fig. 4
figure 4

Dependence of the gas response of the five specimens to 300 ppb formaldehyde on operating temperature

Full size image

The response–recovery curves of the five specimens at 240 °C are shown in Fig. 5. It was found that all the five specimens exhibited good response–recovery characterizations at different concentrations. The response was reversible, and the response values of each specimen increased with increasing gas concentration. Among these specimens, Sn11h exhibited the highest gas response and its response was 6.1, 9.5, 16.3, and 19.6 to 50, 100, 300, and 500 ppb formaldehyde, respectively. It was found that the gas response firstly increased and then decreased with increasing sintering time, while the optimal gas response was shown in SnO2 hollow fibers sintered for 11 h. Additionally, the base line of the sensors’ resistance varied with sintering time. With increasing sintering time, the resistance of SnO2 sensors increased gradually and Sn11h showed the highest resistance. However, the resistance augmented as the sintering time increased from 11 to 24 h. The mechanism for the enhanced gas response of Sn11h and variation of sensors’ resistance were discussed in detail later.

Fig. 5
figure 5

Dynamic response–recovery curves of the five specimens to formaldehyde gas with various concentrations at 240 °C

Full size image

The response and recovery times are key parameters to evaluate the performance of the metal oxide gas sensor. According to the date in the dynamic response–recovery test (as listed in Table 1), the response and recovery time of the five specimens to 300 ppb formaldehyde gas at an operating temperature 240 °C were prepared to investigate the influence of sintering time on the gas-sensing properties. The time taken by the sensor to achieve 90 % of the total resistance change is defined as the response time in the case of injecting the target gas in or the recovery time in the case of keeping the target gas out [27]. Both the response and recovery times of the five specimens, as shown in Table 1, firstly decreased and then increased with increasing sintering time at 240 °C. As plotted in Fig. 6, the sensor of Sn11h exhibited the shortest response (36 s) and recovery time (48 s) among the specimens. Therefore, the surface area of Sn11h cooperated with its relatively wide pore diameter to demonstrate the best gas permeability, which would change the response and recovery time essentially [27, 28]. Although the pore diameter of Sn24h was the widest among the specimens, the low surface reactivity of the larger grain size slowed down the reaction rate of analyte gas reacted on SnO2 surface, so its response and recovery time is longer [29].

Fig. 6
figure 6

Dependence of the response and recovery time (to 300 ppb formaldehyde at 240 °C) on sintering time

Full size image

Discussion

Therefore, the experimental results have shown the sensing performance of SnO2 microfibers assembled by fine nanocrystals sufficiently. First of all, the gas sensors were fabricated by SnO2 hollow microfibers, and the gas molecules could flow through the space between the microfibers due to the loosely stacking, as shown in Fig. 7a. Our previous research has proved that this structure favored the enhancement of gas response of metal oxide materials [1, 18]. Secondly, we will focus on the influence of SnO2 nanocrystals which are elementary units of gas-sensing material. It should be noted that their gas response to formaldehyde did not change with the sintering time monotonously; instead it showed an increase firstly and a decrease later. Herein, we will evaluate these two microstructural parameters (surface area and mesopore size) of SnO2 microfibers on gas-sensing properties and dwell upon the reason for this uncommon variation of sensing performance.

Fig. 7
figure 7

a Schematic of the SnO2 gas sensor and b the dependence of gas response (to 300 ppb formaldehyde at 240 °C), average pore diameter, and SSA on sintering time. c, d, and e shows gas diffusion models of Sn2h, Sn11h, and Sn24h, respectively

Full size image

It is well known that metal oxide sensing materials with high surface area or reduced grain size may exhibit enhanced sensing performance due to the abundant surface-active reaction sites for gas molecule adsorption and surface reaction [30, 31]. As the grain size of SnO2 reduced to a certain value, which was known as “Debye length” (3 nm for SnO2), SnO2 could exhibit an ultra-high gas response because the grain whose size was below 2L d = 6 nm became completely depleted when exposed to air [32]. According to the aforementioned characterizations, Sn2h possessed the highest SSA and smallest grain size, which would provide numerous adsorption and reaction sites for gas-sensing process. Conventionally, these two microstructural properties might imply a relatively high sensing performance.

Surprisingly, in this research an uncommon phenomenon was found that the response of Sn2h was lower than that of Sn6h and Sn11h, and Sn11h exhibited the highest response to ppb-level formaldehyde than other specimens. To investigate the influence of pore diameter and SSA of SnO2 microfibers on their gas response, the dependence of gas response (to 300 ppb formaldehyde at 240 °C), average pore diameter, and SSA on sintering time is depicted and shown in Fig. 7b. This figure demonstrated that the SSA of SnO2 microfibers decreased with increasing sintering time, while the gas response increased gradually. Herein, this result indicated that some other factor dominated their sensing performance in the early stage in which the sintering time is <11 h.

It is worth noting that, besides surface reaction, gas diffusivity also played an important role in influencing the gas-sensing performance. In practical gas-sensing behavior, the individual gas molecule has to reach a certain active region of porous chemiresistive sensing materials through pore or crevice [33]. In this case, the gas molecules diffused into the pores and contributed to a higher gas response because of their concentration gradient in the pores. With regard to these thick-film sensors, whose coating has 50–200 μm depth [34], only the outermost surface of thick-film sensor played a major role in sensing reaction [35]. In this work, the pores in SnO2 microfibers were of typical mesopore, and the transport of gas molecules in mesoporous materials is known as Knudsen diffusion; the diffusion coefficient is given by the equation [36]:

$$ D_{\text{K}} = \frac{4r}{3}\sqrt {\frac{{2{\text{RT}}}}{{\pi {\text{M}}}}} , $$
(ii)

where D K is the Knudsen diffusion coefficient, r is the pore radius (cm), T is the temperature (K), R is the universal gas constant and M is the molecular weight of gas (g mol−1). This formula indicated that the gas molecules can diffuse deeper into the sensing layer through the wide open pores and enable the reaction between formaldehyde and oxygen species(O2 −, O−, and O2−) occurring in deeper layer. Sakai successfully built the relationship between sensitivity and gas concentration, and the formula can be expressed as follows [17]:

$$ S = {1 + }\frac{{aC_{\text{A,s}} }}{{L\sqrt {\frac{k}{{D_{\text{K}} }}} }}{ \tanh }L\sqrt {\frac{k}{{D_{\text{K}} }}}, $$
(iii)

where S is the gas-sensing sensitivity defined as the ratio of R a to R g, C A,s is the target gas concentration outside the material, L is the thickness and k is the rate constant. In this formula, gas sensitivity S increased with decreasing m, namely, the increasing D K may result in an increase in sensitivity.

Here, the effects of the SSA and pore diameter on sensing performance of SnO2 microfibers were discussed in detail. As shown in Fig. 7c, although Sn2h assembled by fine nanocrystals possessed the highest SSA and smallest grain size, its average pore diameter was 2.8 nm, which was too small to let the gas molecules in. According to the Knudsen diffusion formula, it can be inferred that most of the gas molecules were blocked by the small pores, and thus the diffusion rate was vastly limited. The large amount of the inner channels contributed to the highest SSA value, while they were not accessible for gas molecules in practical gas-sensing process. Therefore, the gas response was only ascribed to the adsorption and reaction on the surface SnO2 nanocrystals of the microfibers, resulting in a lower sensing performance. In this early stage (sintering time <11 h), the pore diameter dominated the gas-sensing performance, rather than surface area. When the sintering time was increased to 11 h, the microstructure of SnO2 fibers changed due to grain growth [37]. The SSA decreased from 103.6 to 70.8 m2 g−1, while the pore diameter increased from 2.8 to 8.5 nm. As a result, the gas response increased with increasing pore diameter. As shown in Eq. (ii), larger pore diameter meant a higher Knudsen diffusion rate of gas molecules indicating that the outer and inner surface of materials can interact with detected gas molecules, which will be probably increased [17]. As shown in Fig. 7d, the augmented pore diameter enabled more gas molecules to diffuse in and reacted in deeper layer along the wall. It is worth noting that a larger mesopore diameter accelerated the gas transport in the inner of the fibers, and larger SSA ensured the adsorption–desorption and the chemical reaction on the surface of the inner walls [38]. For the same reason, the resistance of Sn11h was larger than that of Sn2h. Therefore, the moderate mesopore diameter and SSA influenced gas-sensing properties synergistically, and Sn11h exhibited the highest gas response to formaldehyde. In the meantime, the shortest response and recovery time to formaldehyde of Sn11h, as shown in Fig. 6, also benefited from the fast diffusion rate and good reaction activity [19]. When the sintering time was increased from 11 to 24 h, despite the increasing pore diameter (increased from 8.5 to 10.9 nm), the gas response decreased drastically, and the schematic model is shown in Fig. 7e. In the late stage (sintering time more than 11 h), the SSA decreased from 70.8 to 59.8 m2 g−1, which was ascribed to the increasing grain size of SnO2 crystals. As a result, the active adsorption and reaction sites on the surface of SnO2 nanocrystals decreased; in other words, the reaction activity of SnO2 nanocrystals deteriorated [31]. Although the diffusion rate increased to a certain degree, the disadvantages of decreased SSA overwhelmed the positive effects of enlarged mesopore diameter. In addition, the increased response and recovery time was also attributed to the deteriorated surface reaction activity which was closely related to grain size. The smaller grain size provides a more active surface of SnO2 due to the less-coordinated surface Sn atoms [29]; in this sense, the reaction activity of Sn24h was obviously low and its resistance was smaller than Sn11h. Eventually, the gas response of Sn24h decreased and the response and recovery time increased. Thus, the SSA played a dominant role in sensing performance. For mesoporous SnO2 hollow microfibers, surface area and pore diameter were proved to demonstrate their unique effect on the gas-sensing behavior.

To verify the competitive influence of surface area and pore diameter on the sensing performance of mesoporous SnO2 microfibers, we investigated the gas response of the five specimens to hydrogen. Due to its smaller molecular radius and the first-order reaction with oxygen species adsorbed on the surface of metal oxide, hydrogen was suggested to exhibit higher permeability than some other gases [39]. Figure 8 shows their gas response to 10 and 100 ppm hydrogen at 240 °C. This result was consistent with their gas response to formaldehyde, and SnO2 microfibers sintered for 11 h showed the highest response. This result was ascribed to the fact that H2-sensing properties of SnO2-based sensors are determined by the effects of both gas diffusivity and reactivity [40]. It also provided solid evidence that surface area and mesopore size played more significant roles in determining mesoporous gas sensors than molecules’ radius. Therefore, the competitive influence of surface area and mesopore size on gas-sensing properties of SnO2 nanocrystals was revealed.

Fig. 8
figure 8

The response of the five SnO2 specimens to 10 and 100 ppm hydrogen at an operating temperature of 240 °C

Full size image

Conclusions

In summary, this work focused on the gas-sensing properties of mesoporous SnO2 hollow microfibers with different sintering times. Their SSA decreased and mesopore diameter increased gradually with increasing sintering time. Interestingly, the gas response increased with the increasing sintering time from 2 to 11 h and then decreased drastically with further increasing sintering time. The mesopore diameter dominated the sensing performance of mesoporous SnO2 microfibers in the early stage of sintering time, while the surface area dominated in the late stage. This result provided solid evidence of competitive influence of surface area and mesopore diameter on the gas-sensing properties of SnO2. This work proposes a new insight into the research on mesoporous metal oxide materials.