Highly Sensitive and Selective Hydrogen Gas Sensor Using the Mesoporous SnO2 Modified Layers

It is important to improve the sensitivities and selectivities of metal oxide semiconductor (MOS) gas sensors when they are used to monitor the state of hydrogen in aerospace industry and electronic field. In this paper, the ordered mesoporous SnO2 (m-SnO2) powders were prepared by sol-gel method, and the morphology and structure were characterized by X-ray diffraction analysis (XRD), transmission electron microscope (TEM) and Brunauer–Emmett–Teller (BET). The gas sensors were fabricated using m-SnO2 as the modified layers on the surface of commercial SnO2 (c-SnO2) by screen printing technology, and tested for gas sensing towards ethanol, benzene and hydrogen with operating temperatures ranging from 200 °C to 400 °C. Higher sensitivity was achieved by using the modified m-SnO2 layers on the c-SnO2 gas sensor, and it was found that the S(c/m2) sensor exhibited the highest response (Ra/Rg = 22.2) to 1000 ppm hydrogen at 400 °C. In this paper, the mechanism of the sensitivity and selectivity improvement of the gas sensors is also discussed.


Introduction
As one of the most important clean energies, H 2 is widely used in various fields such as fuel cell vehicles, aerospace industry, petrochemical industry, and electronic field [1][2][3]. In consideration of the leakage in the applications of H 2 whose explosive limit is very low, it is essential to monitor the state of hydrogen. Gas sensor is one of the most effective detectors [4,5].
Gas sensing performances of SnO 2 gas sensors can be highly improved by using mesoporous material because of its high specific surface area (SSA) [21]. Japanese researchers fabricated nano-SnO 2 powders coated by mesoporous SnO 2 (m-SnO 2 ), and this kind of SnO 2 films highly increased the responses to H 2 [21]. It was found in the work of Pijolat et al. that thin SiO 2 films deposited on the SnO 2 thick films could improve the selectivity to H 2 [22]. Dhawale et al. synthesized mesoporous ZnO thin films which showed high selectivity towards liquefied petroleum [23]. The aim of the present study was to improve the selectivity and sensitivity simultaneously in hydrogen detection. The uses of such mesoporous materials enable enhancement of the adsorption and reaction of test gas because of the high specific surface area. On the other hand, mesoporous materials are potential molecular sieves for gas sensors to improve their selectivity because of the mesoporous structure.
In this paper, the m-SnO 2 powders were synthesized with a simple and low cost sol-gel method. The sensors were fabricated using commercial SnO 2 (c-SnO 2 ) films as the basic layer and the m-SnO 2 films as the modified layers by screen printing method. Their sensing performances were tested with hydrogen, ethanol and benzene. The relationships between selectivity and the thickness of the films were studied. The present study aims to develop a low cost and highly sensitive and selective hydrogen gas sensor.

Preparation of m-SnO 2 Powders
Employing Na 2 SnO 3 ·4H 2 O as the Sn source, n-cetylpyridinium chloride (C 16 PyCl) as the template and trimethylbenzene (C 6 H 3 (CH 3 ) 3 ) as the surfactant, m-SnO 2 powders were prepared in a similar way to that reported previously [21]. The typical preparation manner was as follows. C 16 PyCl was added to the deionized water at 2.6 wt.%, while Na 2 SnO 3 ·4H 2 O was dissolved in the deionized water at 3.6 wt.%. In this case, Na 2 SnO 3 ·4H 2 O aqueous was mixed with the C 16 PyCl solution at a molar ratio [C 16 PyCl]/[Na 2 SnO 3 ·4H 2 O] = 2.0. Then, trimethylbenzene was added to the solution at a molar ratio [C 6 H 3 (CH 3 ) 3 ]/[Na 2 SnO 3 ·4H 2 O] = 2.5. The pH of the mixture was then adjusted to 10 with an aqueous 35 wt.% HCl solution. The resultant emulsion solution was aged for 2 days at 25 • C. After suction filtration with deionized water and drying, the resultant solid products were treated with a 0.1 M aqueous phosphoric acid (PA) solution for 2 h with magnetic stirrers. Then, it was filtered off, washed and dried at 60 • C for 12 h. Eventually, the solid was calcined at 600 • C for 5 h in air. After calcination, the powders were subjected to mechanical grinding with an agate mortar.
The crystal phases of the m-SnO 2 powders were characterized via X-ray diffraction analysis (XRD, D8 Adwance, Bruker, Karlsruhe, Germany). The specific surface area and pore size distribution were measured by the Brunauer-Emmett-Teller (BET) method using a N 2 adsorption isotherm (BET, ASAP 2020, Micromeritics, Norcross, GA, USA). Morphology of the m-SnO 2 powders was observed by a transmission electron microscope (TEM, JEM2100F STEM/EDS, JEOL, Tokyo, Japan) and the morphology of the commercial SnO 2 (c-SnO 2 ) powders was observed by a scanning electron microscope (SEM, JSM-IT300, JEOL, Tokyo, Japan).

Fabrication of SnO 2 Sensors
Pastes of the c-SnO 2 powders and the as-prepared m-SnO 2 powders were applied on a substrate (30 mm × 6 mm × 0.625 mm), on which interdigitated Pt electrodes had been printed with mechanically automated screen printing technology, as shown in Figure 1.
The thick film gas sensors were fabricated using screen printing technology. For the first layer, the c-SnO 2 powders were mixed with the printing oil (YY-1010, Wuhan Huachuang Ruike Tech. Co. LTD, Wuhan, China) at the mass ratio of 1:1 as the paste. Furthermore, to improve the stability of the gas sensors, the frit of PbO, B 2 O 3 , and SiO 2 (mass ratio [PbO]/[B 2 O 3 ]/[SiO 2 ] = 45/35/20) was added into the c-SnO 2 powders at the level of 2 wt.%. The substrates were treated with drying at room temperature for 10 min and 50 • C for 1 h when the pastes were printed on them. For modified layer, the paste was mixed with the m-SnO 2 powders and the printing oil at the same mass ratio of 1:1. To prepare more modified layers, simply repeat the printing step above. Eventually, the gas sensors were dried at 50 • C for 1 h and calcined at 650 • C for 2 h. The different fabricated gas sensors are listed in Table 1.
were dried at 50 °C for 1 h and calcined at 650 °C for 2 h. The different fabricated gas sensors are listed in Table 1.   The surface morphology of the prepared gas sensors was observed by a scanning electron microscope (SEM, Zeiss Utral Plus, Cari Zeiss AG, Jena, Germany). The cross-sections of the different SnO2 films were observed by a scanning electron microscope (SEM, S-4800, HITACHI, Tokyo, Japan).

Measurement of Sensing Performance
The gas sensors were measured by a commercial SD-101 gas sensing performance testing device (Wuhan Huachuang Ruike Tech. Co. LTD, Wuhan, China) which can be used with four gas sensors to test their gas sensing performance simultaneously ( Figure 2). The operating temperature can be controlled via adjusting the power of the heater coil by a microprocessor. The operating temperature of the gas sensors is in the range of room temperature to 450 °C .   The surface morphology of the prepared gas sensors was observed by a scanning electron microscope (SEM, Zeiss Utral Plus, Cari Zeiss AG, Jena, Germany). The cross-sections of the different SnO 2 films were observed by a scanning electron microscope (SEM, S-4800, HITACHI, Tokyo, Japan).

Measurement of Sensing Performance
The gas sensors were measured by a commercial SD-101 gas sensing performance testing device (Wuhan Huachuang Ruike Tech. Co. LTD, Wuhan, China) which can be used with four gas sensors to test their gas sensing performance simultaneously ( Figure 2). The operating temperature can be controlled via adjusting the power of the heater coil by a microprocessor. The operating temperature of the gas sensors is in the range of room temperature to 450 • C. were dried at 50 °C for 1 h and calcined at 650 °C for 2 h. The different fabricated gas sensors are listed in Table 1.   The surface morphology of the prepared gas sensors was observed by a scanning electron microscope (SEM, Zeiss Utral Plus, Cari Zeiss AG, Jena, Germany). The cross-sections of the different SnO2 films were observed by a scanning electron microscope (SEM, S-4800, HITACHI, Tokyo, Japan).

Measurement of Sensing Performance
The gas sensors were measured by a commercial SD-101 gas sensing performance testing device (Wuhan Huachuang Ruike Tech. Co. LTD, Wuhan, China) which can be used with four gas sensors to test their gas sensing performance simultaneously ( Figure 2). The operating temperature can be controlled via adjusting the power of the heater coil by a microprocessor. The operating temperature of the gas sensors is in the range of room temperature to 450 °C .  The prepared gas sensors were measured to sense 1000 ppm H 2 with dynamic method and 10 ppm ethanol and benzene with static method at the temperature of 200 • C, 250 • C, 300 • C, 350 • C and 400 • C. In the process of dynamic measurement, the SD-101 gas sensing performance testing device was placed in a cylinder of 60 mm in diameter, which is made of polymethyl methacrylate (PMMA). The testing gas flowchart is shown in Figure 3. The prepared gas sensors were measured to sense 1000 ppm H2 with dynamic method and 10 ppm ethanol and benzene with static method at the temperature of 200 °C , 250 °C , 300 °C , 350 °C and 400 °C . In the process of dynamic measurement, the SD-101 gas sensing performance testing device was placed in a cylinder of 60 mm in diameter, which is made of polymethyl methacrylate (PMMA). The testing gas flowchart is shown in Figure 3. The synthetic air, whose flow rate was set as 250 mL/min, consisted of N2 and O2 at the volume ratio of 4:1. To match with the synthetic air, the volume ratio of the 1000 ppm H2 in N2 and O2 was also set as 4:1 with the flow rate of 200 mL/min and 50 mL/min, respectively. During the testing process, the synthetic air was replenished by adjusting the four-way valve. The four-way valve is first turned to let the hydrogen in when the response was stabilized. When the response was stabilized, the four-way valve is turned to lead the synthetic air to go through the cylinder until the sensors recover from the hydrogen. The response transients of the gas sensors to 1000 ppm H2 at 400 °C is shown in Figure 4. It is obvious that all the gas sensors exhibit stable and quick response. In the process of dynamic measurement, the SD-101 gas sensing performance testing device was placed in a cubic evaporated cavity for 50 L. During the testing process, the corresponding quantities of the organic solution (ethanol and benzene) were injected by a micro-injector on a heating panel in the evaporated cavity, when the gas sensors responses to air stabilized. When the response to the test gas stabilized, the cubic testing cavity was opened for recovery. The response is defined as Ra/Rg, where Ra and Rg are the sensor resistances in air and in the test gas, respectively. The response time is generally defined as the time necessary for achieving a The synthetic air, whose flow rate was set as 250 mL/min, consisted of N 2 and O 2 at the volume ratio of 4:1. To match with the synthetic air, the volume ratio of the 1000 ppm H 2 in N 2 and O 2 was also set as 4:1 with the flow rate of 200 mL/min and 50 mL/min, respectively. During the testing process, the synthetic air was replenished by adjusting the four-way valve. The four-way valve is first turned to let the hydrogen in when the response was stabilized. When the response was stabilized, the four-way valve is turned to lead the synthetic air to go through the cylinder until the sensors recover from the hydrogen. The response transients of the gas sensors to 1000 ppm H 2 at 400 • C is shown in Figure 4. It is obvious that all the gas sensors exhibit stable and quick response. In the process of dynamic measurement, the SD-101 gas sensing performance testing device was placed in a cubic evaporated cavity for 50 L. During the testing process, the corresponding quantities of the organic solution (ethanol and benzene) were injected by a micro-injector on a heating panel in the evaporated cavity, when the gas sensors responses to air stabilized. When the response to the test gas stabilized, the cubic testing cavity was opened for recovery. The prepared gas sensors were measured to sense 1000 ppm H2 with dynamic method and 10 ppm ethanol and benzene with static method at the temperature of 200 °C , 250 °C , 300 °C , 350 °C and 400 °C . In the process of dynamic measurement, the SD-101 gas sensing performance testing device was placed in a cylinder of 60 mm in diameter, which is made of polymethyl methacrylate (PMMA). The testing gas flowchart is shown in Figure 3. The synthetic air, whose flow rate was set as 250 mL/min, consisted of N2 and O2 at the volume ratio of 4:1. To match with the synthetic air, the volume ratio of the 1000 ppm H2 in N2 and O2 was also set as 4:1 with the flow rate of 200 mL/min and 50 mL/min, respectively. During the testing process, the synthetic air was replenished by adjusting the four-way valve. The four-way valve is first turned to let the hydrogen in when the response was stabilized. When the response was stabilized, the four-way valve is turned to lead the synthetic air to go through the cylinder until the sensors recover from the hydrogen. The response transients of the gas sensors to 1000 ppm H2 at 400 °C is shown in Figure 4. It is obvious that all the gas sensors exhibit stable and quick response. In the process of dynamic measurement, the SD-101 gas sensing performance testing device was placed in a cubic evaporated cavity for 50 L. During the testing process, the corresponding quantities of the organic solution (ethanol and benzene) were injected by a micro-injector on a heating panel in the evaporated cavity, when the gas sensors responses to air stabilized. When the response to the test gas stabilized, the cubic testing cavity was opened for recovery. The response is defined as Ra/Rg, where Ra and Rg are the sensor resistances in air and in the test gas, respectively. The response time is generally defined as the time necessary for achieving a 90% resistance change to the steady-state value. The recovery time is defined as the time for sensor resistance to reach 90% of air resistance. 90% resistance change to the steady-state value. The recovery time is defined as the time for sensor resistance to reach 90% of air resistance. Figure 5 shows XRD patterns of the c-SnO2 powders (Figure 5a) and the m-SnO2 powders (Figure 5b). The c-SnO2 powders have peaks corresponding to the SnO2 crystalline phase (PDF 41-1445). This implies that the c-SnO2 powders are well-crystallized, and have a tetragonal SnO2 phase. The crystallite size of the c-SnO2, calculated by Scherrer's equation (Jade), is about 65.5 nm. It is also confirmed by the SEM image ( Figure 6). The XRD pattern of the m-SnO2 powders (Figure 5b) shows that they have some main peaks corresponding to SnO2 crystalline phase. It reveals that the prepared m-SnO2 powders have low crystallinity. In addition, the ordered mesoporous structure is confirmed clearly by the TEM observation of the m-SnO2 powders in Figure 7. The pore size distribution and the specific surface area of the m-SnO2 powders are shown in Figure 8. It is clear that the m-SnO2 powders show a large SSA of 262.30 m 2 /g with a small pore size of 2.6 nm.     Figure 9 shows the SEM images of the surface morphology of the gas sensors. It was found that the basic layers of the S(c) sensor were dense and have flat surfaces (see Figure 9a). Moreover, the    Figure 9 shows the SEM images of the surface morphology of the gas sensors. It was found that the basic layers of the S(c) sensor were dense and have flat surfaces (see Figure 9a). Moreover, the    Figure 9 shows the SEM images of the surface morphology of the gas sensors. It was found that the basic layers of the S(c) sensor were dense and have flat surfaces (see Figure 9a). Moreover, the  Figure 9 shows the SEM images of the surface morphology of the gas sensors. It was found that the basic layers of the S(c) sensor were dense and have flat surfaces (see Figure 9a). Moreover, the S(c) sensor appeared to have a particle size of approximately dozens of nanometers, despite a small quantity of lager particles. The calcination resulted in some sintered macropores with a size of several hundred nanometers. In contrast, the film of the S(m) sensor show rough and loosened surfaces, as shown in Figure 9b. It is obvious that the S(m) sensor film showed lager particles (100-200 nm) than that of the S(c) sensor film due to the agglomerations of the particles. The agglomerations of the m-SnO 2 were extremely distinct from those of c-SnO 2 . Furthermore, the calcination of the S(m) sensor resulted in lager sintered macropores. The surface morphology of the other sensors is similar to those of the S(m) sensor (see Figure 9c-e) because of the same printing materials and printing process.  The SEM images of the cross-sections of the gas sensors are shown in Figure 10. The cross-sectional morphology of the S(c) sensor shows that the calcined c-SnO2 was more compact with the thickness of about 5 μm (see Figure 10a), but the modified layers of the m-SnO2 showed relatively loosened morphology (see Figure 10b). It is apparent that there is an obvious stratification between the c-SnO2 basic layer and the m-SnO2 modified layer (see Figure 10c). In addition, fabricated with the same materials and printing manner, the m-SnO2 modified layers had no stratification to each other. The thickness of each m-SnO2 modified layer was confirmed with SEM observation to be about 10-15 µ m. Thus, the thickness of the m-SnO2 modified layers of the S(c/m1), S(c/m2) and S(c/m3) sensors were confirmed to be 15 µ m, 31 µ m and 41 µ m, respectively (see Figure 10c-e). The SEM images of the cross-sections of the gas sensors are shown in Figure 10. The cross-sectional morphology of the S(c) sensor shows that the calcined c-SnO 2 was more compact with the thickness of about 5 µm (see Figure 10a), but the modified layers of the m-SnO 2 showed relatively loosened morphology (see Figure 10b). It is apparent that there is an obvious stratification between the c-SnO 2 basic layer and the m-SnO 2 modified layer (see Figure 10c). In addition, fabricated with the same materials and printing manner, the m-SnO 2 modified layers had no stratification to each other. The thickness of each m-SnO 2 modified layer was confirmed with SEM observation to be about 10-15 µm. Thus, the thickness of the m-SnO 2 modified layers of the S(c/m1), S(c/m2) and S(c/m3) sensors were confirmed to be 15 µm, 31 µm and 41 µm, respectively (see Figure 10c-e). The SEM images of the cross-sections of the gas sensors are shown in Figure 10. The cross-sectional morphology of the S(c) sensor shows that the calcined c-SnO2 was more compact with the thickness of about 5 μm (see Figure 10a), but the modified layers of the m-SnO2 showed relatively loosened morphology (see Figure 10b). It is apparent that there is an obvious stratification between the c-SnO2 basic layer and the m-SnO2 modified layer (see Figure 10c). In addition, fabricated with the same materials and printing manner, the m-SnO2 modified layers had no stratification to each other. The thickness of each m-SnO2 modified layer was confirmed with SEM observation to be about 10-15 µ m. Thus, the thickness of the m-SnO2 modified layers of the S(c/m1), S(c/m2) and S(c/m3) sensors were confirmed to be 15 µ m, 31 µ m and 41 µ m, respectively (see Figure 10c-e).   Figure 11 shows the temperature dependence of the resistances of the gas sensors in air. As a semiconductor material, the resistance of SnO 2 shows decrement trend due to the increase of carriers at the condition of thermal excitation, as confirmed in Figure 11. The higher is the operating temperature in which the gas sensors work, the lower is the resistances in air.  Figure 11 shows the temperature dependence of the resistances of the gas sensors in air. As a semiconductor material, the resistance of SnO2 shows decrement trend due to the increase of carriers at the condition of thermal excitation, as confirmed in Figure 11. The higher is the operating temperature in which the gas sensors work, the lower is the resistances in air. The values of the resistance in air of the S(c) sensor were slightly decreased due to the frit. The resistance of the S(m) sensor in air was much higher, which leads to difficult measurement problem in its application. It can be ascribed to the mesoporous structure, which leads to the extreme decrease of conductive path [21]. However, the resistance of the S(m) sensor in air decreased obviously when the operating temperature increased to 400 °C , owing to the condition of thermal excitation [24]. Using the m-SnO2 as the modified layers, the resistances of the (S(c/m1), S(c/m2) and S(c/m3)) sensors changed obviously. Since the resistance of SnO2 semiconductors was affected by thermal excitation, the resistance of the (S(c/m1), S(c/m2) and S(c/m3)) sensors in air decreased remarkably when the operating temperature increased to 300 °C . As for the S(c/m1), S(c/m2) and S(c/m3) sensors, the resistance of the S(c/m2) sensor in air appeared to be the lowest at all the tested operating temperature. The value of the resistance in air of the S(c/m2) sensor was 3.7 × 10 5 Ω at 400 °C . The above results demonstrate that both thermal excitation and adsorption affect the resistance of MOS gas sensors in air. The oxygen adsorbates were considered as the main reason to change the resistance of MOS gas sensors in air. The absorbed O2 on the surface of SnO2 films implies the formation of O − or O 2− , which result in a decrease in the quantity of carrier. Thus, the resistance of the sensors fabricated with the modified m-SnO2 layers in air increased, due to the marked improvement of the adsorption capacity of surface oxygen. However, since the increase of the modified m-SnO2 layers resulted in larger distance for the oxygen adsorbates diffusing to the basic c-SnO2 layer, as well as the diffusion inhibition of mesoporous to oxygen, the S(c/m3) sensor showed lower resistance in air than the S(c/m2) sensor. The values of the resistance in air of the S(c) sensor were slightly decreased due to the frit. The resistance of the S(m) sensor in air was much higher, which leads to difficult measurement problem in its application. It can be ascribed to the mesoporous structure, which leads to the extreme decrease of conductive path [21]. However, the resistance of the S(m) sensor in air decreased obviously when the operating temperature increased to 400 • C, owing to the condition of thermal excitation [24]. Using the m-SnO 2 as the modified layers, the resistances of the (S(c/m1), S(c/m2) and S(c/m3)) sensors changed obviously. Since the resistance of SnO 2 semiconductors was affected by thermal excitation, the resistance of the (S(c/m1), S(c/m2) and S(c/m3)) sensors in air decreased remarkably when the operating temperature increased to 300 • C. As for the S(c/m1), S(c/m2) and S(c/m3) sensors, the resistance of the S(c/m2) sensor in air appeared to be the lowest at all the tested operating temperature. The value of the resistance in air of the S(c/m2) sensor was 3.7 × 10 5 Ω at 400 • C. The above results demonstrate that both thermal excitation and adsorption affect the resistance of MOS gas sensors in air. The oxygen adsorbates were considered as the main reason to change the resistance of MOS gas sensors in air. The absorbed O 2 on the surface of SnO 2 films implies the formation of O − or O 2− , which result in a decrease in the quantity of carrier. Thus, the resistance of the sensors fabricated with the modified m-SnO 2 layers in air increased, due to the marked improvement of the adsorption capacity of surface oxygen. However, since the increase of the modified m-SnO 2 layers resulted in larger distance for the oxygen adsorbates diffusing to the basic c-SnO 2 layer, as well as the diffusion inhibition of mesoporous to oxygen, the S(c/m3) sensor showed lower resistance in air than the S(c/m2) sensor.

Sensing Responses to the Testing Gas
The temperature dependence of the responses to ethanol, benzene and hydrogen are depicted in Figure 12. The response of the S(c) sensor to ethanol at 10 ppm increased slightly with the increasing of operating temperature up to 400 • C, as shown in Figure 12a. However, the responses to ethanol of the S(c) sensor were lower than those of the S(c/m1), S(c/m2) and S(c/m3) sensors, due to the large specific surface area (262.30 m 2 /g) of the m-SnO 2 . In addition, the S(c/m3) sensor showed the largest response (Ra/Rg = 11.4) to ethanol at 300 • C. The responses to benzene of the S(c/m1), S(c/m2) and S(c/m3) sensors showed a similar tendency: the response decreased at the relatively low operating temperature due to a slight effect of thermal diffusion. While the reaction between testing gas and the basic c-SnO 2 was controlled by gas absorption, the response to benzene was improved with the increasing of operating temperature from 300 • C to 400 • C mainly due to the thermal diffusion. The S(c/m3) sensor exhibited the largest response (Ra/Rg = 4.31) to benzene at 200 • C, as shown in Figure 12b. The response to hydrogen increased from 200 • C to 400 • C (Figure 12c), while the S(c/m2) sensor showed the largest response (Ra/Rg = 22.2) to hydrogen at 400 • C. It can be deduced from the hydrogen molecular diffusion that the small molecular dimension of hydrogen benefits the gas diffusion. While gas adsorbing capacity was enhanced, the response to hydrogen was highly improved (see Figure 12c). All of the above results confirmed that the modified layers of the m-SnO 2 contribute to improving the response of the S(c) sensor sufficiently. Moreover, the magnitude of the response enhancement is not directly proportional to the amount of the m-SnO 2 .

Sensing Responses to the Testing Gas
The temperature dependence of the responses to ethanol, benzene and hydrogen are depicted in Figure 12. The response of the S(c) sensor to ethanol at 10 ppm increased slightly with the increasing of operating temperature up to 400 °C , as shown in Figure 12a. However, the responses to ethanol of the S(c) sensor were lower than those of the S(c/m1), S(c/m2) and S(c/m3) sensors, due to the large specific surface area (262.30 m 2 /g) of the m-SnO2. In addition, the S(c/m3) sensor showed the largest response (Ra/Rg = 11.4) to ethanol at 300 °C . The responses to benzene of the S(c/m1), S(c/m2) and S(c/m3) sensors showed a similar tendency: the response decreased at the relatively low operating temperature due to a slight effect of thermal diffusion. While the reaction between testing gas and the basic c-SnO2 was controlled by gas absorption, the response to benzene was improved with the increasing of operating temperature from 300 °C to 400 °C mainly due to the thermal diffusion. The S(c/m3) sensor exhibited the largest response (Ra/Rg = 4.31) to benzene at 200 °C , as shown in Figure 12b. The response to hydrogen increased from 200 °C to 400 °C ( Figure  12c), while the S(c/m2) sensor showed the largest response (Ra/Rg = 22.2) to hydrogen at 400 °C . It can be deduced from the hydrogen molecular diffusion that the small molecular dimension of hydrogen benefits the gas diffusion. While gas adsorbing capacity was enhanced, the response to hydrogen was highly improved (see Figure 12c). All of the above results confirmed that the modified layers of the m-SnO2 contribute to improving the response of the S(c) sensor sufficiently. Moreover, the magnitude of the response enhancement is not directly proportional to the amount of the m-SnO2. To further investigate the effects of the modified m-SnO2 layers, the evolutions of the response versus thickness of the modified films of the gas sensors at 400 °C are depicted in Figure 13. It is clear that the response of the S(c/m1), S(c/m2) and S(c/m3) sensors to ethanol, benzene and hydrogen were all improved to a certain extent in comparison to the S(c) sensor response, which means higher gas sensitivities. Especially, the response of the S(c/m2) sensor to hydrogen appeared To further investigate the effects of the modified m-SnO 2 layers, the evolutions of the response versus thickness of the modified films of the gas sensors at 400 • C are depicted in Figure 13. It is clear that the response of the S(c/m1), S(c/m2) and S(c/m3) sensors to ethanol, benzene and hydrogen were all improved to a certain extent in comparison to the S(c) sensor response, which means higher gas sensitivities. Especially, the response of the S(c/m2) sensor to hydrogen appeared to improve by 11.4 times, compared to benzene and ethanol improvements of 2.03 and 2.18 times, respectively.   Table 2 shows the response and recovery times of the gas sensors to ethanol, benzene and hydrogen. The response and recovery times of some of the gas sensors were difficult to summarize due to the lower response at low operating temperature. The response time of the S(c) sensor to ethanol was markedly short from 350 °C (response time = 115 s) to 400 °C (response time = 54 s), while the values of response time to hydrogen at 350 °C and 400 °C were 80 s and 89 s, respectively. However, the response times of benzene were hard to summarize because of the relatively large molecular dimension (0.65-0.68 nm), which led to the low response to benzene. In the case of the S(c/m1), S(c/m2) and S(c/m3) sensors, the response times increased with the thicker modified m-SnO2 layers for all of the tested gases at 350 °C and 400 °C . In addition, the smallest response time appeared to be 74 s with the S(c/m1) sensors to hydrogen at 400 °C . It can be ascribed to the large specific surface area and the molecular diffusion at high temperature, both of which lead to easy gas diffusion inside the mesopores.

The Response and Recovery Times of the Gas Sensors
All gas sensors showed longer recovery times to ethanol and benzene in comparison to hydrogen from 350 °C to 400 °C . Especially, the S(c/m1), S(c/m2) and S(c/m3) sensors tended to show a longer response time (>500 s) to hydrogen than to ethanol (<467 s) and benzene (<207 s). Table 2. The response and recovery times of the gas sensors to ethanol, benzene and hydrogen.

200
- -286  496  549  --292  519  >600  250  --252  277  180  --281  355  349  300  126  19  197  106  87  14  90  178  182  142  350  115  41  127  289  430  26  271  246  235  226  400  54  56  152  210  295  37  201  467  467  462   Benzene   200  -->600  >600  >600  -->600  >600  >600  250  --360  486  >600  --60  158  >600  300  --110  225  403  --14  12  13  350  --411  422  478  --89  63  78  400  --173  218 Table 2 shows the response and recovery times of the gas sensors to ethanol, benzene and hydrogen. The response and recovery times of some of the gas sensors were difficult to summarize due to the lower response at low operating temperature. The response time of the S(c) sensor to ethanol was markedly short from 350 • C (response time = 115 s) to 400 • C (response time = 54 s), while the values of response time to hydrogen at 350 • C and 400 • C were 80 s and 89 s, respectively. However, the response times of benzene were hard to summarize because of the relatively large molecular dimension (0.65-0.68 nm), which led to the low response to benzene. In the case of the S(c/m1), S(c/m2) and S(c/m3) sensors, the response times increased with the thicker modified m-SnO 2 layers for all of the tested gases at 350 • C and 400 • C. In addition, the smallest response time appeared to be 74 s with the S(c/m1) sensors to hydrogen at 400 • C. It can be ascribed to the large specific surface area and the molecular diffusion at high temperature, both of which lead to easy gas diffusion inside the mesopores.

The Response and Recovery Times of the Gas Sensors
All gas sensors showed longer recovery times to ethanol and benzene in comparison to hydrogen from 350 • C to 400 • C. Especially, the S(c/m1), S(c/m2) and S(c/m3) sensors tended to show a longer response time (>500 s) to hydrogen than to ethanol (<467 s) and benzene (<207 s).

Discussion
Among these, the possible gas sensing mechanism of the gas sensors are shown in Figure 14. It is considered that the adsorption/desorption properties of the mesoporous influenced the gas sensing performances of the gas sensors. Owing to the large specific surface area of m-SnO 2 , which could enhance the adsorption of gas molecules, the S(c/m1) S(c/m2) and S(c/m3) sensors exhibited higher gas (ethanol, benzene, and hydrogen) responses than those of the S(c) sensor. The responses of benzene are lower than those of ethanol and hydrogen because of its weak reducibility and larger size of benzene ring, which is difficult to pass through the m-SnO 2 modified layer. However, the sintered macropores ( Figure 14) among the m-SnO 2 is helpful to adsorb more benzene molecules. This is why the responses of the S(c/m1), S(c/m2) and S(c/m3) sensors to benzene are higher than those of the S(c) sensor to benzene. In contrast, the smaller molecular size of the hydrogen is beneficial to pass through the ordered structure of m-SnO 2 and the sintered macropores. Thus, thickness of the films and the ordered level of mesoporous influenced the gas sensing performance of the gas sensors fabricated with the modified m-SnO 2 layers. In addition, further approaches to control the amount of sintered macropores, the thickness of the films, surface contact of the films and the ordered level of mesoporous would be effective to improve the sensing performance of the gas sensors.

Discussion
Among these, the possible gas sensing mechanism of the gas sensors are shown in Figure 14. It is considered that the adsorption/desorption properties of the mesoporous influenced the gas sensing performances of the gas sensors. Owing to the large specific surface area of m-SnO2, which could enhance the adsorption of gas molecules, the S(c/m1) S(c/m2) and S(c/m3) sensors exhibited higher gas (ethanol, benzene, and hydrogen) responses than those of the S(c) sensor. The responses of benzene are lower than those of ethanol and hydrogen because of its weak reducibility and larger size of benzene ring,which is difficult to pass through the m-SnO2 modified layer. However, the sintered macropores ( Figure 14) among the m-SnO2 is helpful to adsorb more benzene molecules. This is why the responses of the S(c/m1), S(c/m2) and S(c/m3) sensors to benzene are higher than those of the S(c) sensor to benzene. In contrast, the smaller molecular size of the hydrogen is beneficial to pass through the ordered structure of m-SnO2 and the sintered macropores. Thus, thickness of the films and the ordered level of mesoporous influenced the gas sensing performance of the gas sensors fabricated with the modified m-SnO2 layers. In addition, further approaches to control the amount of sintered macropores, the thickness of the films, surface contact of the films and the ordered level of mesoporous would be effective to improve the sensing performance of the gas sensors.  that the 0.25% Pd doped gas sensor response towards 1000 ppm hydrogen at 50 • C is 0.95. The gas sensors showed zero response to ethanol, LPG, NH 3 and acetone [25]. Seftel et al. obtained gas sensing material by combining Pt with SnO 2 or In 2 O 3 based on SBA-15. The response of the gas sensor based on Pt/SnO 2 /SBA-15 is about 1.4 to 1000 ppm hydrogen at 350 • C [26]. Although the selectivity of the sensors was improved by doping, the responses of the sensors to hydrogen are no more than 2, which limits the applications of the sensors in hydrogen measurement.
We can find a large number of examples of sensitivities improvement of gas sensors to hydrogen when mesoporous structures are employed. Shen et al. reported that the influence of the different morphology of SnO 2 nanomaterials on hydrogen sensing properties. They obtained the response of about 2.1 to 1000 ppm hydrogen at 250 • C for nanofilms [27]. Yeow et al. reported the gas sensors based on SnO 2 nanospheres with various degrees porosity. The reference (SSA SnO 2 = 101.4 m 2 /g) gas sensor showed the largest response: 5.2 to 500 ppm hydrogen at 350 • C [28]. Zhao et al. prepared ordered mesoporous SnO 2 and mesoporous Pd/SnO 2 via nanocasting method using the hexagonal mesoporous SBA-15 as template. The maximum response of the sensor based on the ordered mesoporous SnO 2 is 16.4 to 1000 ppm hydrogen at 300 • C [29]. Hayashi et al. prepared SnO 2 gas sensors based on various m-SnO 2 powders from two kinds of combination of tin source and surfactant template. The largest response of the gas sensors to 1000 ppm hydrogen at 350 • C appeared to be 42 [30]. It is evidential that the responses of the sensors based on the ordered mesoporous SnO 2 to hydrogen have been dramatically increased. The one limitation of these studies is that the selectivity of the mesoporous SnO 2 has not been studied.
Shahabuddin et al. reported the sputter deposited SnO 2 thin film gas sensors with 9 nm thin Pt clusters. The Pt/SnO 2 sensor shows an improvement in sensing response: 168 towards 500 ppm of hydrogen at 110 • C. The sensor revealed negligible cross sensing signals against acetone, IPA, NO 2 , methane, LPG, etc. [10]. Gong et al. reported the mesoporous nanocrystalline SnO 2 gas sensor based on the fabricated SnO 2 sputtering with Pt thin film. The gas sensor showed the response of about 1.8 to 1000 ppm hydrogen at 250 • C [31]. The sensitivities and selectivity of the SnO 2 gas sensors could be significantly improved by sputtering with Pt thin film. However, the method requires an expensive facility and complex sample preparing process.
In our work, the ordered mesoporous SnO 2 was prepared by simple sol-gel method. The gas sensors were prepared with a simple and low cost screen printing method while the mesoporous SnO 2 worked as the modified layers. It was shown that both the sensitivities and the selectivity of the gas sensors to hydrogen were improved. The S(c/m2) sensor showed the largest response 22.2 to 1000 ppm hydrogen at 400 • C. The response to hydrogen is >10 times higher than that of the sensor without the modified layer (the S(c) sensor). Compared with the responses of the S(c) sensor, the responses of the S(c/m2) sensor to benzene and ethanol did not change significantly.

Conclusions
Ordered mesoporous SnO 2 powders were prepared by employing Na 2 SnO 3 ·4H 2 O, C 16 PyCl and trimethylbenzene. The specific surface area of the m-SnO 2 powder was 262.30 m 2 /g after calcination at 600 • C. The gas sensors were fabricated using m-SnO 2 films as the modified layers. It was proven that the gassensing performance of the gas sensors could be highly improved, especially to hydrogen, compared with ethanol or benzene gas. In addition, the S(c/m2) sensor exhibited the highest sensitivity (response: Ra/Rg = 22.2) to 1000 ppm hydrogen at 400 • C. The main reason for the high selectivity may be the diffusivity of hydrogen molecules in the ordered mesopores is easier than that of ethanol and benzene molecules.