Preparation of Mesoporous and/or Macroporous SnO2-Based Powders and Their Gas-Sensing Properties as Thick Film Sensors

Mesoporous and/or macroporous SnO2-based powders have been prepared and their gas-sensing properties as thick film sensors towards H2 and NO2 have been investigated. The mesopores and macropores of various SnO2-based powders were controlled by self-assembly of sodium bis(2-ethylhexyl)sulfosuccinate and polymethyl-methacrylate (PMMA) microspheres (ca. 800 nm in diameter), respectively. The introduction of mesopores and macropores into SnO2-based sensors increased their sensor resistance in air significantly. The additions of SiO2 and Sb2O5 into mesoporous and/or macroporous SnO2 were found to improve the sensing properties of the sensors. The addition of SiO2 into mesoporous and/or macroporous SnO2 was found to increase the sensor resistance in air, whereas doping of Sb2O5 into mesoporous and/or macroporous SnO2 was found to markedly reduce the sensor resistance in air, and to increase the response to 1,000 ppm H2 as well as 1 ppm NO2 in air. Among all the sensors tested, meso-macroporous SnO2 added with 1 wt% SiO2 and 5 wt% Sb2O5, which were prepared with the above two templates simultaneously, exhibited the largest H2 and NO2 responses.


Introduction
In recent years, the development for porous materials is been an essential objective of materials science research. This interest is the result of the progress in all fields of industry and technology [1][2][3][4][5]. According to the IUPAC definition, microporous materials are those with pore diameters less than 2 nm, mesoporous materials are those that have pore diameters between 2 and 50 nm, and macroporous materials are those with pores bigger than 50 nm [6]. Among them, macroporous and mesoporous silica with sufficient thermal stability has been applied to catalysts [7,8] membranes [9], adsorbents [10], chemical sensors [11] and templates for nanowires [12]. On the contrary, the poor thermal stability of non-silica mesoporous materials limits their applications. Over the past 50 years, semiconductor metal oxides such as SnO 2 , ZnO and In 2 O 3 have been extensively studied as gas sensing materials due to their various advantages such as the facile fabrication process of thin and thick films, low cost and high thermal stability [13,14]. Among the various metal oxides, SnO 2 is one of the most attractive materials for semiconductor gas sensors [13][14][15][16][17][18][19][20][21][22][23] operated at elevated temperatures (200-600 °C). The gas sensing property of semiconductor gas sensors is largely dependent on various factors such as shape and size of the oxide particles [24][25][26][27][28][29][30]. In addition, strict control of nanostructure of the oxide powders is also quite effective in improving the gas sensing properties [31][32][33][34]. Thus, our group's efforts have so far been directed to preparing thermally stable mesoporous (m-) [15][16][17] and macroporous (mp-) [18,21] oxide films. However, the H 2 sensing properties of the m-SnO 2 sensors were relatively lower than expected from their large specific surface area and mp-SnO 2 showed rather excellent sensing properties to H 2 . Moreover, our recent studies have demonstrated the successful preparation of thermally stable meso-macroporous (m·mp-) SnO 2 and the improvement of gas sensing properties by employing pellet-type sensor structures [19]. However, the mechanical strength of the m· mp-SnO 2 pellets was not enough for long-term operation and this then became a subject for further investigation.
The present study is thus directed to developing m-, mp-and m· mp-SnO 2 thick film sensors. The sensors were fabricated by screen-printing of their as-prepared powders, which were produced by employing sodium bis(2-ethylhexyl)sulfosuccinate (aerosol-OT, AOT) as a mesopore template and PMMA microspheres with an average diameter of 800 nm as a macropore template, and then subsequent calcination at 600 °C for 5 h. The effects of the addition of SiO 2 and Sb 2 O 5 to m-, mp-and m· mp-SnO 2 powders on their H 2 and NO 2 sensing properties were also examined.

Preparation of Mesoporous and/or Macroporous SnO 2 -Based Powders
Various SnO 2 -based powders with well-developed mesopores and/or macropores were prepared by a sol-gel method using SnCl 4 · 5H 2 O (Kishida Chem. Co., Ltd.) as a Sn source, AOT (Kishida Chem. Co., Ltd.) as a mesopore template and PMMA microspheres with an average diameter of 800 nm (MP-1600, Soken Chem. & Eng. Co., Ltd.) as a macropore template. A given amount of SnCl 4 · 5H 2 O (1.75 g) was mixed in 400 mL of ultra pure water together with an appropriate amount of AOT and/or PMMA microspheres. In some cases, appropriate amounts of tetraethoxysilane (TEOS, Kishida Chem. Co., Ltd.) and/or SbCl 3 (Kishida Chem. Co., Ltd.) were also added to the solution, in order to prepare SnO 2 powders added with the given amounts of SiO 2 and/or Sb 2 O 5 . Then the pH value of the resulting mixture was adjusted to 8.5 by adding an aqueous solution of NH 3 . The solid product obtained was aged in the solution at 20 °C for 3 days, then the resulting product was separated from the solution by centrifugation. After drying the product in an oven at 80 °C overnight, the resulting powder product was treated with a 0.1 mol L −1 phosphoric acid solution for about 2 h, and the resulting product was dried in an oven at 80 °C overnight. The powder product resulting after pulverization is referred to as-prepared powder. The as-prepared powders were used for fabricating thick film sensors, but for the characterization tests, as-prepared powders were subjected to calcination at 600 °C for 5 h in air, which are the same conditions adopted for the thick film sensors after the screen-printing of a paste of as-prepared powders. The preparation conditions and compositions of all SnO 2 -based powders obtained in this study and their abbreviations are summarized in Table 1.  In our study, the SnO 2 -based powders prepared using AOT or PMMA microspheres as a template are identified by using abbreviations such as m-TxSy or mp-TxSy, respectively, and the SnO 2 -based powder prepared using both AOT and PMMA microspheres as templates is indicated as m· mp-TxSy, as shown in Table 1. Here, T and S mean the addition of TEOS and SbCl 3 in the precursor solution, respectively, and x (x = 0, 1, 5) and y (y = 0, 1, 5) represent the added amounts of SiO 2 and Sb 2 O 5 (wt%) with respect to the weight of SnO 2 , respectively, on the basis of the expected weight of constituent oxides after calcination.
Crystal phase and crystallite size of SnO 2 -based powders were characterized with X-ray diffraction (XRD, CuKα, Shimadzu Corp., RINT-2200). The crystallite size was calculated by using Scherrer's formula: where λ is the wavelength of CuKα, β is the full-width at the half-maximum of the (110) line and θ is the diffraction angle of the (110) peak. The specific surface area, pore volume and pore size distribution of SnO 2 -based powders were measured by the BET method using a N 2 sorption isotherm (Micromeritics Instrument Corp., TriStar3000). Morphology of SnO 2 -based thick films was observed by a scanning electron microscope (SEM, JEOL Ltd., JCM-5700).

Fabrication of Mesoporous and/or Macroporous SnO 2 -Based Thick Film Sensors
The as-prepared SnO 2 -based powder was mixed with a printing oil which is composed of an alkyl ester of methacrylic acid as a binder, a toluene-based solvent, and an ammonium salt of polyacrylic acid as a plasticizer, and the resulting paste was screen printed on an alumina substrate, on which a pair of interdigitated Pt electrodes (gap between electrodes: 130 μm) had been printed (the thickness of the film was controlled to be about 20 μm after calcination for all the sensors fabricated). Then the printed film was subjected to heat treatment at 600 °C for 5 h in air prior to response measurements. The gas-sensing properties of the thick film sensors to 1,000 ppm H 2 and 1 ppm NO 2 balanced with air were measured in the temperature range of 350-500 °C. The magnitude of the gas response was defined as the ratio (R a /R g ) of the sensor resistance in air (R a ) to that in a target gas (R g ) for H 2 , but the reverse ratio (R g /R a ) was used for NO 2 .

Characterization of Mesoporous and/or Macroporous SnO 2 -Based Powders
Pore size distribution and specific surface area (SSA) of representative m-SnO 2 , mp-SnO 2 and m· mp-SnO 2 powders after calcination are shown in Figure 1. As shown in Figures 1(a-i), m-T0S0 powder, which was prepared only with the addition of AOT, showed a SSA of 150.9 m 2 g −1 and a larger pore volume of 0.153 cm 3 g −1 with a pore diameter of ca. 3.1 nm at the maximum pore volume (hereafter, it will be referred to as the maximum pore diameter). The characterization data of representative SnO 2 -based powders is summarized in Table 2. The addition of 1 wt% SiO 2 to m-T0S0 induced a slight increase in SSA (162.3 m 2 g −1 ) and reduced the maximum pore diameter to ca. 2.7 nm (see m-T1S0). This result implies the repression of growth of SnO 2 crystallites and/or grains by the added SiO 2 , as was reported by Fukuoka et al. [12]. Simultaneous addition of 1 wt% SiO 2 and 5 wt% Sb 2 O 5 to m-T0S0 resulted in further increase in SSA slightly to a value of 176.5 m 2 g −1 (see m-T1S5). Thus, the addition of Sb 2 O 5 was suggested to be also effective in controlling grain growth, which will be confirmed by the change in SnO 2 crystallite size discussed later.
It was revealed that the introduction of macropores into m-T1S5 was very effective for increasing SSA to a value of 262.7 m 2 g −1 (see m· mp-T1S5, Figures 1(c-ii)). This arises undoubtedly from the decrease in the maximum pore diameter and the increase in pore volume, as summarized in Table 2.
On the other hand, the introduction of macropores only (mp-T1S5), instead of mesopores (m-T1S5), into SnO 2 -based powder reduced SSA to a value of 112.0 m 2 g −1 (compare Figure 2(b-ii) with Figure 2(a-iii)) and then decreased pore volume (see Table 2). From these results, it is confirmed that the introduction of mesopores is essential for obtaining both large specific surface and large pore volume of SnO 2 -based powders.     Figure 3 shows the SEM images of the fracture surface of m-T1S5, mp-T1S5 and m· mp-T1S5 thick film sensors. No formation of macropores in m-T1S5 is reasonable, since no PMMA microspheres were added, as shown in Figure 3(a). But, Figure 3(b,c) confirm the formation of many spherical macropores originating from the morphology of PMMA microspheres as a template in the mp-T1S5 and m· mp-T1S5 thick film sensors. However, the diameter of macropores observed was in the 400-750 nm range, which was smaller than that of the diameter of raw PMMA microspheres, due to shrinkage of resulting voids during the growth of SnO 2 crystallites. As shown in Figure 1 and Table 1, mp-SnO 2 powder prepared in the present study showed relatively larger SSA than the conventional SnO 2 powder (8.4 m 2 g −1 , [35]), indicating the formation of a certain amount of mesopores, irrespective of the addition or not of AOT as a mesoporous structure template. This result implies penetration or diffusion of PMMA fragments into the dried SnO 2 precursor material during the calcination and such fragments may act as a mesoporous template at the interface between the PMMA microsphere and surrounding dried SnO 2 precursor. Thus, after the calcination at 600 °C for 5 h of the mp-SnO 2 thick film, a thin mesoporous layer may be formed at the interface between the pores and SnO 2 particles, as shown schematically in Figure 4(a) [36]. As for m· mp-SnO 2 , it is considered that mesopores are formed uniformly inside all the SnO 2 particles and the whole thick film structure, as shown in Figure 4  Another notable finding in Figure 3 is a relatively longer distance of the macropores in m· mp-T1S5 than that in mp-T1S5. Since AOT was used as a mesoporous template in fabricating m· mp-SnO 2 powder, thermal decomposition and subsequent firing along with generation of combustion gases may induce sponge and/or bulky structure with mesopores, leading to a longer distance of the macropores, as also shown schematically in Figure 4(b). Figure 5 shows XRD patterns of representative m-SnO 2 , mp-SnO 2 and m· mp-SnO 2 powders. Diffraction peaks of all powders were rather broad, indicating low crystallinity, but all peaks could be ascribed to those of tetragonal SnO 2 . The CS value which was calculated for each powder using Scherrer's formula is summarized in Table 2. Variations in CS of representative SnO 2 -based powders are shown Figure 6. On the whole, the CS values were small and were in a range of 2.7-7 nm in diameter, due to the limitation of crystallite growth induced by the phosphoric acid treatment before the calcination [15,16]. Exceptionally, m-T0S0 showed the largest CS value of 7 nm. The CS value was decreased drastically to 3.8 nm by the addition of 1 wt% SiO 2 to m-T0S0 (see m-T1S0 in Table 2 and Figure 6). Thus, the repression of the growth of SnO 2 crystallites by the added SiO 2 could be confirmed from these results [37]. As for the powders containing 1 wt% SiO 2 , CS values were almost comparable, whereas they tended to decrease slightly with increasing amounts of the Sb 2 O 5 additive in each series. In addition, the kind of porous structure, i.e., mesopore, macropore, and mesopore plus macropore, was found to have only a little effect on controlling the CS values. Thus, we can confirm again that the addition of 1 wt% SiO 2 was the most powerful method in reducing the CS value among several factors. The CS values decreased slightly by the Sb 2 O 5 addition in each series, but no diffraction peaks other than SnO 2 were observed in XRD patterns even for the cases of 5 wt% Sb 2 O 5 addition ( Figure 5). This implies that Sb ions added were sufficiently incorporated into the SnO 2 crystal lattice and this solid-solution is also effective for the repression of the crystal growth among SnO 2 -based crystallites [23,38,39]. These results demonstrate that the pore size distribution, SSA and CS values of SnO 2 -based powders can be controlled by selecting the kinds of templates, the kind of additives and their additive amounts.

H 2 and NO 2 Sensing Properties of Mesoporous and/or Macroporous SnO 2 -Based Sensors
Variations in sensor resistance of SnO 2 -based thick film sensors in air at 450 °C with amounts of Sb 2 O 5 added are shown in Figure 7. The m-T0S0 sensor showed the lowest resistance in air, but the addition of 1 wt% SiO 2 to m-T0S0 increased the sensor resistance in air (see m-T1S0). The sensor resistance of other two series sensors, i.e., mp-T1S0 and m· mp-T1S0, in air was also very high. Even if Si 4+ ions would be substituted for Sn 4+ ion sites, no valency control effect could be expected. Therefore, SiO 2 added was anticipated to be segregated among SnO 2 crystallites and/or grains and then to reduce electronic conduction of SnO 2 -based thick film sensors, although the segregation of SiO 2 was not confirmed by the XRD measurements due to its small amount added. Introduction of macropores into SnO 2 by using PMMA microspheres (see mp-T1Sy series), instead of the introduction of mesopores, and/or the simultaneous introduction of macropores (see m· mp-T1Sy series) also resulted in an increase in sensor resistance. This phenomenon can be considered to arise mainly from the introduction of air voids, which are electrical insulators, via various pores in the thick film sensors, but the mp-T1Sy sensor with macropores showed the largest resistance in air, irrespective of the smallest pore volume, among three series of sensors. This fact implies the existence of another factor, besides the pore volume, in determining the sensor resistance in air, such as the manner of distribution of pores in the thick film and so on, though the details are not clear at present.
In each sensor series, the sensor resistance in air decreased with increasing amounts of Sb 2 O 5 additive. This behavior can be explained by the valency control, i.e., partial substitution of Sn 4+ sites with Sb 5+ ions, producing free electrons, as described in Equation (2) These results also confirm the existence of substituted Sb 5+ ions, i.e., the solid-solution of between Sb 2 O 5 and SnO 2 and then little amount of segregated Sb 2 O 5 among SnO 2 -based particles. Figures 8 and 9 show temperature dependence of response of SnO 2 -based thick film sensors to 1,000 ppm H 2 balanced with air and 1 ppm NO 2 balanced with air. Almost all sensors showed the maximum response to 1,000 ppm H 2 at a temperature of 450 °C. In contrast, the response to 1 ppm NO 2 of all sensors tended to increase as the operating temperature decreased, and showed the largest response in the temperature range studied at 350 °C.  Response transients of SnO 2 -based thick film sensors to 1,000 ppm H 2 at 450 °C and 1 ppm NO 2 at 350 °C balanced with air are shown in Figures 10 and 11, respectively. In this study, 50% response time is defined as a period necessary to achieve 50% of resistance value of R a − R g , while 50% recovery time is defined as that necessary to achieve 50% of resistance value of R g − R a for H 2 .
The 50% response and recovery times to NO 2 are also defined in the similar manner, but by using R g − R a for response time and R a − R g for recovery time. Hereafter, they are simply expressed as response time and recovery time, respectively. Response and recovery times of SnO 2 -based thick film sensors to 1,000 ppm H 2 at 450 °C and 1 ppm NO 2 at 350 °C were summarized in Table 3. The m-T0S0 sensor showed the longest response and recovery times to H 2 among the sensors listed in Table 3. The simultaneous addition of 1 wt% SiO 2 and 5 wt% Sb 2 O 5 to m-T0S0 was found to shorten slightly response and recovery times to H 2 (see m-T1S5).  However, the m-T1S5 sensor showed longer response and recovery times to 1 ppm NO 2 than m-T0S0. Thus, the effect of additive on the response and recovery times varied with the kind of target gas. The introduction of macropores into m-T1S5 shortens the response and recovery times to H 2 . More remarkable shortening of the recovery time to H 2 as well as response and recovery times to NO 2 were observed with m· mp-T1S5. It is reasonable to consider that the response time to H 2 is closely related to the diffusivity of H 2 , while the recovery time is controlled by the diffusivity of O 2 which has a larger molecular size than H 2 . As for NO 2 , on the other hand, both the response and recovery times are considered to affected by the diffusivity of NO 2 itself, which has a larger molecular size than H 2 , from its gas sensing mechanism. Such considerations predict a shorter recovery time to H 2 as well as shorter response and recovery times to NO 2 by the introduction of macropores into the sensor materials. The results obtained with m· mp-T1S5 were in good agreement with this prediction. Thus, the mp-T1S5 sensor, which was fabricated only by the introduction of macropores, showed the fastest response to H 2 as well as the fastest response and recovery times to NO 2 among the sensors tested. But, the reason for the longer recovery time to H 2 of mp-T1S5 than m· mp-T1S5 is not clear at present. Anyway, such behavior undoubtedly arises from more easy diffusion of a target gas as well as oxygen through mesopores rather than macropores. On the other hand, all of the response and recovery times to NO 2 are much longer than those to H 2 . This may arise not only from slow diffusivity of NO 2 in comparison to H 2 , but also from slow adsorption rate and strong interaction of NO 2 − species on the oxide surface.  Figures 12 and 13 show variations in responses of SnO 2 -based sensors to 1,000 ppm H 2 at 450 °C and to 1 ppm NO 2 at 350 °C in air with amounts of Sb 2 O 5 added, respectively. From these figures, it is also apparent that the m-T0S0 sensor showed the smallest responses to both H 2 and NO 2 among the sensors studied. The addition of 1 wt% SiO 2 to m-T0S0 enhanced responses to both H 2 and NO 2 to a certain level for every series of sensors. In addition, H 2 and NO 2 responses increased with increasing amounts of Sb 2 O 5 additive in each series of sensors. On the whole, the magnitude of the response was in the order of m· mp-T1Sy > mp-T1Sy > m-T1Sy, when the comparison was made at the same additive amount of Sb 2 O 5 , with only one exception observed for the NO 2 response of the m· mp-T1S0 sensor. It is worth noting that the mp-T1Sy series sensors showed higher H 2 and NO 2 responses than those of m-T1Sy series sensors, irrespective of their smaller surface area. This implies that all the surface of sensor materials including the inner surface of mesopores is not utilized effectively for gas detection, and that easy diffusion of a target gas as well as oxygen to the active surface, i.e., the existence of certain amounts of macropores inside the thick film sensors, is more important for improving gas response. The highest H 2 and NO 2 responses observed with the m· mp-T1Sy sensors may be a result of good combination of mesopore and macropores in the thick film sensors. From these results, it was revealed that the strict control of microstructure having well-developed mesoporous and macroporous is indispensable to enhancing gas reactivity and diffusivity and thus to improving responses to H 2 and NO 2 in air.

Conclusions
Mesoporous and/or macroporous SnO 2 -based powders have been prepared by a sol-gel method by employing SnCl 4 · 5H 2 O, ATO as a mesopore template, PMMA microspheres as a macropore template, and their gas-sensing properties as thick film sensors towards 1 ppm NO 2 as well as 1,000 ppm H 2 in air have been investigated. The addition of SiO 2 into mesoporous and/or macroporous SnO 2 was found to increase SSA of mesoporous SnO 2 . However, the SSA of all samples increased and their CS tended to decrease slightly with the addition of the Sb 2 O 5 . The additions of SiO 2 and Sb 2 O 5 into mesoporous and/or macroporous SnO 2 were found to improve the sensing properties of the resulting sensors. The addition of SiO 2 into mesoporous and/or macroporous SnO 2 was found to increase the sensor resistance in air. However, the doping of Sb 2 O 5 into mesoporous and/or macroporous SnO 2 was found to markedly reduce the sensor resistance in air, and to increase the response to 1,000 ppm H 2 as well as 1 ppm NO 2 in air. Among all the sensors tested, meso-macroporous SnO 2 mixed with 1 wt% SiO 2 and 5 wt% Sb 2 O 5 , which were prepared with above two templates simultaneously, exhibited the largest H 2 and NO 2 responses.