Effects of catalytic combustion behavior and adsorption/desorption properties on ethanol-sensing characteristics of adsorption/combustion-type gas sensors

ABSTRACT Adsorption/combustion-type gas sensors, subspecies of catalytic combustion-type gas sensors, show large dynamic responses to volatile organic compounds (VOCs) under the operation with a mode of pulsed temperature heating, because of the flash catalytic combustion of target VOCs adsorbed on the gas-sensing films. Catalytic combustion behavior of ethanol over γ-Al2O3 powders loaded with and without 1 wt% Pt and/or 10 wt% metal oxide (MO: CeO2 or Bi2O3) and their adsorption/desorption properties of ethanol (adsorption temperature: 150°C) were investigated, and then the ethanol-sensing characteristics of the sensors utilizing the gas-sensing materials (low and high temperatures under dynamic operation: 150°C and 450°C, respectively) have been discussed on the basis of the findings on both their catalytic combustion behavior and adsorption/desorption properties. Especially, the co-loading of Pt with CeO2 onto γ-Al2O3 was the most effective in enhancing the dynamic response, because the relatively small amounts of various compounds that were adsorbed on the surface at 150°C efficiently oxidized at the initial stage of the pulse-driven heating to form CO2. The effects of low and high temperatures under the pulse-heating operation on the ethanol-sensing properties of the sensor utilizing γ-Al2O3 co-loaded with Pt and CeO2 were also clarified on the basis of the above findings.


Introduction
Catalytic combustion-type gas sensors are operated by monitoring the heat, which is generated by the combustion of target gases, and thus they have generally been used for monitoring inflammable gases such as hydrogen, carbon monoxide, methane, and propane [1][2][3][4][5]. The practical sensors are quite useful in different application fields, because the magnitude of responses to the inflammable gases is almost proportional to the gas concentration until the lower explosive limit (LEL). Recently, the sensors have also been adopted as hydrogen detectors for fuel-cell vehicles as well as common gas-leak detectors for multiple uses. However, the catalytic combustion-type gas sensors do not show so much larger responses to the small amounts of volatile organic compounds (VOCs) under general operation. To overcome the drawback, Sasahara et al. proposed adsorption/combustion-type gas sensors, which actively utilize the dynamic sensorsignal profile under pulse-heating operation, although the sensor structure is based on the catalytic combustion-type gas sensors [6][7][8][9][10][11]. The expected gassensing mechanism is as follows. First, VOC molecules adsorb on both the sensing and reference films of the adsorption/combustion-type gas sensors at a low temperature for a certain period. Next, the adsorbed VOC molecules burn especially on the sensing film at the initial stage of the pulse high-temperature heating, while they mainly just desorb from the inactive reference film. The sensor-signal profile obtained typically shows a quite large dynamic output by the flash catalytic combustion of VOC molecules adsorbed on the sensing film. The microsensor platform with extremely small thermal capacity, which is fabricated by MEMS (Micro Electro Mechanical Systems) technology, has been utilized to detect even a slight amount of heat generated by the combustion of the low concentration of VOCs adsorbed on the sensing film. Sasahara et al. have thus far demonstrated that the adsorption/ combustion-type gas sensors can sufficiently detect as low as 40 ppb formaldehyde [8] and 10 ppb toluene [9,10] in ambient air by optimizing the composition of the sensing-film materials as well as the operating conditions. Besides, the adsorption/combustion-type gas sensors show specific dynamic sensor-signal profiles largely dependent on the kinds of target VOCs [6,7] and detect the amounts of total VOCs generated in a newly constructed building [10] as well as at early stages of fire [11] sensitively. More recently, our group have reported that the compositional and microstructural control of the gas-sensing materials is quite effective in drastically enhancing the magnitude of dynamic responses of the adsorption/combustiontype gas sensors [12][13][14][15][16][17][18]. Our fundamental strategy to improve the VOC-sensing properties is to increase the VOC adsorbability and VOC-oxidation activity of the gas-sensing materials while maintaining the thermal conductivity. Mesoporous γ-alumina (γ-Al 2 O 3 ) powders with large specific surface area, which were prepared by microwave-assisted solvothermal technique, were promising as base materials of the sensing and reference films [12], and the simultaneous loading of Au and Pd onto the γ-Al 2 O 3 powder even by impregnation technique was effective in improving the ethanol-sensing properties of the adsorption/combustion-type gas sensors [13]. The loading of core(Au)/ shell(Pd) nanoparticles onto the γ-Al 2 O 3 powder by sonochemical reduction technique achieved much higher dispersion of Au and PdO nanoparticles on the surface, and thus largely contributed to the excellent sensing properties to ethanol, toluene, and n-hexane [14]. The addition of the α-Al 2 O 3 powder with relatively high thermal conductivity to the sensing film also improved the ethanol-sensing properties, even though the specific surface area of the α-Al 2 O 3mixed γ-Al 2 O 3 powders was much lower than that of the pristine γ-Al 2 O 3 powder [13,17]. Moreover, various oxides (Bi 2 O 3 , CeO 2 , Co 3 O 4 , CuO, Fe 2 O 3 , Mn 2 O 3 , NiO, RuO 2 , ZrO 2 , etc.) were co-loaded onto the γ-Al 2 O 3 powder with Pt, to improve the ethanol-sensing properties of the sensors [15,16]. Among these candidates for the gas-sensing materials, the co-loading of Pt with CeO 2 on the γ-Al 2 O 3 powder effectively enhanced the sensing properties to various VOCs [15], while the coloading of Pt with Bi 2 O 3 on the γ-Al 2 O 3 powder was effective in enhancing the sensing properties to only a specified VOC (e.g., toluene) [16]. However, an incomplete understanding of the essential VOCsensing mechanism prevents us from optimizing the composition and microstructure of the gas-sensing materials to further improve the sensitivity as well as the selectivity. Therefore, the catalytic combustion behavior of ethanol as a fundamental target VOC gas over mesoporous γ-Al 2 O 3 powders co-loaded with and without Pt and/or metal oxide (CeO 2 or Bi 2 O 3 ), has been clarified in this study. The adsorption/desorption properties of ethanol over these materials have also been investigated, and these effects on the ethanolsensing characteristics of the adsorption/combustiontype gas sensors employing these materials as a sensing-film material have been discussed in detail.

Preparation of mesoporous γ-Al 2 O 3 and MOloaded γ-Al 2 O 3 powders (MO: metal oxide)
Aluminum secondary butoxide (4.2 × 10 −2 mol) and behenic acid (mesopore-template material, 1.0 × 10 −2 mol) were mixed with 1-propanol (300 cm 3 ). After deionized water (10 cm 3 ) was added to the solution, it was stirred at room temperature for 45 h to hydrolyze the aluminum secondary butoxide. Then, the solution was solvothermally treated at 110°C for 1 h. After the obtained precipitates were centrifuged at 2500 rpm for 10 min and then washed with ethanol, the mesoporous γ-Al 2 O 3 powder was prepared after firing the precipitates at 700°C for 2 h in air [12]. In some cases, an appropriate amount of Ce(NO 3 ) 3 · 6H 2 O or Bi(NO 3 ) 3 · 5H 2 O was subsequently mixed with the γ-Al 2 O 3 powder in a small amount of de-ionized water, they were evaporated to dryness, and then the obtained precipitates were fired at 700°C for 1 h in air. The obtained MO-loaded γ-Al 2 O 3 powders (MO: CeO 2 or Bi 2 O 3 ) were denoted as nMO/γ-Al 2 O 3 (n: the amount of MO loaded, 10 (wt%) in this study).

Loading with Pt nanoparticles onto the surface of γ-Al 2 O 3 and 10MO/γ-Al 2 O 3 powders
We have already demonstrated that the noble-metal loading onto γ-Al 2 O 3 by utilizing sonochemical reduction technique was quite effective in increasing the ethanol response, in comparison with that by general impregnation technique [14]. Therefore, Pt nanoparticles were synthesized by the sonochemical reduction technique, in this study. The appropriate amount of the dihydrogen hexachloroplatinate(IV) hexahydrate was dissolved into de-ionized water, and then the 10 mM aqueous solution obtained (10 cm 3 ) and a 4 mM polyethylene glycol monostearate (C 17

Sensor fabrication and measurement
A schematic drawing of the adsorption/ combustion-type gas sensor fabricated is shown in Fig. S1.  [15]. The sensors obtained are referred to as the kind of VOC-sensing materials in this study. The sensing properties of the adsorption/combustion-type gas sensors fabricated were measured in an acrylic chamber (inner volume: 50 dm 3 ), in which 10-1000 ppm ethanol was evaporated with a compact heater. The sensor was incorporated into a bridge circuit with two fixed resistive elements, which was used to get output signals from the sensor. All sensors were operated with a mode of pulsed temperature heating (high temperature (T H ) at 250-450°C for 0.4 s after low temperature (T L ) at RT-150°C for 9.6 s with a cycle of 10 s) by applying rectangular pulse voltage, as shown in Fig. S2(a) and Fig. S2(b). The rate of temperature rise in the sensing and reference films on a microsensor chip was quite fast (the period required for the rapid increase in temperature from RT to 450°C: merely ca. 30 ms [6]). Figure S2(c) shows a typical sensor-signal profile of an adsorption/combustion-type gas sensor. The wave profile of the sensor response generally has one dynamic signal mainly by flash combustion of ethanol and the related compounds adsorbed on the sensing-film surface and a subsequent static signal by general catalytic combustion of ethanol during the pulsed high-temperature heating. The magnitude of general ethanol response, ΔV MAX , was defined as the maximum of the difference between output voltages in ethanol balanced with air and that in air. Besides, two kinds of response values calculated by integrating the wave profiles, i.e. approximately integrated dynamic response (IDR) and approximately integrated static response (ISR) were also defined as shown in Fig.  S2(d).

Catalytic combustion activity and temperature-programmed desorption and oxidation properties
All the powders were pressed into a disc, and then crushed into granules (ca. 9-20 mesh). The granules obtained (ca. 0.6 g) were fixed in a glass reactor connected with a flow apparatus. After they were pretreated at 500°C for 1 h and then cooled down to selected temperatures (30-500°C, rates of temperature rise and temperature drop: 3°C min −1 and −3°C min −1 , respectively) in dry air (30 cm 3 min −1 ), the catalytic combustion behavior of 923 ppm ethanol over all samples was characterized at each temperature in dry air (30 cm 3 min −1 , space velocity (SV): 1664 h −1 ). Temperature-programmed desorption (TPD) and temperature-programmed oxidation (TPO) properties of all samples were also investigated by using the same experimental setup. After they were pre-treated at 500°C for 1 h in dry air (30 cm 3 min −1 ), they were cooled down to 150°C (rate of temperature drop: −3° C min −1 ), which is the typical low-power heating temperature under sensor operation. The flow gas was changed from dry air to 923 ppm ethanol balanced with dry air, and then all the samples were treated at 150°C for 1 h in 923 ppm ethanol balanced with dry air (30 cm 3 min −1 ). After an abrupt drop in temperature to 30°C in helium (for TPD) or dry air (for TPO), the TPD or TPO profile was measured at a rate of temperature rise of 3°C min −1 in the same gas (30 cm 3 min −1 , space velocity: 1664 h −1 ). The kind and concentration of outlet gases were analyzed by using a gas chromatograph/mass spectroscope (GC-MS; Shimadzu Corp., GCMS-QP5050 (capillary column used: PoraPLOT Q)) and a GC equipped with an FID detector (Shimadzu Corp., GC-2010 (capillary column used: DB-5)).  Figure 1(b) shows ethanolconcentration dependences of three kinds of responses, ΔV MAX , IDR, and ISR of all the sensors. In particular, the magnitudes of ΔV MAX and IDR of 1Pt/γ-Al 2 O 3 and 1Pt/ 10MO/γ-Al 2 O 3 sensors, which mainly originate from their dynamic responses, seem to be intimately related to the adsorption of ethanol and/or the related components onto these surfaces, because the relationship between "ΔV MAX or IDR" and "concentration" was quite similar to general adsorption isotherm [22][23][24]. On the other hand, the magnitude of ISR of the sensors was roughly proportional to ethanol concentration, and the behavior approximately corresponds to that of general catalytic combustion-type gas sensors [25,26]. Figure 2 shows catalytic combustion behavior of 923 ppm ethanol over unloaded γ-Al 2 O 3 , 1Pt/γ-Al 2 O 3 , 10MO/γ-Al 2 O 3 , and 1.0Pt/10MO/γ-Al 2 O 3 in dry air (dependences of ethanol-conversion ratio, CO 2production ratio (the ratio of the amount of CO 2 produced to the amount of total carbon in 923 ppm ethanol), and the concentration of products on temperature). Besides, various parameters on their catalytic combustion behavior are shown in Table 1.

Catalytic combustion properties
The conversion onset temperature of ethanol, T CO (E), and the complete conversion temperature of ethanol, T CC (E), of the γ-Al 2 O 3 were quite high (ca. 120°C and ca. 280°C, respectively), and thus the γ-Al 2 O 3 showed the lowest oxidation activity of ethanol (the temperature at which 50% of ethanol was converted, T 50 (E): ca. 170°C) among all the samples. On the other hand, the production onset temperature of CO 2 (T PO (CO 2 ), ca. 120°C) was quite comparable to T CO (E), while the complete production temperature of CO 2 (T CP (CO 2 ), ca. 300°C) were much higher than T CC (E). Therefore, the temperature at which 50% of ethanol was oxidized to CO 2 (T 50 (CO 2 ), ca. 235°C) was much higher than T 50 (E) (ca. 170°C). Three kinds of gases (ethylene, acetaldehyde, and diethyl ether) were confidently produced as reaction intermediates, and acetaldehyde and diethyl ether especially generated in the temperature range from ca. 120°C (i.e. T PO (CO 2 )) to ca. 300°C (i.e. T CP (CO 2 )). On the other hand, ethylene generated over the whole temperature range of more than 150°C, and the amount of ethylene was the largest around 400°C, even though ethanol was almost perfectly oxidized to produce CO 2 . These catalytic reactions were generally given as follows [27,28].
The ratio of each reaction to complete catalytic oxidation reaction of C 2 H 5 OH, were largely dependent on the acid-base property and the catalytic oxidation/reduction activity on the surface [29][30][31][32]. At least, the generation of the large amounts of these reaction intermediates obviously shows that the oxidation activity of ethanol over γ-Al 2 O 3 was quite low. Especially, the conversion ratio was ca. 14%, and the CO 2 -production ratio was almost 0% at 150°C (T L in Figure 1), only with the generation of a little amount of diethyl ether. The mass-balance ratio (MBR) in this study is defined as (5) where [ethanol]  The MBR is nearly 1.0 at 150°C, which means that unknown reaction intermediates were hardly produced on the γ-Al 2 O 3 surface. On the other hand, both the ethanol-conversion ratio and the CO 2production ratio were almost 100% at 450°C (T H in Figure 1), only with the generation of a little amount of ethylene (ca. 56 ppm). The loading of 10 wt% MO (CeO 2 or Bi 2 O 3 ) onto γ-Al 2 O 3 slightly improved the conversion property over γ-Al 2 O 3 . Namely, all parameters on the ethanolconversion ratio (T 50 (E), T CO (E), and T CC (E)) and CO 2production ratio (T 50 (CO 2 ), T PO (CO 2 ), and T CP (CO 2 )) were a little lower than those of γ-Al 2 O 3 . Especially according to the reduction in T PO (CO 2 ) of both the samples, their production onset temperature of acetaldehyde (T PO (AA)) substantially decreased from more than 150°C to ca. 50°C by the MO loading (the temperature at which the largest amount of acetaldehyde produced: ca. 180°C for both 10MO/mp-Al 2 O 3 samples), and the total amount of acetaldehyde produced over 10Bi 2 O 3 /γ-Al 2 O 3 and 10CeO 2 /γ-Al 2 O 3 became larger and smaller than that over γ-Al 2 O 3 , respectively. On the other hand, the amount of ethylene produced was drastically decreased by both the Bi 2 O 3 and CeO 2 loading, and the Bi 2 O 3 loading especially reduced the amount of ethylene produced in the hightemperature range of more than 270°C. In terms of diethyl ether, the CeO 2 loading lowered the production onset temperature (T PO (DE), less than 100°C) to exceed the amount of acetaldehyde produced at 150°C (T L ), whereas the Bi 2 O 3 loading raised the T PO (DE) to inhibit the production of diethyl ether at 150°C. The MBR of both 10MO/γ-Al 2 O 3 samples at 150°C was also ca. 1, just like mp-Al 2 O 3 , even though the conversion ratio (ca. 50% for both 10MO/γ-Al 2 O 3 samples) and the CO 2 -production ratio (ca. 15% for 10CeO 2 /γ-Al 2 O 3 and ca. 37% for 10Bi 2 O 3 /γ-Al 2 O 3 ) at 150°C were much larger than those of γ-Al 2 O 3 . These results indicate that the loading of Bi 2 O 3 or CeO 2 hardly produced other reaction intermediates emitted leastwise to the gaseous atmosphere at 150°C. Considering these results on the catalytic combustion behavior of ethanol over the 10MO/γ-Al 2 O 3 surfaces, the effects of the MO loading onto γ-Al 2 O 3 at the operating temperature of 150°C can be explained by the following scenario. The CeO 2 loading enhanced the production reaction of acetaldehyde and diethyl ether produced through equations (2) and (3), respectively, on the surface at 150°C, and certain amounts of these intermediates were also oxidized to form CO 2 . The Bi 2 O 3 loading mainly enhanced the dehydrogenation of ethanol to acetaldehyde at 150°C, and the large part of acetaldehyde desorbed from the surface without further oxidation. On the other hand, both the conversion and the CO 2production ratio were nearly 100% at 450°C, with little production of reaction intermediates.
The loading of only 1 wt% Pt also enhanced the catalytic activity of γ-Al 2 O 3 , and the effect of the Pt loading was larger than that of the MO loading.
On the other hand, T PO (CO 2 ) and T CP (CO 2 ) of 1Pt/γ-Al 2 O 3 was slightly higher than those of 10MO/γ-Al 2 O 3 , whereas T 50(CO2) of 1Pt/γ-Al 2 O 3 (ca. 152°C) was lower than those of 10MO/γ-Al 2 O 3 . The Pt loading largely decreased the amounts of ethylene and diethyl ether produced over γ-Al 2 O 3 , and drastically reduced the T PO (AA) with a decrease in temperature at which the largest amount of acetaldehyde was produced (ca. 180°C), as is the case with the loading of 10 wt% MO. The MBR of 1Pt/γ-Al 2 O 3 at 150°C (ca. 0.8) was smaller than those of γ-Al 2 O 3 and 10MO/mp-Al 2 O 3 . Otherwise, the conversion ratio (ca. 88%) and the CO 2 -production ratio (ca. 51%) at 150°C were much larger than those of 10MO/γ-Al 2 O 3 . These results indicate that other reaction intermediates adsorbing on the surface and/or desorbing to the gaseous atmosphere (not detected by GC-MS), which were different from either ethylene, acetaldehyde, or diethyl ether, were produced by the Pt loading onto the γ- The co-loading of Pt with MO quite further enhanced the catalytic oxidation activity of ethanol over γ-Al 2 O 3 . Namely, the co-loading of Pt with MO was quite effective in improving the ethanol conversion of γ-Al 2 O 3 , compared with the loading of only Pt or MO, and the 1Pt/10CeO 2 /γ-Al 2 O 3 especially showed the most excellent ethanolconversion property (the lowest T 50 (E), ca. 68°C). The co-loading of Pt with CeO 2 effectively improved the complete oxidation activity as well, and reduced the amounts of three reaction intermediates produced. T 50 (CO 2 ) and T PO (CO 2 ) of the 1Pt/10CeO 2 /γ-Al 2 O 3 (ca. 145°C and <30°C, respectively) were the lowest among those of all samples. T PO (AA) was quite low (less than 30°C), and the largest amount of acetaldehyde produced around at T 50 (E). Besides, only a little amount of ethylene was produced in a wide temperature range of 30-400°C, whereas the production of diethyl ether was not confirmed. On the other hand, the co-loading of Pt with Bi 2 O 3 was ineffective in improving the complete oxidation activity. Namely, the T 50 (CO 2 ) of 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 was higher than those of 10Bi 2 O 3 /γ-Al 2 O 3 and 10CeO 2 /γ-Al 2 O 3 . The relatively large amount of acetaldehyde was intermittently produced at lower temperatures (30-270°C). The production of diethyl ether was hardly confirmed in the whole temperature range, while the small amount of ethylene produced at around 210°C. The conversion ratio at 150°C (ca. 100% and ca. 90% for 1Pt/10CeO 2 /γ-Al 2 O 3 and 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 , respectively) was quite large in comparison with that of other samples. However, the CO 2production ratio of 1Pt/10CeO 2 /γ-Al 2 O 3 at 150°C was not so much large (ca. 53%), while that of 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 was extremely small (ca. 23%, which was much larger than that of γ-Al 2 O 3 but smaller than those of 10MO/mp-Al 2 O 3 ). Thus, the MBRs at 150°C of 1Pt/10CeO 2 /γ-Al 2 O 3 and 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 were ca. 0.7 and ca. 0.2, respectively. There was no doubt that the co-loading of Pt with MO enhanced the oxidation activity of ethanol, and thus their relatively enhanced oxidation activity probably induced to produce the large amounts of other reaction intermediates which were not detected, by the synergistic effect of Pt with MO. The high-molecular-weight and/or highly polarized intermediates (e.g. with carboxylic, aldehyde, and/ or ester group) is strongly expected to be produced especially on the surface of 1Pt/10Bi 2 O 3 /γ- T 50 : temperature at which 50% of ethanol was converted, T CO : conversion onset temperature of ethanol, T CC : complete conversion temperature of ethanol, T 50 (CO 2 ): temperature at which 50% of ethanol was oxidized to CO 2 , T PO (CO 2 ): production onset temperature of CO 2 , T CP (CO 2 ): complete production temperature of CO 2 , and MBR at 150°C: mass-balance ratio at 150°C, which was defined by eq. (5).
Al 2 O 3 , and thus we have been attempting to get the evidence on the production on the surface and/or in the gaseous atmosphere.

TPD and TPO properties
TPD and TPO profiles generally show temperature dependences of desorption behavior of the target gas adsorbed and/or chemical species produced on the surface, in an inactive gas (helium in this study) and in gas containing oxygen (21% oxygen in this study, i.e. dry air), respectively. When the adsorption/ combustion-type gas sensors operate at a mode of pulse-driven heating, the amounts and kinds of adsorbates derived from the target gas are greatly dependent on the concentration of oxygen (i.e. the amounts of oxygen adsorbates and lattice oxygen at the surface of the sensing and reference materials). Therefore, the comparative characterization of TPD and TPO profiles offers significant findings to clarify the gas-sensing characteristics of the sensors. However, we should discuss these data carefully, because the rate of temperature rise in TPD and TPO profiles (ca. 3°C min −1 ) are drastically different from that of the pulse-driven heating for the adsorption/combustion-type gas sensors (ca. 6 × 10 4°C min −1 , which was roughly estimated).  The TPD and TPO profiles were shown in Figures 3 and  4, respectively.

γ-Al 2 O 3 3.3.1.1. TPD profiles.
The temperature at which ethanol (target gas) started to desorb (onset temperature) was around 100°C, and the desorption amount of ethanol abruptly increased with an increase in temperature. A large amount of ethanol intermittingly desorbed, especially in a temperature range of 150-320°C (the largest concentration: ca. 200 ppm at 180° C), probably due to the direct adsorption on various adsorption sites on the surface, and then the desorption ended at ca. 300°C (offset temperature). Diethyl ether also started to desorb at around 100°C. However, the amount of diethyl ether desorbed at less than low-power heating (ethanol adsorption) temperature, T L (i.e. 150°C), was extremely small, and it abruptly increased with an increase in temperature and it reached the maximum value (ca. 380 ppm) at 215°C. The offset temperature of diethyl ether was also ca. 300°C. Considering that diethyl ether was catalytically produced in the similar temperature range (120-300° C) under steady-state condition (see Figure 2) and that the offset temperature of ethanol was also ca. 300°C in the TPD profile, it is expected that both the forward (dehydration condensation) reaction and reverse (hydrolysis) reaction in the following equation simultaneously proceeded in the temperature range [33].
2C 2 H 5 OHþO nÀ oxygen species on surface ð Þþ 2À n ð Þe À ! C 2 H 5 OC 2 H 5 þ2OH À (6) Ethylene and acetaldehyde started to desorb at much higher temperatures (ca. 180°C and ca. 190°C, respectively), and the maximum amounts of ethylene and acetaldehyde generated was ca. 1950 ppm and ca. 220 ppm, respectively, at ca. 250°C. The offset temperature of both gases was ca. 320°C. Namely, the temperature range in which these gases were generated was relatively higher than that in which ethanol and diethyl ether were generated. Therefore, ethanol and the related adsorbates, which strongly adsorbed on the oxide surface, probably reacted with oxygen species on the γ-Al 2 O 3 surface to produce these gases, according to the following equations, because the catalytic activity of oxygen species (oxygen adsorbates and lattice oxygen) on the γ-Al 2 O 3 surface is quite poor.
C 2 H 5 OHþO nÀ oxygen species on surface ð Þþ 2À n ð Þe À ! C 2 H 4 þ2OH À (7) Besides, these reactions must be severely restricted on the TPD condition, because the oxygen species hardly increase due to the lack of oxygen in the flow gas (impurity oxygen may be slightly immixed from ambient air). Thus, the relative amount of acetaldehyde adsorbs on the γ-Al 2 O 3 surface through equation (8) during the pretreatment at 150°C in ethanol balanced with dry air, and then it desorbs in the temperature range of more than 150°C. The fact that acetaldehyde was hardly produced at 150°C in dry air in the catalytic combustion behavior of ethanol (see Figure 2(c)) strongly supports the above process of acetaldehyde desorption. Ethylene cannot originally adsorb on the γ-Al 2 O 3 surface at 150°C. On the other hand, the TPD profile (Figure 3(e)) showed a little amount of CO 2 (several ten ppm) intermittingly desorbed only in the temperature range of more than 300°C. This onset temperature is close to offset temperatures of three reaction intermediates. This is probably because the small number of relatively high-molecular-weight components (C x H y O z ), which were produced by the polycondensation of ethanol, reacted with the oxygen species on the γ-Al 2 O 3 surface to form CO 2 at the higher temperatures, according to the following equation.

TPO profiles. The amounts of ethanol and
diethyl ether observed in the TPO profiles were extremely reduced due to the mixing of oxygen into the carrier gas (He). Onset and offset temperatures of ethanol generated were ca. 210°C and ca. 300°C, respectively, and the largest concentration was ca. 9.5 ppm at 270°C. The existence of oxygen in the carrier gas smoothly increases oxygen adsorbates (e.g. gen. Therefore, adsorbed ethanol and the related compounds can react with the oxygen species even during the TPO process at lower temperatures to form other reaction intermediates (actually, the large amounts of acetaldehyde and ethylene generated at higher temperatures, as shown in Figure 4) and unknown highermolecular-weight reaction intermediates adsorbed on the γ-Al 2 O 3 surface. Onset and offset temperatures of diethyl ether generated were ca. 155°C and ca. 300°C, respectively, and both the values were almost coincident with those of its TPD profile. Besides, the concentration of diethyl ether generated was quite low. Thus, the small amount of diethyl ether in the TPO profile is expected to desorb from the γ-Al 2 O 3 surface by the same mechanism as that of TPD.
The amount of ethylene observed in the TPO profile, which was also smaller than that desorbed in the TPD profile, was much larger than that of diethyl ether. On the contrary, the amount of acetaldehyde in the TPO profile was about three times larger than that in the TPD profile. These results indicate that the dehydration processes (equations (7) and (6)) retracted and the dehydrogenation (equation (8)) proceeded with an increase in the amount of oxygen in the flow gas. Namely, the dehydrogenation is relatively dominant on the γ-Al 2 O 3 surface in an oxygen-rich gaseous atmosphere at elevated temperatures, since some kinds of oxygen species can effectively adsorb on the surface (equation (10)).
The TPO profile in Figure 4(e) also showed that CO 2 intermittingly desorbed in the temperature range of more than ca. 190°C, and the onset temperature of CO 2 in the TPO profile is lower than that in the TPD profile. Besides, ca. 700 ppm and ca. 400 ppm of CO 2 were produced at ca. 250°C and ca. 430°C, respectively. Considering that the amount of CO 2 produced in the TPO profile is much larger than that in the TPD profile, it is expected that various ethanol and the related compounds adsorbed during the pretreatment were converted and/or transformed to various high-molecular-weight reaction intermediates on the γ-Al 2 O 3 surface, and thus they were all oxidized at commensurate temperatures in the range of ca. 190-500°C. However, the amount of CO 2 generated from γ-Al 2 O 3 was much smaller than that from Pt and/ or MO-loaded γ-Al 2 O 3 samples as shown below, which means that the amounts of ethanol and the related compounds adsorbed on the γ-Al 2 O 3 surface during the pretreatment at 150°C was quite smaller than those adsorbed on the Pt and/or MO-loaded γ-Al 2 O 3 surfaces.

10MO/γ-Al 2 O 3
The amounts of ethanol, ethylene, and diethyl ether generated were drastically decreased by the MO loading in the TPD profiles. However, the generation behavior of acetaldehyde was different from that of others. Namely, both the CeO 2 and Bi 2 O 3 loading slightly lowered the temperature at which the maximum amount of acetaldehyde was generated. Besides, the CeO 2 loading increased the maximum amount of acetaldehyde generated at ca. 215°C, while the Bi 2 O 3 loading decreased the maximum amount of acetaldehyde generated at ca. 235°C. On the other hand, ethanol and three kinds of reaction intermediates were hardly generated in the whole temperature range in the TPO profiles.
Contrary to our expectations, both the TPD and TPO profiles of CO 2 were unique. In these TPD profiles, CO 2 started to be generated at a lower temperature than pretreatment temperature (less than 100°C), and the amount of CO 2 increased with a rise in temperature. The amount of CO 2 generated from 10CeO 2 /γ-Al 2 O 3 was larger than that from 10Bi 2 O 3 /γ-Al 2 O 3 at less than 350°C. However, the amount of CO 2 generated from 10Bi 2 O 3 /γ-Al 2 O 3 abruptly increased with a rise in temperature, and the amount of CO 2 generated from 10Bi 2 O 3 /γ-Al 2 O 3 was larger than that from 10CeO 2 /γ-Al 2 O 3 at more than ca. 350°C. On the other hand, the onset temperature of CO 2 generated from both the 10MO/γ-Al 2 O 3 samples in the TPO profiles was comparable to that in the TPD profiles, and the amounts of CO 2 generated from both the 10MO/γ-Al 2 O 3 in the TPO profiles were much larger than those in the TPD profiles. Especially, the CeO 2 loading intermittingly produced a quite large amount of CO 2 (more than 6000 ppm) at 150-250°C, which was overwhelmingly larger than that from 10Bi 2 O 3 /γ-Al 2 O 3 . The amount of CO 2 generated from both the 10MO/γ-Al 2 O 3 samples shifted to decrease with a further rise in temperature, and then CO 2 was hardly produced by 10CeO 2 /γ-Al 2 O 3 and 10Bi 2 O 3 /γ-Al 2 O 3 at more than 400°C and 460°C, respectively. Considering that the amount of CO 2 increased at lower temperatures with an increase in oxygen concentration in the flow gas, the MO loading promoted the production reaction of highmolecular-weight components (C x H y O z ) during the pretreatment to adsorb them on the 10MO/γ-Al 2 O 3 surface. In the TPD process, they were oxidized with oxygen adsorbates and/or lattice oxygen, which was quantitively limited on the surface, due to the lack of oxygen in the flow gas. The number of active oxygen adsorbates at lower temperatures was originally small, and thus only a small amount of CO 2 was generated at the lower temperatures. By contrast, many inactive oxygen adsorbates and/or lattice oxygen which originally existed became active at higher temperatures, and thus they oxidized the C x H y O z species, to produce the large amount of CO 2 despite no oxygen in the flow gas. On the other hand, in the TPO process, the active oxygen adsorbates at lower temperatures were constantly supplied from the flow gas, which seems to be the most important factor to generate the large amount of CO 2 only at lower temperatures. The mixture of oxygen in the flow gas increased the number of active oxygen adsorbates and/or lattice oxygen at various temperatures on the 10MO/γ-Al 2 O 3 surface, too, and thus the C x H y O z species had hardly remained after most of them reacted with the active oxygen adsorbates at lower temperatures.

1Pt/γ-Al 2 O 3
The Pt loading onto γ-Al 2 O 3 also drastically reduced the amounts of all reaction intermediates and perfectly eliminated the desorption of ethanol in the TPD profile. Besides, the generation temperature of these reaction intermediates from 1Pt/γ-Al 2 O 3 (the temperature at which the largest amounts of ethylene, acetaldehyde, and diethyl ether: ca. 310°C, ca. 400°C, and ca. 365°C, respectively) was hardly coincident with that of these reaction intermediates from 10MO/γ-Al 2 O 3 , which means that the adsorption/desorption properties of 1Pt/γ-Al 2 O 3 , as well as the catalytic combustion behavior, were largely different from those of 10MO/γ-Al 2 O 3 . The fact that MBR of 1Pt/γ-Al 2 O 3 at 150°C (0.8) is smaller than that of 10MO/γ-Al 2 O 3 (1.0) in their catalytic combustion behavior (see Table 1) promises more efficient production of higher-molecular-weight components on the 1Pt/γ-Al 2 O 3 surface at T L (150°C) than that on 10MO/ γ-Al 2 O 3 , which might increase the generation temperatures of the reaction intermediates on the 1Pt/γ-Al 2 O 3 surface. On the other hand, the TPO profiles showed that ethanol and three kinds of reaction intermediates were hardly generated in the whole temperature range.
The Pt loading onto γ-Al 2 O 3 also increased the amounts of CO 2 generated in the TPD and TPO profiles. The amount of CO 2 intermittingly generated from 1Pt/γ-Al 2 O 3 was smaller than that from 10MO/γ-Al 2 O 3 in the TPD profile, and the largest amount of CO 2 generated from 1Pt/γ-Al 2 O 3 in the temperature range of 300-370°C. On the other hand, the amount of CO 2 in the TPO profile was comparable to that in the TPD profile at low temperatures (less than 230°C), but it abruptly increased with a rise in temperatures at more than 230°C and reached the maximum (ca. 7000 ppm) at around 320°C. Namely, there is a large difference in the amount of CO 2 generated between TPD and TPO, even though there is little difference in the temperature dependence of CO 2 -generation behavior between TPD and TPO. This indicates that the larger number of high-molecular-weight components produced at T L (150°C) and adsorbed on the surface was oxidized under oxygen-richer gas flowing, but the quality of the reaction sites for catalytic combustion was not so much dependent on oxygen concentration in the flow gas at the slow rate of temperature rise (3°C min −1 ). A certain number of active reaction sites was formed on the surface of Pt and/or the boundary between Pt and γ-Al 2 O 3 , even under the oxygen-free gas flowing (i.e. under the TPD process) and the increase in the oxygen concentration in the flow gas probably incremented the number of active reaction sites on the 1Pt/γ-Al 2 O 3 surface.

1Pt/10MO/γ-Al 2 O 3
The co-loading of Pt with MO onto γ-Al 2 O 3 had a great influence on the TPD and TPO profiles of 1Pt/γ-Al 2 O 3 as well as 10MO/γ-Al 2 O 3 . The generation of ethanol, acetaldehyde, and diethyl ether from 1Pt/10CeO 2 /γ-Al 2 O 3 was hardly confirmed in the TPD profile. Ethylene was only generated from 1Pt/10CeO 2 /γ-Al 2 O 3 at around 300°C, which is almost the same temperature as that for 1Pt/γ-Al 2 O 3 . The amount of ethylene generated (up to 200 ppm) was a little larger than that from 1Pt/γ-Al 2 O 3 , whereas it was much smaller than that from γ-Al 2 O 3 . On the other hand, the amounts of acetaldehyde and diethyl ether from 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 (ca. 125 ppm at 375°C and ca. 220 ppm at 330°C, respectively) were larger than those from 1Pt/γ-Al 2 O 3 in the TPD profile. The generation temperature range was comparable to that from 1Pt/γ-Al 2 O 3 , which was higher than that from γ-Al 2 O 3 . Furthermore, the small amount of ethanol (ca. 30 ppm at 170°C) also desorbed in the lower temperature range (less than 220°C). Ethylene was hardly generated from 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 . These results indicate that the reaction intermediates adsorbed on the 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 surface were quite different from those of the 1Pt/10CeO 2 /γ-Al 2 O 3 surface. The CeO 2 loading probably enhanced the production and adsorption of high-molecular-weight components on the 1Pt/γ-Al 2 O 3 surface at pretreatment at T L (150°C), which were decomposed to form ethylene mainly at elevated temperatures. The Bi 2 O 3 loading seemed to enhance the production and adsorption of high-molecular-weight components on the 1Pt/γ-Al 2 O 3 surface during pretreatment at T L (150° C), which was decomposed only to form both acetaldehyde and diethyl ether at elevated temperatures. Nevertheless, the generation of ethanol, as well as three kinds of reaction intermediates from both 10MO/γ-Al 2 O 3 surfaces, was hardly confirmed in the TPO profiles, which means that these specific highmolecular-weight components adsorbed on the surfaces were totally oxidized under the oxygen-rich gas flowing (TPO process).
The co-loading of Pt with MO onto γ-Al 2 O 3 also increased the amounts of CO 2 generated from γ-Al 2 O 3 in the TPD profile. The CO 2 -generation behavior from 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 was comparable to that from 1Pt/γ-Al 2 O 3 , and the amount of CO 2 intermittingly generated at elevated temperatures was slightly larger than that from 1Pt/γ-Al 2 O 3 . The co-loading of Pt with CeO 2 reduced the CO 2 onset temperature and increased the amount of CO 2 generated in the temperature range of less than ca. 320°C (the largest amount of CO 2 generated: ca. 1000 ppm). However, the amounts of CO 2 generated from 1Pt/10MO/γ-Al 2 O 3 were much smaller than those from 10MO/γ-Al 2 O 3 , especially at lower temperatures (less than ca. 220°C) and higher temperatures (more than ca. 330°C). In the TPO profiles, on the other hand, the co-loading of Pt with MO reduced the CO 2 onset temperatures and decreased the amount of CO 2 generated from 1Pt/γ-Al 2 O 3 . The co-loading of Pt with CeO 2 was slightly effective in decreasing the CO 2 onset temperature than the co-loading of Pt with Bi 2 O 3 , while the amount of CO 2 generated from 1Pt/10CeO 2 /γ-Al 2 O 3 was smaller than that from 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 . Nevertheless, the CO 2 onset temperatures of 1Pt/10MO/γ-Al 2 O 3 were higher than those of 10MO/γ-Al 2 O 3 , and the amounts of CO 2 from 1Pt/10MO/γ-Al 2 O 3 were much smaller than those of 10MO/γ-Al 2 O 3 . The co-loading of 1Pt with 10CeO 2 obviously enhanced the catalytic oxidation activity of 1Pt/γ-Al 2 O 3 and 10CeO 2 /γ-Al 2 O 3 , but the co-loading of Pt with MO lowered their MBR at 150°C (ca. 0.7 for 1Pt/10CeO 2 /γ-Al 2 O 3 and ca. 0.2 for 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 (ca. 0.2)., as shown in Figure 2 and Table 1. The TPO profiles show that the number of the high-molecular-weight reaction intermediates adsorbed during the pretreatment at T L (150°C) on the 1Pt/10MO/γ-Al 2 O 3 surfaces was smaller than that on the surfaces of 10MO/γ-Al 2 O 3 (especially, 10CeO 2 /γ-Al 2 O 3 ). Besides, the loading of only Pt and the coloading of Pt with Bi 2 O 3 largely increased the amount of CO 2 generated at around 300 and 330°C, respectively, but the co-loading of Pt with CeO 2 only slightly enhanced the amount of CO 2 generated in lower temperature range, with a decrease in the oxygen concentration in the flow gas. These results indicate that the co-loading of Pt with CeO 2 is quite effective in totally oxidizing the high-molecular-weight reaction intermediates on the surface, even under the oxygen-poor gas flowing.

Effects of catalytic combustion behavior and adsorption/desorption properties on ethanol-sensing characteristics of adsorption/ combustion-type micro VOC sensors
Based on the catalytic combustion behavior and TPD and TPO properties of all samples in sections 3.2 and 3.3, the ethanol-sensing characteristics of adsorption/ combustion-type gas sensors are discussed in this section, along with the role of each of the sensing materials.

Validity of γ-Al 2 O 3 as a reference material
The catalytic activity of the γ-Al 2 O 3 surface was so low at T L (150°C) that only diethyl ether was slightly produced as a reaction intermediate, through dehydration condensation without the production of CO 2 . Ethanol and diethyl ether were adsorbed at 150°C and they were directly desorbed with a rise in temperature. Ethanol may be slightly oxidized to form high-molecular-weight intermediates, but their amounts can be vanishingly low, because the MBR at 150°C was almost 1.0. Besides, the rise in temperature of up to ca. 300°C generated ethylene and acetaldehyde with low active oxygen adsorbates on the γ-Al 2 O 3 surface and totally oxidized them (and/or the related adsorbates) to produce a small amount of CO 2 , according to TPD and TPO spectra. These results indicate that ethanol, diethyl ether, and the related adsorbates were not effectively oxidized on the γ-Al 2 O 3 surface at the initial stage of the pulse-driven heating from 150°C to 450°C (i.e. ca. several tens of milliseconds for the dynamic response). Namely, it is expected that endothermic process such as just desorption of the adsorbates is superior to exothermic process such as oxidation at the stage. Based on these discussions, γ-Al 2 O 3 is quite suitable as a reference material of adsorption/combustion-type gas sensors operable at 150°C as T L . On the other hand, the CO 2production ratio was almost 100% on the γ-Al 2 O 3 surface at higher than 300°C. Ethanol was totally oxidized at 450°C, which is a temperature value over an almost entire period of pulsed-driven heating, whereas the adsorption/combustion-type gas sensors employing γ-Al 2 O 3 as a reference material showed ethanol responses appropriate for the catalytic materials used (see Figure 1). Namely, the γ-Al 2 O 3 sufficiently worked as a reference material. We would like to discuss the differences in catalytic combustion behavior and adsorption/desorption (i.e. TPD and TPO) properties between γ-Al 2 O 3 and catalytic gassensing materials (Pt and/or MO-loaded γ-Al 2 O 3 ) in the following sections.

Effects of MO loading on ethanol-sensing properties
The MO loading onto γ-Al 2 O 3 certainly activated the catalytic activity of ethanol. On the other hand, the TPD profiles of 10MO/γ-Al 2 O 3 indicated that even the rate of temperature rise of 3°C min −1 was too fast to completely oxidize the high-molecularweight components adsorbed on the surface, only with oxygen adsorbates formed during the pretreatment. Namely, further enhanced oxidation activities over these oxide surfaces were essential in proceeding with the complete oxidation. The increase in oxygen concentration in the flow gas drastically reduced the generation temperature of the large amount of CO 2 , according to the TPO profile. This result also indicates that the complete-oxidation rate of the high-molecular-weight components on the 10MO/γ-Al 2 O 3 surface was quite slow. The intermitting CO 2 generation in the wide temperature range shows that various types of high-molecularweight components are largely produced and subsequently adsorbed on the surface. In the case of static catalytic combustion of ethanol over the 10MO/γ-Al 2 O 3 surface (see Figure 2); therefore, T 50 (E) and T 50 (CO 2 ) of 10MO/γ-Al 2 O 3 were quite high and the MBR at 150°C was about 1.0, while the high-molecular-weight components on the surface were decomposed to desorb as each of three kinds of reaction intermediates as well as CO 2 . On the other hand, the rate of abrupt temperature rise of adsorption/combustion-type gas sensors at pulsedriven heating (flash combustion) was roughly ca. 6 × 10 4°C min −1 . The MO loading improved the oxidation activity of γ-Al 2 O 3 , but the rate of complete oxidation of ethanol over 10MO/γ-Al 2 O 3 cannot follow the abrupt temperature rising. This is the main reason why the MO loading onto γ-Al 2 O 3 did not contribute to enhancing the dynamic oxidation of ethanol and the 10MO/γ-Al 2 O 3 sensor showed an extremely small dynamic response to ethanol. In addition, these sensor-signal profiles negatively shifted to a lower concentration of ethanol immediately after the dynamic response. The thermal conductivity of ceramic materials generally decreases with an increase in the mass of the constituent metal (M), lattice defects, grain boundaries, impurities, and so on [34]. Therefore, the MO loading is likely to cause the reduction in thermal conductivity of the sensing γ-Al 2 O 3 film. The balance between the thermal conductivity and the catalytic oxidation activity of 10MO/γ-Al 2 O 3 (gas-sensing material) and γ-Al 2 O 3 (reference) are quite important for determining the shift direction of the sensor-signal profiles.
These 10MO/γ-Al 2 O 3 sensors also showed a quite small static response to ethanol. This is probably because the rate of catalytic combustion of ethanol at 450°C directly influenced the magnitude of ethanol response. Namely, the rate of catalytic combustion of ethanol over 10MO/γ-Al 2 O 3 was only slightly faster than that over γ-Al 2 O 3 , considering that only ethylene generated over 10MO/γ-Al 2 O 3 was smaller than that over γ-Al 2 O 3 at 450°C. This slight difference in rates of catalytic combustion between 10MO/γ-Al 2 O 3 and γ-Al 2 O 3 determined the magnitude of static ethanol response.

Effects of Pt loading and Pt/MO co-loading on ethanol-sensing properties
The Pt loading onto γ-Al 2 O 3 largely improved the CO 2production ratio as well as ethanol conversion, and thus both T 50 (E) and T 50 (CO 2 ) shifted to lower temperatures. Therefore, the amounts of various adsorbates on the 1Pt/γ-Al 2 O 3 surface were relatively small, in comparison with that of on the 10MO/γ-Al 2 O 3 surface. However, the activity of the reaction sites for catalytic combustion was not so much dependent on oxygen concentration in the flow gas, because the 1Pt/γ-Al 2 O 3 has active oxygen species on the surface, which can effectively oxidize the relatively small amounts of ethanol, reaction intermediates, and high-molecularweight components adsorbed at 150°C, even under the oxygen-poor TPD conditions. These highly qualified catalytic combustion properties of the 1Pt/γ-Al 2 O 3 surface were the main reason why the magnitude of the dynamic response of the 1Pt/γ-Al 2 O 3 sensor was much larger than that of the 10MO/γ-Al 2 O 3 sensors. Besides, the 1Pt/γ-Al 2 O 3 sensor naturally shows a quite large static response, in comparison with that of the 10MO/γ-Al 2 O 3 sensors, because the catalytic combustion rate of ethanol over 1Pt/γ-Al 2 O 3 at 450°C was much larger than that over 10MO/γ-Al 2 O 3 .
Furthermore, the co-loading of Pt with MO onto γ-Al 2 O 3 modified the catalytic combustion behavior and adsorption/desorption properties of ethanol over 1Pt/γ-Al 2 O 3 and 10MO/γ-Al 2 O 3 , and then it made an important impact on their dynamic responses to ethanol. The co-loading of Pt with Bi 2 O 3 onto γ-Al 2 O 3 produced and then adsorbed the large amounts of relatively high-molecular-weight components on the surface at 150°C, in comparison with those produced on the surface of other samples, as expected from the small MBR (ca. 0.2, see Figure 2). Therefore, it is difficult to efficiently oxidize them on the 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 surface at the initial stage of the pulse-driven heating from 150°C to 450°C. This is the reason why the 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 sensor showed a relatively small and slow dynamic response to ethanol (see Figure 1). On the other hand, the coloading of Pt with CeO 2 onto γ-Al 2 O 3 showed the largest catalytic combustion properties, with relatively large MBR at 150°C (ca. 0.7, see Figure 2) and the small amounts of reaction intermediates. Besides, the TPD and TPO profiles (Figure 3(e) and 4(e)) show that the high-molecular-weight components adsorbed on the 1Pt/10CeO 2 /γ-Al 2 O 3 surface, which probably had relatively small molecular weight, are efficiently oxidized to form CO 2 at low temperatures in comparison with those on the 1Pt/γ-Al 2 O 3 and 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 surfaces. In addition, CeO 2 has a large oxygen storage/release capacity [35][36][37][38][39]. The considerable amounts of oxygen components are liberated from the bulk of CeO 2 with a rise in temperature, especially under oxygen-poor gaseous atmosphere, and thus they promoted the oxidation of various adsorbates on the surface, even at the quite fast flash heating (ca. 6 × 10 4 C min −1 ). They are the reasons why the 1Pt/10CeO 2 /γ-Al 2 O 3 sensor showed a quite large dynamic response.
On the other hand, the magnitude of static responses of the 1Pt/10CeO 2 /γ-Al 2 O 3 sensor was quite comparable with those of the 1Pt/γ-Al 2 O 3 and 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 sensors. The catalytic combustion behavior, as well as TPD and TPO profiles at 450°C, were quite similar between 1Pt/γ-Al 2 O 3 , 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 , and 1Pt/10CeO 2 /γ-Al 2 O 3 . These results strongly show that the rate of catalytic combustion of ethanol over their surfaces determines the total amount of heat on their surface at 450°C, which directly reflects the magnitude of static response.

Effects of T L and T H on the ethanol-sensing properties of 1Pt/10CeO 2 /γ-Al 2 O 3 sensor
Based on these results and discussion, the effects of T L and T H on the ethanol-sensing properties of the adsorption/combustion-type gas sensors are discussed in this section. Figure 5 shows the sensor-signal profiles of the 1Pt/10CeO 2 /γ-Al 2 O 3 sensor to ethanol under the operation with a mode of pulsed temperature heating of T H at 450°C after T L -operation at RT, 100, and 150°C and ethanol-concentration dependences of the various responses. The reduction in T L naturally had an insignificant effect on the magnitude of static responses. However, it was surprisingly ineffective in improving the magnitude of dynamic responses, too. The magnitude of dynamic response to 1000 ppm ethanol slightly increased with a rise in T L , while that to 100 ppm ethanol inversely decreased with a rise in T L . The T L dependences of its ΔV MAX and IDR of the sensor ( Figure 5(b) (i) and (ii)) obviously shows that the amounts of ethanol and/or various adsorbates containing reaction intermediates and high-molecular-weight components adsorbed on the surface are the important key in exhibiting the large responses. Therefore, the amount and kind of adsorbed components are expected to largely depend on T L , according to the catalytic combustion behavior (Figure 1). However, the actual amount of heat generated by the combustion of these adsorbates seems not to be much dependent on T L . The fact indicates that the amounts of various adsorbates at 150°C on the 1Pt/10CeO 2 /γ-Al 2 O 3 surface were almost comparable to that at RT, from the viewpoint of their combustion heat. The operation of T L at 150°C is discouraged in terms of energy saving. However, the relatively hightemperature T L operation above 100°C is generally quite effective in reducing the influence of moisture on the magnitude of dynamic responses of the sensor.
On the other hand, T H had a large influence on the magnitude of both dynamic and static responses, as we had expected. Figure 6 shows the sensor-signal profiles of the 1Pt/10CeO 2 /γ-Al 2 O 3 sensor to ethanol under the operation with a mode of pulsed temperature heating of T H at 250, 350, and 450°C after T Loperation at 150°C and ethanol-concentration dependences of the various responses. Firstly, the magnitude of static response reduced with a decrease in T H . This behavior is similar to that of general catalytic combustion-type gas sensors. Thus, the magnitude of static responses had been quite small, even though both the ethanol conversion and the production ratio of CO 2 at 250°C nearly were 100%. This behavior means that the reaction rate was slow and the amount of heat generation was small at 250°C in comparison with those at elevated temperatures. In contrast, the magnitude of the dynamic response of the 1Pt/10CeO 2 /γ-Al 2 O 3 sensor also reduced with a decrease in T H , but the sensor showed a relatively large dynamic response as low as 10 ppm ethanol, in comparison with the static response. This shows that a certain amount of various adsorbates at 150°C on the surface were efficiently oxidized on the surface even at the initial stage of the flash catalytic combustion process at 250°C and 350°C. The desorption (endothermic process) of reaction intermediates from the reference γ-Al 2 O 3 film is also likely to enhance the magnitude of dynamic responses, especially under the operation of T H at 250°C. As mentioned above, clarification of the catalytic combustion behavior and adsorption/desorption properties over gas-sensing materials are quite essential in understanding the fundamental gas-sensing characteristics of the adsorption/combustion-type gas sensors. Therefore, the detailed investigation on these properties of various target VOCs has recently been conducted in our group, and the compositional and microstructural optimization of the sensing film as well as the optimal setting of the dynamic operating condition have been attempted to detect various VOCs sensitively and selectively, on the basis of these properties.

Conclusions
Catalytic combustion behavior and adsorption/desorption properties of ethanol over these materials of ethanol over γ-Al 2 O 3 powders co-loaded with and without 1Pt and/or 10MO were investigated in detail and the ethanol-sensing characteristics of the adsorption/ combustion-type gas sensors employing these gassensing materials were discussed in this study. γ-Al 2 O 3 shows poor catalytic activity for ethanol oxidation, and thus three kinds of reaction intermediates (ethylene, acetaldehyde, and diethyl ether) were generated in large amounts. The MO loading slightly promoted the catalytic activity and reduced the amounts of reaction intermediates. However, it was not so effective in improving the rate of catalytic combustion of ethanol, and thus the 10MO/γ-Al 2 O 3 sensors showed a quite small magnitude of ethanol responses. On the other hand, the Pt loading drastically improved the catalytic activity and then accelerated the rate of catalytic combustion of ethanol. Therefore, the ethanol response of 1Pt/γ-Al 2 O 3 sensor was much larger than those of the 10MO/γ-Al 2 O 3 sensors. The loading of Pt with Bi 2 O 3 decreased MBR at 150°C and then the produced highmolecular-weight components on the surface are not easily oxidized at the initial stage of the pulse-driven heating. Therefore, the 1Pt/10Bi 2 O 3 /γ-Al 2 O 3 sensor showed a relatively small and slow dynamic response to ethanol. 1Pt/10CeO 2 /γ-Al 2 O 3 showed the largest catalytic activity. The amounts of adsorbates on the surface were the smallest among all samples. However, they were efficiently oxidized on the surface to form CO 2 , probably because of the large oxygen-release capacity of CeO 2 even at quite fast flash heating. They are the reasons why the 1Pt/10CeO 2 /γ-Al 2 O 3 sensor showed the largest dynamic response to ethanol. The magnitude of static responses of the 1Pt/γ-Al 2 O 3 and 1Pt/10MO/γ-Al 2 O 3 sensors is independent of the kind of loading species, because of their comparable catalytic activities at 450°C. The effects of T L and T H on the ethanol-sensing properties of the 1Pt/10CeO 2 /γ-Al 2 O 3 sensor were also explained well enough, on the basis of the catalytic activity for ethanol oxidation and TPD/TPO properties.