Microelectronics Reliability 48 (2008) 1772–1779 Contents lists available at ScienceDirect Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel A low power MEMS gas sensor based on nanocrystalline ZnO thin films for sensing methane P. Bhattacharyya a, P.K. Basu b, B. Mondal b, H. Saha b,* a b Department of Electronics and Telecommunication Engineering, Bengal Engineering and Science University, Shibpur, Howrah 711103, India IC Design and Fabrication Center, Department of Electronics and Telecommunication Engineering, Jadavpur University, S.c. Mallik Road, Kolkata 700032, India a r t i c l e i n f o Article history: Received 13 March 2008 Received in revised form 20 May 2008 Available online 30 October 2008 a b s t r a c t Nanocrystalline ZnO based sensor using micromachined silicon substrate has been reported for efficient detection of methane as opposed to conventional SnO2 based micromachined sensors for its higher compatibility to silicon IC technology and greater response. A suitably designed nickel microheater has been fabricated on to the micromachined Si platform. The optimum temperature for highest response magnitude and lowest response time were found to be 250 °C although relatively high (76.6%) response is obtained even at as low as 150 °C. Our study showed quite high response magnitude (87.3%), appreciably fast response time (8.3 s) and recovery time (17.8 s) to 1.0% methane at 250 °C. The sensor showed appreciably fast response (14.3 s) and recovery time (28.7 s) at 150 °C. The power consumption at an operating temperature of 250 °C was 120 mW and at 150 °C is only 70 mW. Moreover, this type of sensor was found to give fairly appreciable response for lower methane concentrations (0.01%) also. For higher methane concentrations (>0.5%) response is detectable even at 100 °C where the power consumption is only 40 mW. Ó 2008 Published by Elsevier Ltd. 1. Introduction Conventional metal oxide gas sensors, mostly alumina substrate based, which are commonly used for sensing inflammable hydrocarbon gases (like CH4) [1] and other toxic gases (like CO) [2] often suffer from the two principal limitations, viz. (a) their relatively high operating temperature (P300 °C) [3] and (b) large power dissipation (0.5–1 W) [4]. Both these features are unacceptable for continuous gas monitoring in many environmental scenario such as underground coalmines. CMOS or CMOS–MEMS technology has been presently employed in sensor technology in a multitude of applications for miniaturization of the devices, low power consumption, faster sensor response, batch fabrication at industrial standards, low cost and moreover greater sensitivity [5–9]. On the other hand use of nanomaterials for the sensing device for their enormously increased surface to volume ratio compared to their bulk counterpart leads to opportunities to lower the operating temperature of metal oxide semiconductor gas sensors [10]. In our earlier publication [11] we also found that use of sol–gel derived nanocrystalline ZnO sensing film along with Pd (Ag) catalytic contact are particularly suitable for sensing methane at moderate temperature (150–250 °C) with high response magnitude (74%) and appreciably fast response time (16 s). * Corresponding author. Tel./fax: +91 3324146217. E-mail address: sahahiranmay@yahoo.com (H. Saha). 0026-2714/$ - see front matter Ó 2008 Published by Elsevier Ltd. doi:10.1016/j.microrel.2008.07.063 With a view to integrate MEMS technology along with nanodimensional sensing film Fau et al. [12] has recently published a report on nanosized SnO2 sensitive layer on a silicon platform for sensing CH4 and CO where the minimum methane concentration measured was 200 ppm. Mitzner [13] and others have reported on methane sensing using ZnO but with a prolonged recovery. Methane sensing at an optimum temperature of 450 °C, using a catalytic thick film/SnO2 thin film has been reported by Suzuki et al. [14]. Cane [15] and his coauthors worked on detection of NO2, CO and Toluene by SnO2 based micromachined gas sensor arrays. However, very few reports have so far been published on sensing using nano-ZnO thin film on micromachined silicon substrate [16,17]. Most of the papers, on micromachined microheaters, reported so far deal with the design of either platinum [18] or polysilicon microheater [19], particularly suitable for the higher temperature range (400–700 °C) [20] for improved high temperature stability and the CMOS process compatibility. Most of the platforms, for micromachined gas sensor, reported so far are based on a SiO2 and Si3N4 composite layer on a thick (400 lm) silicon substrate [21] with a view to reduce power consumption. Gotz et al. [22] first proposed a thin Si plug underneath the dielectric membrane, for achieving uniform temperature distribution over the active heater area owing to the high thermal conductivity of Si. Recently several works on relatively low temperature gas sensors, using several nanotextured semiconducting oxides, have been reported [11,23]. This type of relatively low temperature (150–250 °C) gas sensors P. Bhattacharyya et al. / Microelectronics Reliability 48 (2008) 1772–1779 1773 [11,23] do not require an expensive Pt or poly-Si microheater; some low cost heating elements (such as nickel) may efficiently serve in this temperature range. The present work is motivated to solve the problems of elevated operating temperature and high power consumption by taking a twin approach of (a) reducing the operating temperature through the use of nanocrystalline metal oxide (ZnO) and (b) reducing the power dissipation through the deployment of MEMS structure with thin membrane for active layer (ZnO) deposition. In this paper we reported on nanocrystalline ZnO based sensor using micromachined silicon substrate for efficient detection of lower concentrations (0.01%) of methane. Also, for the first time, a nickel microheater has been designed and fabricated accordingly to maintain the operating temperature of the sensing layer. The detail design and characterization of this nickel microheater has been discussed. The response magnitude and response time of the nickel microheater based nano-ZnO methane gas sensors has been determined at different operating temperatures in the range of 100– 300 °C for different methane concentrations ranging from 0% to 1.0%. The power consumption of the sensor is also established. 2. Experimental A schematic drawing with requisite dimensions of fabricated sensor structure is shown in Fig. 1(a) and the corresponding process flowchart is shown in Fig. 1(b). The starting wafer was p-Si <1 0 0> of resistivity 1 X cm (100 lm thick), over which a thermal insulating SiO2 membrane (0.8 lm) was grown by thermal oxidation. After opening window by lithographic technique backside bulk micromachining was carried out with EDP (ethylene diamine pyrocatechol) solution at a temperature of 85 °C, resulting into a 20 lm Si membrane. The SEM image of the backside of membrane is shown in the Fig. 2. Most of the paper reported so far uses a freestanding dielectric (SiO2 or SiO2–Si3N4 composite layer) [24,25] for lower power consumption and higher operating temperature at the cost of long-term stability. The use of a backside Si plug provides better temperature homogeneity along the active layer [22]. In our work a 20 lm Si membrane has been used for better mechanical stability along with fairly good temperature uniformity. A backside silicon oxide layer (0.8 lm) was grown in the next step. Purpose of this backside oxidation is to reduce the thermal losses from the backside of the membrane. A 0.2 lm nickel layer was vacuum deposited on the top side of SiO2 covering entire front side of the sample. The microheater was then fabricated using conventional lithography followed by nickel etch-back technique. A 0.6 lm SiO2layer, acting as an electrical isolation between the heater and the active layer, was then deposited on Ni microheater by sol–gel method. The active area has a dimension of 1.3 mm  1.3 mm at the center of a 3 mm  3 mm membrane. The total sensor area is 4 mm  4 mm. The lines of meander shaped microheater were 50 lm wide and were separated also by 50 lm. Fig. 3 shows the SEM image of the fabricated microheater. The fabricated heater resistance was about 167 X. Thermo-electrical simulations using coupled field 3D model of commercial finite Fig. 1(b). Process flow chart for the fabrication of nanocrystalline ZnO based micromachined methene sensor with embedded Ni microheater. Fig. 2. SEM of the backside micromachined Si substrate. element model (FEM) program ANSYS 10.0 has been employed. We consider conduction heat losses through the membrane and through air as the primary component. In the desired temperature range (150–300 °C) radiation losses are not significant. The active ZnO layer was deposited by sol–gel method by spin coating technique. The rotation speed of the coating unit was 1000 rpm and the duration of the single coating was 25 s. Then the samples were Fig. 1(a). Schematic of the MEMS gas sensor structure with requisite dimensions (not to scale). 1774 P. Bhattacharyya et al. / Microelectronics Reliability 48 (2008) 1772–1779 repeated for three times and a ZnO film of 900 nm thickness with the particle size ranging from 45 nm to 75 nm and average pore diameter of 56 nm was produced (Fig. 4). X-ray diffraction (XRD) measurements and the scanning electron microscopy (SEM) analysis were done and the detail of that structural analysis of ZnO layer has been reported earlier [11]. Pd–Ag (26%) catalytic contact [11] was deposited on ZnO by an e-beam deposition method (106 mbar) using Al metal masks. The complete three-dimensional cross-sectional view of nanocrystalline ZnO based MEMS methane sensor is shown in the Fig. 5(a) whereas Fig. 5(b) shows the photograph of nanocrystalline ZnO based MEMS methane sensor mounted on a sensor platform. For studying the performance of the sensor, high purity (100%) methane gas and high purity (99.99%) N2 in desired proportions were allowed to flow to the gas-sensing chamber through a mixing path via an Alicat Scientific mass flow controller and a mass flow meter for keeping the mass flow rate and thus the concentration of the methane gas constant throughout the experiments. The gas pressure over the sensor device was 1 atm during the experiments. The resistance of the sensors in the presence and absence of CH4 was measured by a Keithley 6487 picoammeter/voltage source. Fig. 3. SEM of fabricated Nickel microheater. 3. Sensing mechanism A possible mechanism for methane sensing in the present study is depicted as follows. It is known that addition of small amounts of noble metal, e.g. Pd or Pt to the metal oxide can promote not only Fig. 4. SEM images of the nanocrystalline ZnO surface. heated at 110 °C for 10 min to evaporate the solvent and to remove organic residuals. Finally the samples were annealed at 350 °C for producing nanocrystalline ZnO for 30 min. The entire process was Fig. 5(b). Photograph of nanocrystalline ZnO based MEMS methane sensor mounted on a sensor platform. Fig. 5(a). Three-dimensional cross-sectional view of nanocrystalline ZnO based MEMS methane sensor (not to scale). 1775 P. Bhattacharyya et al. / Microelectronics Reliability 48 (2008) 1772–1779 300 250 o Temperatue (in C) the gas sensitivity but also the rate of response to a great extent due to the catalytic activities of these metals for oxidation of inflammable gases like methane. According to Yamazoe [26] Pd and Ag act as electronic sensitizer as they affect the work function of SnO2 during gas sensing. Following the same arguments we can envisage that the work function of ZnO increase in presence of Pd and Ag in air thereby widening the depletion region on ZnO–Pd (Ag) interface as indicated in Fig. 11(a). However this shift vanishes when the samples are exposed to methane or similar reducing atmosphere. Fig. 11(b) clearly depicts that there is a change in the ZnO Fermi level in presence of hydrogen thereby reducing the band bending facilitating the electron transport and increase in conductivity. As reported by Kohl [3] methane dissociates to a methyl group and hydrogen followed by combining the adsorbed hydrogen atoms producing hydrogen molecule as shown below: 200 150 100 Simulation Experimental With backside SiO2 layer 50 0 1000 CH4gas $ CH3ads þ Hads Hads þ Hads H2 The hydrogen then reacts with the adsorbed oxygen on the Pd (Ag) surface [shown in Fig. 11] and produces H2O. The adsorbed organic radicals undergo a chain of reactions and finally produce CO2as the other product. 2000 3000 4000 Distance (in micrometer) Fig. 6(b). The simulated and experimental temperature distribution profile along the membrane for an input power of 150 mW (5 V) with 20 lm silicon membrane with backside 0.8 lm Silicon dioxide layer. 400 o Fig. 6(a) shows the simulated (ANSYS 10.0) temperature distribution along the membrane with that of experimental one in case of the heater structure shown in Fig. 3 for an input power of 150 mW and with backside 0.8 micron oxide layer. The purpose of this backside oxidation is to reduce the thermal loss from the backside of the membrane. The power consumption of the nickel microheater obtained from the FEM simulations has been compared with experimental results in Fig. 6(b). Clearly at the optimum operating temperature of 250 °C around the active area the corresponding power consumption is 120 mW while at 100 °C (for application in underground mining environment such low operating temperature is extremely desirable) power is only 43 mW. The experimental results are in good agreement with the simulated one within a satisfactory error limit of ±5%. The response magnitude of nanocrystalline ZnO thin film based MEMS sensor at five different concentrations of methane (0.01%, Temperature (in C) 4. Results and discussions 300 200 Simulation Experimental 100 50 100 150 200 250 Power(in mW) Fig. 6(c). The temperature vs the power plots for both the simulated and experimental data for the nanocrystalline ZnO based micromachined gas sensor. 0.05%, 0.1%, 0.5% and 1.0%) was recorded at different operating temperatures (100, 150, 200, 250 and 300 °C). The response magnitude S, is expressed in terms of sensor resistance in air (Ra) and in test gas (Rg) as follows [11]: S ¼ ðRa  Rg Þ=Ra Fig. 6(a). Simulated (ANSYS 10.0) temperature distribution over the membrane for an input power of 150 mW. We calculated the response time as the time to reach the response magnitude to 67% of the saturation value [11]. After the gas pulse was cut off the recovery time was calculated as 67% of the time to fall the sensor response from the saturation value to the baseline value. Fig. 7 shows the variation of response magnitude as a function of temperature (The operating temperature indicated throughout the text is the sensor’s local temperature) for different methane concentrations. It is clear from the figure that temperature for maximum response for all methane concentrations is 250 °C. Another important observation was that though maximum response is obtained at 250 °C for all methane concentrations, the sensor shows quite moderate response (44%) and response time (42 s) even at 100 °C, particularly for higher methane concentrations (>0.1%). In the environment where measurement of somewhat higher methane concentrations are required (e.g. mining environment, alarm is set at 0.1%) lower temperature (100 °C) operation is preferable and in this sensing range power consumption is 1776 P. Bhattacharyya et al. / Microelectronics Reliability 48 (2008) 1772–1779 with increased kinetic energy of the molecules hitting the sensor’s surface and the desorption of the gas molecules from the surface. Fig. 8 shows the variation of response as a function of methane concentration for operating temperatures ranging from 100 °C to 300 °C. It is apparent from the figure that there is a trend of increasing response upon exposure to higher methane concentrations eventually reaching saturation. For the temperatures ranging from 150 °C to 300 °C for concentrations higher than 0.05% the response is quite high and varies linearly with methane concentrations whereas for 100 °C the curve is almost linearly increasing 100 Response (%) 80 60 40 1% Methane 0.5% Methane 0.1% Methane 0.05% Methane 0.01% Methane 20 100 0 100 150 200 250 300 80 Response (%) Temperature (˚C ) Fig. 7. Response magnitude as a function of temperature for five different methane concentrations. small. The sensors show the maximum response at 250 °C. This can be explained following the model by Yamazoe et al. [26]. The rate of adsorption (and therefore sticking coefficient) of methane molecules on the nano-ZnO surface increases with increasing temperature and ultimately reaches a peak at a particular temperature. On the other hand desorption rate of the gas molecules also increases with increasing temperature. The optimum temperature for the maximum response with fastest response time may be the result of the competition between the increased adsorption probability 40 100°C 150°C 200°C 250°C 300°C 20 0 0.0 0.2 0.4 0.6 1.0 Fig. 8. Response magnitude as a function of methane concentrations at different operating temperatures. 60 Resistance(in KOhm) 60 0.01% 50 0.05% 40 0.1% 30 0.5% 150˚C 20 50 40 30 0.01% 0.05% 20 0.1% 250˚C 10 0.5% 1.0% 10 1.0% 0 0 2 4 6 8 0 1 2 3 4 5 Time(x100s) Time(x100s) 60 50 Resistance(in KOhm) Resistance(in KOhm) 0.8 Methane concentrations (%) 70 Resistance(in KOhm) 60 50 40 0.01% 30 0.05% 20 0.1% 200˚C 0.01% 0.05% 20 0.1% 300˚C 2 3 4 Time(x100s) 5 6 0.5% 10 1.0% 1 30 0.5% 10 0 40 7 1.0% 0 1 2 3 4 5 6 Time(x100s) Fig. 9. Transient response characteristics with five different methane concentrations at (a) 150 °C (b) 200 °C (c) 250 °C (d) 300 °C. 7 P. Bhattacharyya et al. / Microelectronics Reliability 48 (2008) 1772–1779 in nature. The same adsorption–desorption kinetics play the primary role with the sticking coefficient initially increasing with increasing CH4 concentration till it reaches almost saturation at a very high methane concentrations. The transient response behavior of the sensor structures in 0.01%, 0.05%, 0.1%, and 0.5% and in 1% methane at 150 °C, 200 °C, 250 °C and 350 °C are shown in Fig. 9a–d respectively. Upon exposure to methane the sensor resistance initially decreased due to release of free electrons (as indicated in Section 3) and then got saturated while on cutting off the methane supply the resistance increased and returned almost to its baseline value. Fig. 6 shows that they do not reach exactly the baseline value possibly because of some gas molecules remaining adsorbed on the sensor surface. The calculated response magnitude, response time and recovery time for different gas concentrations are plotted in Fig. 10a–c and in the form of bar chart. 1777 The response, response time and recovery time are functions of methane concentrations. Increase in methane concentration provides more methane molecules to be adsorbed on the oxide surface Fig. 11(a). Sensitization mechanism of Pd–Ag/ZnO sensor. Fig. 10. (a) Response as a function of gas concentration for five different operating temperatures. (b) Response time as a function of gas concentration for five different operating temperatures. (c) Recovery time as a function of gas concentration for five different operating temperatures. 1778 P. Bhattacharyya et al. / Microelectronics Reliability 48 (2008) 1772–1779 per unit time thereby favoring the fast electron transport kinetics as a result of which the response increases and response time decreases. On the other hand the recovery time is higher due to slow desorption kinetics of the methane gas from the interface at higher concentration. Table 1 shows the response, response time and recovery time of the MEMS methane sensor at different methane concentrations and at different operating temperatures. With 1% methane the maximum sensitivity of 87.3% was obtained at 250 °C, whereas minimum response and recovery time were found to be 8.3 s (1% methane) and 11.7 s (0.01% methane) respectively. The cross sensitivity of the ZnO based resistive sensor structure with Pd–Ag contact was checked by flowing H2 and CH4in separate experiments. The results are shown in Fig. 12. It is clear from the figure that the maximum response for H2 (95%) is at 100 °C where methane response is 44%. On the other hand, for methane, the maximum response (87%) occurs at 250 °C where the hydrogen response value is only 50%. 5. Summary and conclusion Fig. 11(b). Schematic band diagrams of Pd–Ag/ZnO in oxidized state and in reduced state. Table 1 Response magnitude, response time and recovery time at five different gas concentrations and at different operating temperatures of nano-ZnO based MEMS methane sensor Temperature (°C) Response magnitude (%) 100 150 200 250 300 0.01% 0.05% 0.1% 0.5% 1% 4.2 12.4 34.8 44.4 40.1 7.3 27.9 47.8 57.1 52.2 12.3 52.8 61.1 68.4 57.9 Recovery time (s) 26.8 65.2 69.5 75.7 64.3 44.5 76.6 81.1 87. 3 74.4 Response time (s) 100 150 200 250 300 0.01% 0.05% 0.1% 0.5% 1% 0.01% 0.5% 0.1% 0.05% 1% 145.3 124.5 86.7 63.6 71.0 128.0 86.4 65.3 43.8 49.8 62.3 33.7 24.8 21. 2 26.8 50.6 26.4 18.5 16. 4 22.2 42.6 14.3 11.8 8. 3 18.5 35.5 28.7 25.5 11.7 14.3 41.2 31.1 27.6 15.2 19.9 46.7 34.0 29.1 17. 8 24.8 54.4 37.5 32.1 27.7 36.8 76.6 42.8 38.9 34. 8 42.1 100 Acknowledgements 80 Response (%) Nanocrystalline ZnO based MEMS micro-hotplate (with Ni microheater) provides a promising platform for low power sensors for methane sensing even at a lower ppm. The Ni microheater was fabricated using conventional lithography followed by nickel etchback technique. The active area was having a dimension of 1.3 mm  1.3 mm at the center of a 3 mm  3 mm membrane. The total sensor area was 4 mm  4 mm. The lines of meander shaped microheater were 50 lm wide and were separated also by 50 lm. The response magnitude at an optimum temperature of 250 °C was found to be 87% with 1.0% methane with a very fast response time (8 s) and moderately fast recovery time (35 s). Our MEMS sensor showed appreciable response magnitude of 44%, for same methane concentration, at even 100 °C although the response (response time 42 s) is rather sluggish. For a methane concentration as low as 0.01%, the sensor showed almost 44.4% response at 250 °C. The power consumption at optimum temperature (250 °C) was found to be 120 mW and at 150 °C it was 70 mW only. Such low power sensors are highly suitable for underground coalmine environment where continuous monitoring of hazardous gases is required. Moreover, as the microheater is driven by 5 V supply, due to its CMOS compatibility entire signal processing unit can be integrated on a single chip having opportunity for remote communication with the distant control room. Though an optimum operating temperature of 250 °C has been achieved, for ensuring safety in underground coalmine environment, further reduction of operating temperature (6100 °C) is extremely desirable. Through the modification nano-ZnO structure and suitable doping by catalytic metals, the above requirement may be fulfilled. Integration of nano-ZnO thin film along with MEMS technology seems to provide future platform for low power low temperature methane sensor with appreciable response magnitude and response time. Further work is in progress regarding the selectivity in presence of CO and other hydrocarbons and long-term stability of the sensor. P.K. Basu gratefully acknowledges DST, India for a Junior Research Fellowship (JRF). The work has been carried out under the research project sponsored by DST, Government of India. The authors are thankful to Prof. S. Basu for his keen interest and active help in the work. 60 40 CH4 H2 At 1% 20 50 100 150 200 References 250 300 Temperature (˚C) Fig. 12. Selectivity study with methane and hydrogen at different temperatures with 1% gas concentration. [1] Chatterjee K, Chatterjee S, Banerjee A, Raut M, Pal NC, Sen A, et al. The effect of palladium incorporation on methane sensitivity of antimony doped tin oxide. Mater Chem Phys 2003;81:33–8. [2] Hahn SH, Barsan N, Weimar U. Investigation of CO/CH4 mixture measured with differently doped SnO2 sensors. Sensors Actuators B 2001;78:64–8. [3] Kohl D. Function and application of gas sensors. J Phys D: Appl Phys 2001;34:R125–49. P. Bhattacharyya et al. / Microelectronics Reliability 48 (2008) 1772–1779 [4] Figaro Products Catalogue, Figaro gas sensors 2000-series, Figaro Engineering Inc., European Office, Oststrasse 10, 40211 Dusseldorf, Germany. [5] Mitzner KD, Strnhagen J, Glipeau DN. Development of micromachined hazardous gas sensor array. Sensors Actuators B 2003;93:92–9. [6] Suehle JS, Cavicchi RE, Gaitan M, Semancik S. Tin oxide gas sensor fabricated using CMOS micro-hotplates and in situ processing. IEEE Electron Dev Lett 1993;14(3):118–20. [7] S, Cavicchi RE, Gaitan M, Suehle JS, US Patent 1994;5:345,213.. [8] Salcedo JA, Juin JL, Afridi Muhammad Y, Hefner Allen R. On-chip electrostatic discharge protection for CMOS gas sensor systems-on-a-chip (SoC). Microelectron Reliab 2006;46:1285–94. [9] Lafontan X, Pressecq F, Beaudoin F, Rigo S, Dardalhon M, Roux JL, et al. The advent of MEMS in space. Microelectron Reliab 2003;43(7):1061–83. [10] Comini E, Faglia G, Sberveglieri G, Pan Z, Wag ZL. Stable and highly sensitive gas sensors based on semiconducting oxide nanobelts. Appl Phys Lett 2002;81:1869–71. [11] Bhattacharyya P, Basu PK, Saha H, Basu S. Fast response methane sensor using nanocrystalline zinc oxide thin films derived by sol–gel method. Sensors Actuators B 2007;124:62–7. [12] Fau P, Sauvan M, Trautweiler S, Nayrel C, Erades L, Maisonat A, et al. Nanosized tinoxide sensitive layer on a silicon platform for domestic gas application. Sensors Actuators B 2001;78:83–8. [13] Mitzner KD, Sternhagen J, Galipeau DW. Development of a micromachined hazardous gas sensor array. Sensors Actuators B 2003;93:92–9. [14] Suzuki T, Kunihare K, Kobayashi M, Tabata S, Higaki K, Ohnishi H. A micromachined gas sensor based on a catalytic thick film/SnO2thin film bilayer and thin film heater Part 1: CH4 sensing. Sensors Actuators B 2005;109:185–9. [15] Cane C, Gracia I, Gotz A, Fonseca L, Tamay EL, Horrillo MC, et al. Detection of gases with arrays of micromachined tin oxide gas sensors. Sensors Actuators B 2000;65:244–6. 1779 [16] Hassani F, Tigli O, Ahmadi S, Korman C, Zaghloul. Integrated CMOS surface acoustic wave gas sensor: design and characteristics. Proc IEEE Sensors 2003;2:1199–202. [17] Wan Q, Li QH, Chen YJ, Wang TH, He XL, Li JP, et al. Fabrication and ethanol sensing characteristics of ZnO nanowire gas sensors. Appl Phys Lett 2004;84:3654–6. [18] Lee SM, Dyer DC, Gardner JW. Design and optimisation of a high-temperature silicon micro-hotplate for nanoporous palladium pellistors. Microelectron J 2003;43:115–26. [19] Puigcrobe J, Vogel D, Michel B, Vila A, Gracia I, Cane C, et al. Thermal and mechanical analysis of micromachined gas sensors. J Micromech Microeng 2003;13:548–56. [20] Chung W, Shim C, Choi S, Lee D. Tin oxide microsensor for LPG monitoring. Sensors Actuators B 1994;20:139–43. [21] Rossi C, Scheid E, Esteve D. Theoretical and experimental study of silicon micromachined microheater with dielectric stacked membranes. Sensors Actuators A 1997;63:183–9. [22] Gotz A, Gracia I, Cane C, Lora-Tamayo E, Horrilo MC, Getino J, et al. A micromachined solid state integrated gas sensor for the detection of aromatic hydrocarbons. Sensors Actuators B 1997;44:483–7. [23] Nunes P, Fortunato E, Lopes A, Martins R. Influence of the deposition conditions on the gas sensitivity of zinc oxide thin films deposited by spray pyrolysis. Int J Inorg Mater 2001;3:1129–31. [24] Sheng LY, Tang Z, Wu JP, Chan CH, Sin JKO. A low-power CMOS compatible integrated gas sensor using maskless tin oxide sputtering. Sensors Actuators B 1998;49:81–7. [25] Rossi C, Boyer PT, Esteve D. Realization and performance of thin SiO2/Si3Nx membrane for microheater applications. Sensors Actuators A 1998;64:241–5. [26] Yamazoe N, Fuchigama J, Kishikawa M, Seiyama T. Interactions of tin oxide surface with O2, H2O and H2. Surf Sci 1979;86:335–44.