sensors Article
Planar Indium Tin Oxide Heater for Improved Thermal Distribution for Metal Oxide Micromachined Gas Sensors M. Cihan Çakır 1,2, *, Deniz Çalı¸skan 1 , Bayram Bütün 1 and Ekmel Özbay 1,3 1 2 3
*
Nanotechnology Research Center, Bilkent University, Ankara 06800, Turkey; [email protected] (D.Ç.); [email protected] (B.B.); [email protected] (E.Ö.) Department of Nanotechnology and Nanomedicine, Hacettepe University, Ankara 06800, Turkey Department of Electrical and Electronics Engineering, Department of Physics, Bilkent University, Ankara 06800, Turkey Correspondence: [email protected]; Tel.: +90-312-290-3050
Academic Editors: Sang Sub Kim and Hyoun Woo Kim Received: 18 July 2016; Accepted: 23 September 2016; Published: 29 September 2016
Abstract: Metal oxide gas sensors with integrated micro-hotplate structures are widely used in the industry and they are still being investigated and developed. Metal oxide gas sensors have the advantage of being sensitive to a wide range of organic and inorganic volatile compounds, although they lack selectivity. To introduce selectivity, the operating temperature of a single sensor is swept, and the measurements are fed to a discriminating algorithm. The efficiency of those data processing methods strongly depends on temperature uniformity across the active area of the sensor. To achieve this, hot plate structures with complex resistor geometries have been designed and additional heat-spreading structures have been introduced. In this work we designed and fabricated a metal oxide gas sensor integrated with a simple square planar indium tin oxide (ITO) heating element, by using conventional micromachining and thin-film deposition techniques. Power consumption–dependent surface temperature measurements were performed. A 420 ◦ C working temperature was achieved at 120 mW power consumption. Temperature distribution uniformity was measured and a 17 ◦ C difference between the hottest and the coldest points of the sensor at an operating temperature of 290 ◦ C was achieved. Transient heat-up and cool-down cycle durations are measured as 40 ms and 20 ms, respectively. Keywords: Indium tin oxide; Metal oxide gas sensor; Micro hot-plate; SnO2 ; Heat distribution
1. Introduction Metal oxides are widely used in gas sensing applications, such as chemo-resistive [1] and optoelectronic sensors [2,3]. Metal oxide–resistive gas sensors are being investigated, developed, and used due to their simplicity, production flexibility, low cost and sensitivity to a wide variety of volatiles [1]. Due to increasing interest in those sensors, micro-hotplate structures are being developed with enhanced device performance, decreased power consumption, and low cost [1,4]. Research activities aimed at enhancing the sensitivity and selectivity as well as lowering the response time and power consumption are still under active research. Those research studies present innovative integrated micro-hotplate technologies based on Micro Electro Mechanical Systems (MEMS) to achieve optimized thermal properties such as low power consumption and good temperature uniformity across the active layer [4]. With the development of MEMS technologies in the 1990s, new generations of gas sensors with integrated micro-hotplates were realized. Micromachining processes allow for the precise modification of the thermal characteristics of those devices. The optimization of device geometry, membrane, Sensors 2016, 16, 1612; doi:10.3390/s16101612
www.mdpi.com/journal/sensors
Sensors 2016, 16, 1612
2 of 7
and resistor materials affects the heat transfer mechanisms, resulting in the enhancement of the thermal dispersion and Joule heating characteristics of the devices. Besides these enhancements, the sensor’s heat capacity reduction, which is achieved by thermal mass reduction with the help of micromachining methods, also reduces the power consumption and improves the response time of the sensors [5,6]. By improving the response time, selectivity could be enhanced by rapid temperature cycle measurements to produce additional information by getting the response signature as a variable of temperature pulses [7–9]. Metal oxide gas sensors are based on the surface reactions between the target gas species and the sensing metal oxide film. As a result of the surface reactions, gas molecules interact with the film surface and dope it, the surface band bending alters and then the metal oxide layer resistivity changes. Metal oxide sensors give a response to the wide variety of organic and inorganic gases. However, this wide range has a disadvantage, which is nonselectivity. This disadvantage is eliminated once a temperature-dependent measurement is used. When gas species are introduced to the film surface, the total amount of surface reactions, i.e., the adsorbed and desorbed amount of gas molecules, is correlated with the resistance change of the thin film, and it is temperature-dependent and specific to the gas compound type. The base conductance of the sensing film can be defined as the conductance in the absence of the target gas molecules at a given temperature, and it is also strongly dependent on the temperature of the sensing film [6]. Different metal oxides react with the same analytes with different efficiencies at different temperatures [6,10,11]. In other words, the absolute resistivity change caused by the same concentration of the target gas can reach its peak value at different temperatures for different metal oxide films. This information can be used to obtain more selective data extraction from metal oxide gas sensors. For a given temperature-dependent measurement, the nonuniformity of the surface temperature distribution means that the total sensor signal will be an integration of signals obtained from metal oxide sites at different temperatures. Larger nonuniformity in temperature means that the data will be blurred and the peaks in the resistance change with temperature changes will be broadened. To get a characteristic temperature-dependent sensor signal at a high resolution, the temperature distribution on the active sensing film region has to be as uniform as possible [12–14]. To reach good temperature uniformity, power consumption, and working temperature goal by using conventional metal and metal alloy resistor materials, the hotplates with heat-spreading plates, micro-bridges, and resistors with complex geometries based on meanders were designed and fabricated [4–6,14,15]. 2. Design and Fabrication In this work, a planar resistor as a heating element is introduced which can provide high temperature uniformity across the active region of the thin-film metal oxide sensors. A thin sputtered indium tin oxide (ITO) film is used as the resistor material, with simple planar resistor geometry. Silicon nitride (Si3 N4 ) is used for both the membrane material and also as the passivation and insulation layers. Conventional bulk silicon micromachining methods such as potassium hydroxide (KOH) etching are used to obtain the micro-hotplate structures. In order to make a realistic thermal characterization, not only a bare micro-hotplate structure is measured; we also added thermal mass on the hotplate structure by depositing the contact metals for the sensing metal oxide thin film and the metal connection pads by e-beam evaporation and the thin tin oxide (SnO2 ) metal oxide sensing layer by sputtering. The structure of the metal oxide gas sensor is shown in Figure 1, which demonstrates the cross-section of the gas sensor. As shown in Figure 1, the basic concept of the hotplate is the planar-structured, continuous ITO resistor that lies underneath the active metal oxide layer, isolated with a thin Si3 N4 layer.
Sensors Sensors 2016, 2016, 16, 16, 1612 1612
33 of of 77
Figure 1. Cross-section of the gas sensor. Figure 1. Cross-section of the gas sensor.
Sensors on p-type, double-side, polished, 300-µm-thick (100) silicon (Si)silicon substrates. Sensors were werefabricated fabricated on p-type, double-side, polished, 300-µm-thick (100) (Si) As the first step of the fabrication, 1.8 µm thermal SiO is grown on both sides of a wafer in a 2 substrates. As the first step of the fabrication, 1.8 µm thermal SiO2 is grown on both sides of a wafer tube-furnace (MTI,(MTI, Richmond, CA, USA) oxidation. This layer serves an extra in a tube-furnace Richmond, CA, using USA) wet using wet oxidation. This layerasserves as masking an extra layer during the bulk Si etching process and also as a mechanical protection layer for the fabrication masking layer during the bulk Si etching process and also as a mechanical protection layer for the steps on the steps backside silicon substrate, avoiding scratches. A 1-µm-thick Si3 NA4 layer is deposited fabrication on of thethe backside of the silicon substrate, avoiding scratches. 1-µm-thick Si3N4 on the thermal SiO with plasma-enhanced chemical vapor deposition (PECVD) (Samco, Kyoto, Japan) 2 layer is deposited on the thermal SiO2 with plasma-enhanced chemical vapor deposition (PECVD) to form an etch stop layer for theanwet anisotropic etching process. Si3 N4 Si is etching preferred due to its very (Samco, Kyoto, Japan) to form etch stop layerSifor the wet anisotropic process. Si3N 4 is low etch rate in KOH solution. preferred due to its very low etch rate in KOH solution. The The thermal thermal oxide oxide on on the the front front side side of of the the wafer wafer is is removed removed by by reactive reactive ion ion etching etching in in an an inductively coupled plasma reactor (ICP-RIE) (Samco, Kyoto, Japan) using fluorine chemistry. inductively coupled plasma reactor (ICP-RIE) (Samco, Kyoto, Japan) using fluorine chemistry. The The etching parameters W100 ICP,W100 bias power, Pa chamber and etching parameters were were 75 W 75 ICP, RF W biasRFpower, 0.25 Pa 0.25 chamber pressure,pressure, and 30 sccm 30 sccm CHF process gas flow. Then, 1-µm-thick Si N is grown with PECVD as a membrane 3 3 4 CHF3 process gas flow. Then, 1-µm-thick Si3N4 is grown with PECVD as a membrane of the sensor. of the sensor. Si3 N4 was preferred due to its low thermal[3] conductivity [3] in order to reduce PECVD Si3N4 PECVD was preferred due to its low thermal conductivity in order to reduce the heat losses the losses due to thermal Si3 N parameters 130 W 275 ◦ C 4 deposition dueheat to thermal conduction. Siconduction. 3N4 deposition parameters were 130 W RFwere power, 275RF °Cpower, temperature, temperature, 75 pressure Pa chamber andgas precursor gasSiH flow wassccm)/NH SiH4 (160 3sccm)/NH sccm)/N 3 2(6(500 75 Pa chamber andpressure precursor flow was 4 (160 (6 sccm)/N sccm).2 (500 sccm). Resistor byphotolithography. reversal photolithography. ITO resistor is deposited Resistor patterns arepatterns definedare bydefined reversal The ITOThe resistor is deposited by RF by RF magnetron sputtering system (Leybold Oerlikon, Köln, Germany) using an indium oxide/tin magnetron sputtering system (Leybold Oerlikon, Köln, Germany) using an indium oxide/tin oxide oxide %)with target with a purity. 99.99%Deposition purity. Deposition is performed at power, 75 W RF (90/10 (90/10 by wt. by %) wt. target a 99.99% is performed at 75 W RF 2.2power, × 10−3 − 3 ˙ 2.2 × 10 mbarwith pressure a rate of Ar 1.4 atmosphere A/s in Ar atmosphere resulting an ITO layer thickness of mbar pressure a ratewith of 1.4 Ȧ/s in resulting in an ITOin layer thickness of 140 nm. 140 nm. After deposition, lift off is done. The 1-µm-thick Si N is grown with PECVD as a passivation 3 4 After deposition, lift off is done. The 1-µm-thick Si3N4 is grown with PECVD as a passivation layer. layer. Thesilicon bulk silicon etch mask is defined by photolithography and patterned 3 NICP-RIE. 4 by ICP-RIE. The bulk etch mask is defined by photolithography and patterned into into Si3NSi 4 by The The openings on the passivation etched with ICP-RIE using CHF A 200-nm-thick SnO 3 plasma. 2 openings on the passivation are are etched with ICP-RIE byby using CHF 3 plasma. A 200-nm-thick SnO 2 is is deposited as an active layer with RF magnetron sputtering system (Nanovak, Ankara, Turkey) by deposited as an active layer with RF magnetron sputtering system (Nanovak, Ankara, Turkey) by using target withwith 99.99% purity.purity. Prior toPrior the active material Cr/Au/Pt 2 sputtering using the theSnO SnO 2 sputtering target 99.99% to the active deposition, material deposition, contact metals are deposited by e-beam evaporation as resistor contacts. Cr/Au thick metal contact Cr/Au/Pt contact metals are deposited by e-beam evaporation as resistor contacts. Cr/Au thick metal pads for the sensing layer are deposited with an e-beam evaporator system (Leybold Oerlikon, Köln, contact pads for the sensing layer are deposited with an e-beam evaporator system (Leybold ◦ Germany) as well. Bulk silicon etchBulk was performed 33% KOH solution at KOH 90 C solution for the formation of Oerlikon, Köln, Germany) as well. silicon etchin was performed in 33% at 90 °C for the membrane by using a special wafer holder with backside protection capability. The microscope the formation of the membrane by using a special wafer holder with backside protection capability. photograph of thephotograph final sensor of device is shown in Figure totalin heated which is also thearea, ITO The microscope the final sensor device2.isThe shown Figurearea, 2. The total heated 2 2 resistor 820 × 820 µm while membrane area was 920 × 920 µmarea . The SnO 2 while the 2 active which isarea, also was the ITO resistor area, wasthe 820overall × 820 µm overall membrane was 920 × 920 2 and positioned geometrically sensing area was 500 × 500 µm at the center of both the membrane and 2 2 µm . The SnO2 active sensing area was 500 × 500 µm and positioned geometrically at the center of ITO bothmicro-heater. the membrane and ITO micro-heater.
Sensors 2016, 2016, 16, 16, 1612 1612 Sensors Sensors 2016, 16, 1612
of 77 44 of 4 of 7
Figure 2. Fabricated sensor’s optical microscope photograph. Figure 2. Fabricated sensor’s sensor’s optical optical microscope microscope photograph.
3. Results and Discussion Resultsand andDiscussion Discussion 3. Results The resistance of the square-shaped ITO heater resistor was measured as 35 Ohms by the The resistance the square-shaped ITO heater resistor was measured as 35 by the four-point resistanceof measurement technique. By using a thermal microscope, theOhms relationship resistance measurement technique. By using a thermal microscope, the relationship between four-point resistance measurement technique. By using a thermal microscope, the relationship between the input power and active area temperature was measured, as shown in Figure 3. The the inputthe power and active area temperature was measured, as The shown in 3.in The calibration between input power and active area temperature measured, asFigure shown Figure 3. The calibration temperature used for the measurements waswas 50 C. graph implies that the heating ◦ temperature used for the measurements was 50 C. The graph implies that the heating efficiency calibration temperature used for the measurements was 50 C. The graph implies that the heating efficiency of the structure is weakly dependent on the temperature. At higher heating powers, the of the structure weakly isdependent onslope the temperature. At higher powers, the heating efficiency of the is structure weakly The dependent temperature. Atheating higher heating powers, heating efficiency decreases slightly. of on the the linear fitting line gives the heating efficiency ofthe the efficiency decreases slightly. The slope of the linear fitting line gives the heating efficiency of the sensor heating efficiency decreases slightly. The slope of the linear fitting line gives the heating efficiency of the sensor as 3 C/mW. as 3 ◦ C/mW. sensor as 3 C/mW.
Figure Figure 3. 3. Power Power consumption consumption versus versus measured measured maximum maximum metal metal oxide oxide film film temperature. temperature. Figure 3. Power consumption versus measured maximum metal oxide film temperature.
Temperature with QFI InfraScope II (Quantum Focus Istruments, Vista, CA, Temperaturemapping mappingisisperformed performed with QFI InfraScope II (Quantum Focus Istruments,Vista, Temperature mapping is performed with QFI InfraScope II (Quantum Focus Istruments,Vista, USA), with with a liquid nitrogen–cooled InSb detector, using radiance and uniformity calibration targets. CA USA), a liquid nitrogen–cooled InSb detector, using radiance and uniformity calibration CA USA), with a liquid nitrogen–cooled InSb using radiance and uniformity calibration Measurement calibration is done by keeping thedetector, sample the microscope stage with a high thermal targets. Measurement calibration is done by keeping theon sample on the microscope stage with a high ◦ targets. Measurement calibration is done by keeping the sample on the microscope stage with a high conductance liquid between the surfaces. The stage temperature is fixed at C,at no50bias thermal conductance liquid between the surfaces. The stage temperature is 50 fixed °C,isnoapplied bias is thermal conductance liquid between the surfaces. The stage temperature is fixed at 50 °C, no bias is to the micro-heater and the and reference radianceradiance is taken.isThis way, different emissivity values values of the applied to the micro-heater the reference taken. This way, different emissivity applied to the micro-heater theand reference radiance isareas taken. way, different emissivity values metal ITO heater and dielectric coated areas calibrated using measured radianceradiance data as of the electrodes, metal electrodes, ITO and heater dielectric coatedare areThis calibrated using measured of the metal electrodes, ITO heater and dielectric coated areas are calibrated using measured radiance references. Then biasThen is applied to applied the micro-sensor, and then temperature is data as references. bias is to the micro-sensor, and then distribution temperatureuniformity distribution data as references. Then bias is applied to theto micro-sensor, and then temperature distribution measured and corrected according to the reference map. uniformity is measured and corrected according the reference map. uniformity corrected according to of thethe reference map. Figure shows theand thermal microscope view operating device at a 80 mW mW heating heating power. power. Figureis4 measured Figure 4 shows the thermal microscope view of the operating device at a 80 mW heating power. As As can can be be seen seen from from the the line line scan scan through through the the center center of of the the heated heated zone, zone, the the temperature temperature difference difference As can bethe seen fromresistor the linecontacts scan through the30% center heated zone, the temperature difference between planar is about of of thethe maximum temperature at the center of the between the planar resistor contacts is about 30% of the maximum temperature at the center of the
Sensors 2016, 16, 1612 Sensors 2016, 16, 1612
5 of 7 5 of 7
heater. However, the temperature change in the SnO2 active layer area was measured to be lower thanSensors 7% with a maximum 17 C temperature difference between the hottest and coldest points. 2016, 16, 1612 resistor contacts is about 30% of the maximum temperature at the center 5 of 7of the between the planar Sudden temperature jumps on the sensor electrodes and heater electrodes were probably linked to heater. However, the temperature change in the SnO2 active layer area was measured to be lower heater. However, the temperature changeemissivities, in the SnO2 whereas active layer was measured to be lower the temperature dependence of the metal thearea thermal microscope uses a single than than 7% withwith a maximum 17 ◦ C temperature differencebetween between hottest coldest points. a maximum difference thethe hottest and and coldest points. emissivity7% value for each pixel17atC thetemperature reference radiance measurement temperature. Sudden temperature jumps on the sensor electrodes and heater electrodes were probably linked to Sudden temperature jumps on the sensor electrodes and heater electrodes were probably linked to the temperature dependence of the metal emissivities, whereas the thermal microscope uses a single the temperature dependence of the metal emissivities, whereas the thermal microscope uses a single emissivity value for for each pixel atatthe measurement temperature. emissivity value each pixel thereference reference radiance radiance measurement temperature.
Figure 4. Temperaturedistribution distribution image image taken (a) (a) and and temperature Figure 4. Temperature distribution image taken by bya athermal thermalmicroscope microscope temperature 4. Temperature taken by and temperature distribution line scan through center of the device at 80 mW (b). distribution distribution line line scan scan through through center center of of the the device device at at 80 80 mW mW (b). (b).
Figure 5 shows the results of transient measurements taken with the thermal microscope. Figure 55 shows shows the the results results of of transient measurements measurements taken with the thermal microscope. Figure Measurements were performed at 2 transient Hz with a pulse width of 200taken ms. with the thermal microscope.
Measurements were were performed performed at at 22 Hz Hz with with aa pulse pulse width width of of 200 200ms. ms. Measurements
Figure 5. Heat-up and cool-down times as measured at 100 C, 150 C, and 200 C average active area temperature. ◦ C average Figure 5. 5. Heat-up at at 100 C,◦ C, 150150 C,◦and 200 200 C average active area Figure Heat-upand andcool-down cool-downtimes timesasasmeasured measured 100 C, and active temperature. area temperature.
Sensors 2016, 16, 1612
6 of 7
As is commonly used, the rise time can be defined as the time for the sensor to heat up from 10% to 90% of the maximum stabilized temperature. Similarly, the fall time is defined as the time for the sensor to cold down from 90% to 10% of its maximum temperature. Transient measurements were made for three different working temperatures. Heat-up and cool-down times of 40 ms and 20 ms were obtained, respectively. These values can be improved by reducing the size of the heater areas inside the thermally insulating Si3 N4 membranes to reduce heat leakages. Also, by using a low-pressure PECVD (LPECVD) membrane and passivation layers with low intrinsic stress, the film thicknesses can be further reduced. In this way the thermal mass can be reduced and the rise-fall time values can be improved. 4. Conclusions This work describes the design and fabrication of a novel micro-hotplate structure for thin-film metal oxide gas sensors, using silicon micromachining methods. ITO is used as the resistor material with a square planar geometry. The structure and fabrication techniques are simple but they meet the required performance specifications of low power consumption and linearity of heating efficiency for such devices. Further, the demonstrated heating element achieves high temperature uniformity with a 17 ◦ C gradient at a 300 ◦ C working temperature on the active film area with 80 mW power consumption. The rise and fall times are measured to be 40 ms and 20 ms, respectively. The hotplate structure with the square planar ITO geometry that we introduced solves the temperature nonuniformity problem. Cold spots at elevated temperatures, which arise from the non-continuous nature of conventional resistor meanders [14], are avoided. Key advantages of the planar ITO resistors are their continuity underneath the active metal oxide layers and the lacking necessity of extra microstructures, such as heat-spreading plates. The micro-hotplate with a planar ITO resistor can serve as the platform for integrated thin-film gas sensors. It brings simplicity to sensor design and fabrication. Both the membrane structure and the resistor are transparent; it could be a promising technology for sensor applications based on both of the electrical and optical measurements. Acknowledgments: This work is supported by the project DPT-HAMIT and the Turkish Science Industry and Technology Ministry SANTEZ Program. One of the authors (E.O.) also acknowledges partial support from the Turkish Academy of Sciences. Author Contributions: M.C.Ç. and D.C. conceived, designed the experiments and made the measurements. M.C.Ç. made the fabrication. M.C.Ç. and B.B. prepared the manuscript and figures. E.Ö. provided consumables, processing equipment and characterization setup, followed all the work from the beginning, including the design, fabrication and measurements, read the manuscript, commented on the design and manuscript, and made corrections on the manuscript. Conflicts of Interest: The authors declare no conflict of interest.
References 1. 2.
3. 4. 5. 6.
Wang, C. Metal Oxide Gas Sensors: Sensitivity and Influencing Factors. Sensors 2010, 10, 2088–2106. [CrossRef] [PubMed] Çalı¸skan, D.; Bütün, B.; Çakır, M.C.; Özcan, S¸ .; Özbay, E. Low dark current and high speed ZnO metal–semiconductor–metal photodetector on SiO2 /Si substrate. Appl. Phys. Lett. 2014, 105, 161108. [CrossRef] Razeghi, M.; Rogalski, A. Semiconductor ultraviolet detectors. J. Appl. Phys. 1996, 79, 7433–7473. [CrossRef] Hwang, W.; Shin, K.; Roh, J.; Lee, D.; Choa, S. Development of Micro-Heaters with Optimized Temperature Compensation Design for Gas Sensors. Sensors 2010, 11, 2580–2591. [CrossRef] [PubMed] Manginell, R.P.; Smith, J.H.; Ricco, A.J. An overview of micromachined platforms for thermal sensing and gas detection. Proc. SPIE 1997, 3046. [CrossRef] Semancik, S.; Cavicchia, R.E.; Wheelera, M.C.; Tiffanya, J.E.; Poiriera, G.E.; Walton, R.M.; Suehlea, J.S.; Panchapakesanb, B.; DeVoeb, D.L. Microhotplate platforms for chemical sensor research. Sens. Actuators B Chem. 2010, 77, 579–591. [CrossRef]
Sensors 2016, 16, 1612
7. 8. 9. 10.
11. 12.
13. 14. 15.
7 of 7
Min, Y. Properties and Sensor Performance of Zinc Oxide Thin Films. Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, MA, USA, 2003. Semancik, S.; Cavicchi, R. Kinetically controlled chemical sensing using micromachined stuctures. Acc. Chem. Res. 1998, 31, 279–287. [CrossRef] Steffes, H.; Imawan, C.; Solzbacher, F.; Obermeier, E. Enhancement of NO2 sensing properties of In2 O3 -based thin films using an Au or Ti surface modification. Sens. Actuators B Chem. 2001, 78, 106–112. [CrossRef] Kolmakov, A.; Klenov, D.O.; Lilach, Y.; Stemmer, S.; Moskovits, M. Enhanced Gas Sensing by Individual SnO2 Nanowires and Nanobelts Functionalized with Pd Catalyst Particles. Nano Lett. 2005, 5, 667–673. [CrossRef] [PubMed] Jing, Z.; Zhan, J. Fabrication and Gas-Sensing Properties of Porous ZnO Nanoplates. Adv. Mater. 2008, 20, 4547–4551. [CrossRef] Sidek, O.; Ishak, M.Z.; Khalid, M.A.; Abu Bakar, M.Z.; Miskam, M.A. Effect of heater geometry on the high temperature distribution on a MEMS micro-hotplate. In Proceedings of the 3rd Asia Symposium on Quality Electronic Design (ASQED), Kuala Lumpur, Malaysia, 19–20 July 2011; pp. 100–104. Martinelli, E. Self-adapted temperature modulation in metal-oxide semiconductor gas sensors. Sens. Actuators B Chem. 2010, 161, 534–541. [CrossRef] Briand, D.; Heimgartner, S.; Gétillat, M.-A.; van der Schoot, B.; de Rooij, N.F. Thermal optimization of micro-hotplates that have a silicon island. J. Micromech. Microeng. 2002, 12, 971–978. [CrossRef] Guha, P.K.; Ali, S.Z.; Lee, C.C.C.; Udrea, F.; Milne, W.I.; Iwaki, T.; Covington, J.A.; Gardner, J.W. Novel design and characterization of SOI CMOS micro-hotplates for high temperature gas sensors. Sens. Actuators B Chem. 2007, 127, 260–266. [CrossRef] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).
Planar Indium Tin Oxide Heater for Improved Thermal Distribution for Metal Oxide Micromachined Gas Sensors M. Cihan Çakır 1,2, *, Deniz Çalı¸skan 1 , Bayram Bütün 1 and Ekmel Özbay 1,3 1 2 3
*
Nanotechnology Research Center, Bilkent University, Ankara 06800, Turkey; [email protected] (D.Ç.); [email protected] (B.B.); [email protected] (E.Ö.) Department of Nanotechnology and Nanomedicine, Hacettepe University, Ankara 06800, Turkey Department of Electrical and Electronics Engineering, Department of Physics, Bilkent University, Ankara 06800, Turkey Correspondence: [email protected]; Tel.: +90-312-290-3050
Academic Editors: Sang Sub Kim and Hyoun Woo Kim Received: 18 July 2016; Accepted: 23 September 2016; Published: 29 September 2016
Abstract: Metal oxide gas sensors with integrated micro-hotplate structures are widely used in the industry and they are still being investigated and developed. Metal oxide gas sensors have the advantage of being sensitive to a wide range of organic and inorganic volatile compounds, although they lack selectivity. To introduce selectivity, the operating temperature of a single sensor is swept, and the measurements are fed to a discriminating algorithm. The efficiency of those data processing methods strongly depends on temperature uniformity across the active area of the sensor. To achieve this, hot plate structures with complex resistor geometries have been designed and additional heat-spreading structures have been introduced. In this work we designed and fabricated a metal oxide gas sensor integrated with a simple square planar indium tin oxide (ITO) heating element, by using conventional micromachining and thin-film deposition techniques. Power consumption–dependent surface temperature measurements were performed. A 420 ◦ C working temperature was achieved at 120 mW power consumption. Temperature distribution uniformity was measured and a 17 ◦ C difference between the hottest and the coldest points of the sensor at an operating temperature of 290 ◦ C was achieved. Transient heat-up and cool-down cycle durations are measured as 40 ms and 20 ms, respectively. Keywords: Indium tin oxide; Metal oxide gas sensor; Micro hot-plate; SnO2 ; Heat distribution
1. Introduction Metal oxides are widely used in gas sensing applications, such as chemo-resistive [1] and optoelectronic sensors [2,3]. Metal oxide–resistive gas sensors are being investigated, developed, and used due to their simplicity, production flexibility, low cost and sensitivity to a wide variety of volatiles [1]. Due to increasing interest in those sensors, micro-hotplate structures are being developed with enhanced device performance, decreased power consumption, and low cost [1,4]. Research activities aimed at enhancing the sensitivity and selectivity as well as lowering the response time and power consumption are still under active research. Those research studies present innovative integrated micro-hotplate technologies based on Micro Electro Mechanical Systems (MEMS) to achieve optimized thermal properties such as low power consumption and good temperature uniformity across the active layer [4]. With the development of MEMS technologies in the 1990s, new generations of gas sensors with integrated micro-hotplates were realized. Micromachining processes allow for the precise modification of the thermal characteristics of those devices. The optimization of device geometry, membrane, Sensors 2016, 16, 1612; doi:10.3390/s16101612
www.mdpi.com/journal/sensors
Sensors 2016, 16, 1612
2 of 7
and resistor materials affects the heat transfer mechanisms, resulting in the enhancement of the thermal dispersion and Joule heating characteristics of the devices. Besides these enhancements, the sensor’s heat capacity reduction, which is achieved by thermal mass reduction with the help of micromachining methods, also reduces the power consumption and improves the response time of the sensors [5,6]. By improving the response time, selectivity could be enhanced by rapid temperature cycle measurements to produce additional information by getting the response signature as a variable of temperature pulses [7–9]. Metal oxide gas sensors are based on the surface reactions between the target gas species and the sensing metal oxide film. As a result of the surface reactions, gas molecules interact with the film surface and dope it, the surface band bending alters and then the metal oxide layer resistivity changes. Metal oxide sensors give a response to the wide variety of organic and inorganic gases. However, this wide range has a disadvantage, which is nonselectivity. This disadvantage is eliminated once a temperature-dependent measurement is used. When gas species are introduced to the film surface, the total amount of surface reactions, i.e., the adsorbed and desorbed amount of gas molecules, is correlated with the resistance change of the thin film, and it is temperature-dependent and specific to the gas compound type. The base conductance of the sensing film can be defined as the conductance in the absence of the target gas molecules at a given temperature, and it is also strongly dependent on the temperature of the sensing film [6]. Different metal oxides react with the same analytes with different efficiencies at different temperatures [6,10,11]. In other words, the absolute resistivity change caused by the same concentration of the target gas can reach its peak value at different temperatures for different metal oxide films. This information can be used to obtain more selective data extraction from metal oxide gas sensors. For a given temperature-dependent measurement, the nonuniformity of the surface temperature distribution means that the total sensor signal will be an integration of signals obtained from metal oxide sites at different temperatures. Larger nonuniformity in temperature means that the data will be blurred and the peaks in the resistance change with temperature changes will be broadened. To get a characteristic temperature-dependent sensor signal at a high resolution, the temperature distribution on the active sensing film region has to be as uniform as possible [12–14]. To reach good temperature uniformity, power consumption, and working temperature goal by using conventional metal and metal alloy resistor materials, the hotplates with heat-spreading plates, micro-bridges, and resistors with complex geometries based on meanders were designed and fabricated [4–6,14,15]. 2. Design and Fabrication In this work, a planar resistor as a heating element is introduced which can provide high temperature uniformity across the active region of the thin-film metal oxide sensors. A thin sputtered indium tin oxide (ITO) film is used as the resistor material, with simple planar resistor geometry. Silicon nitride (Si3 N4 ) is used for both the membrane material and also as the passivation and insulation layers. Conventional bulk silicon micromachining methods such as potassium hydroxide (KOH) etching are used to obtain the micro-hotplate structures. In order to make a realistic thermal characterization, not only a bare micro-hotplate structure is measured; we also added thermal mass on the hotplate structure by depositing the contact metals for the sensing metal oxide thin film and the metal connection pads by e-beam evaporation and the thin tin oxide (SnO2 ) metal oxide sensing layer by sputtering. The structure of the metal oxide gas sensor is shown in Figure 1, which demonstrates the cross-section of the gas sensor. As shown in Figure 1, the basic concept of the hotplate is the planar-structured, continuous ITO resistor that lies underneath the active metal oxide layer, isolated with a thin Si3 N4 layer.
Sensors Sensors 2016, 2016, 16, 16, 1612 1612
33 of of 77
Figure 1. Cross-section of the gas sensor. Figure 1. Cross-section of the gas sensor.
Sensors on p-type, double-side, polished, 300-µm-thick (100) silicon (Si)silicon substrates. Sensors were werefabricated fabricated on p-type, double-side, polished, 300-µm-thick (100) (Si) As the first step of the fabrication, 1.8 µm thermal SiO is grown on both sides of a wafer in a 2 substrates. As the first step of the fabrication, 1.8 µm thermal SiO2 is grown on both sides of a wafer tube-furnace (MTI,(MTI, Richmond, CA, USA) oxidation. This layer serves an extra in a tube-furnace Richmond, CA, using USA) wet using wet oxidation. This layerasserves as masking an extra layer during the bulk Si etching process and also as a mechanical protection layer for the fabrication masking layer during the bulk Si etching process and also as a mechanical protection layer for the steps on the steps backside silicon substrate, avoiding scratches. A 1-µm-thick Si3 NA4 layer is deposited fabrication on of thethe backside of the silicon substrate, avoiding scratches. 1-µm-thick Si3N4 on the thermal SiO with plasma-enhanced chemical vapor deposition (PECVD) (Samco, Kyoto, Japan) 2 layer is deposited on the thermal SiO2 with plasma-enhanced chemical vapor deposition (PECVD) to form an etch stop layer for theanwet anisotropic etching process. Si3 N4 Si is etching preferred due to its very (Samco, Kyoto, Japan) to form etch stop layerSifor the wet anisotropic process. Si3N 4 is low etch rate in KOH solution. preferred due to its very low etch rate in KOH solution. The The thermal thermal oxide oxide on on the the front front side side of of the the wafer wafer is is removed removed by by reactive reactive ion ion etching etching in in an an inductively coupled plasma reactor (ICP-RIE) (Samco, Kyoto, Japan) using fluorine chemistry. inductively coupled plasma reactor (ICP-RIE) (Samco, Kyoto, Japan) using fluorine chemistry. The The etching parameters W100 ICP,W100 bias power, Pa chamber and etching parameters were were 75 W 75 ICP, RF W biasRFpower, 0.25 Pa 0.25 chamber pressure,pressure, and 30 sccm 30 sccm CHF process gas flow. Then, 1-µm-thick Si N is grown with PECVD as a membrane 3 3 4 CHF3 process gas flow. Then, 1-µm-thick Si3N4 is grown with PECVD as a membrane of the sensor. of the sensor. Si3 N4 was preferred due to its low thermal[3] conductivity [3] in order to reduce PECVD Si3N4 PECVD was preferred due to its low thermal conductivity in order to reduce the heat losses the losses due to thermal Si3 N parameters 130 W 275 ◦ C 4 deposition dueheat to thermal conduction. Siconduction. 3N4 deposition parameters were 130 W RFwere power, 275RF °Cpower, temperature, temperature, 75 pressure Pa chamber andgas precursor gasSiH flow wassccm)/NH SiH4 (160 3sccm)/NH sccm)/N 3 2(6(500 75 Pa chamber andpressure precursor flow was 4 (160 (6 sccm)/N sccm).2 (500 sccm). Resistor byphotolithography. reversal photolithography. ITO resistor is deposited Resistor patterns arepatterns definedare bydefined reversal The ITOThe resistor is deposited by RF by RF magnetron sputtering system (Leybold Oerlikon, Köln, Germany) using an indium oxide/tin magnetron sputtering system (Leybold Oerlikon, Köln, Germany) using an indium oxide/tin oxide oxide %)with target with a purity. 99.99%Deposition purity. Deposition is performed at power, 75 W RF (90/10 (90/10 by wt. by %) wt. target a 99.99% is performed at 75 W RF 2.2power, × 10−3 − 3 ˙ 2.2 × 10 mbarwith pressure a rate of Ar 1.4 atmosphere A/s in Ar atmosphere resulting an ITO layer thickness of mbar pressure a ratewith of 1.4 Ȧ/s in resulting in an ITOin layer thickness of 140 nm. 140 nm. After deposition, lift off is done. The 1-µm-thick Si N is grown with PECVD as a passivation 3 4 After deposition, lift off is done. The 1-µm-thick Si3N4 is grown with PECVD as a passivation layer. layer. Thesilicon bulk silicon etch mask is defined by photolithography and patterned 3 NICP-RIE. 4 by ICP-RIE. The bulk etch mask is defined by photolithography and patterned into into Si3NSi 4 by The The openings on the passivation etched with ICP-RIE using CHF A 200-nm-thick SnO 3 plasma. 2 openings on the passivation are are etched with ICP-RIE byby using CHF 3 plasma. A 200-nm-thick SnO 2 is is deposited as an active layer with RF magnetron sputtering system (Nanovak, Ankara, Turkey) by deposited as an active layer with RF magnetron sputtering system (Nanovak, Ankara, Turkey) by using target withwith 99.99% purity.purity. Prior toPrior the active material Cr/Au/Pt 2 sputtering using the theSnO SnO 2 sputtering target 99.99% to the active deposition, material deposition, contact metals are deposited by e-beam evaporation as resistor contacts. Cr/Au thick metal contact Cr/Au/Pt contact metals are deposited by e-beam evaporation as resistor contacts. Cr/Au thick metal pads for the sensing layer are deposited with an e-beam evaporator system (Leybold Oerlikon, Köln, contact pads for the sensing layer are deposited with an e-beam evaporator system (Leybold ◦ Germany) as well. Bulk silicon etchBulk was performed 33% KOH solution at KOH 90 C solution for the formation of Oerlikon, Köln, Germany) as well. silicon etchin was performed in 33% at 90 °C for the membrane by using a special wafer holder with backside protection capability. The microscope the formation of the membrane by using a special wafer holder with backside protection capability. photograph of thephotograph final sensor of device is shown in Figure totalin heated which is also thearea, ITO The microscope the final sensor device2.isThe shown Figurearea, 2. The total heated 2 2 resistor 820 × 820 µm while membrane area was 920 × 920 µmarea . The SnO 2 while the 2 active which isarea, also was the ITO resistor area, wasthe 820overall × 820 µm overall membrane was 920 × 920 2 and positioned geometrically sensing area was 500 × 500 µm at the center of both the membrane and 2 2 µm . The SnO2 active sensing area was 500 × 500 µm and positioned geometrically at the center of ITO bothmicro-heater. the membrane and ITO micro-heater.
Sensors 2016, 2016, 16, 16, 1612 1612 Sensors Sensors 2016, 16, 1612
of 77 44 of 4 of 7
Figure 2. Fabricated sensor’s optical microscope photograph. Figure 2. Fabricated sensor’s sensor’s optical optical microscope microscope photograph.
3. Results and Discussion Resultsand andDiscussion Discussion 3. Results The resistance of the square-shaped ITO heater resistor was measured as 35 Ohms by the The resistance the square-shaped ITO heater resistor was measured as 35 by the four-point resistanceof measurement technique. By using a thermal microscope, theOhms relationship resistance measurement technique. By using a thermal microscope, the relationship between four-point resistance measurement technique. By using a thermal microscope, the relationship between the input power and active area temperature was measured, as shown in Figure 3. The the inputthe power and active area temperature was measured, as The shown in 3.in The calibration between input power and active area temperature measured, asFigure shown Figure 3. The calibration temperature used for the measurements waswas 50 C. graph implies that the heating ◦ temperature used for the measurements was 50 C. The graph implies that the heating efficiency calibration temperature used for the measurements was 50 C. The graph implies that the heating efficiency of the structure is weakly dependent on the temperature. At higher heating powers, the of the structure weakly isdependent onslope the temperature. At higher powers, the heating efficiency of the is structure weakly The dependent temperature. Atheating higher heating powers, heating efficiency decreases slightly. of on the the linear fitting line gives the heating efficiency ofthe the efficiency decreases slightly. The slope of the linear fitting line gives the heating efficiency of the sensor heating efficiency decreases slightly. The slope of the linear fitting line gives the heating efficiency of the sensor as 3 C/mW. as 3 ◦ C/mW. sensor as 3 C/mW.
Figure Figure 3. 3. Power Power consumption consumption versus versus measured measured maximum maximum metal metal oxide oxide film film temperature. temperature. Figure 3. Power consumption versus measured maximum metal oxide film temperature.
Temperature with QFI InfraScope II (Quantum Focus Istruments, Vista, CA, Temperaturemapping mappingisisperformed performed with QFI InfraScope II (Quantum Focus Istruments,Vista, Temperature mapping is performed with QFI InfraScope II (Quantum Focus Istruments,Vista, USA), with with a liquid nitrogen–cooled InSb detector, using radiance and uniformity calibration targets. CA USA), a liquid nitrogen–cooled InSb detector, using radiance and uniformity calibration CA USA), with a liquid nitrogen–cooled InSb using radiance and uniformity calibration Measurement calibration is done by keeping thedetector, sample the microscope stage with a high thermal targets. Measurement calibration is done by keeping theon sample on the microscope stage with a high ◦ targets. Measurement calibration is done by keeping the sample on the microscope stage with a high conductance liquid between the surfaces. The stage temperature is fixed at C,at no50bias thermal conductance liquid between the surfaces. The stage temperature is 50 fixed °C,isnoapplied bias is thermal conductance liquid between the surfaces. The stage temperature is fixed at 50 °C, no bias is to the micro-heater and the and reference radianceradiance is taken.isThis way, different emissivity values values of the applied to the micro-heater the reference taken. This way, different emissivity applied to the micro-heater theand reference radiance isareas taken. way, different emissivity values metal ITO heater and dielectric coated areas calibrated using measured radianceradiance data as of the electrodes, metal electrodes, ITO and heater dielectric coatedare areThis calibrated using measured of the metal electrodes, ITO heater and dielectric coated areas are calibrated using measured radiance references. Then biasThen is applied to applied the micro-sensor, and then temperature is data as references. bias is to the micro-sensor, and then distribution temperatureuniformity distribution data as references. Then bias is applied to theto micro-sensor, and then temperature distribution measured and corrected according to the reference map. uniformity is measured and corrected according the reference map. uniformity corrected according to of thethe reference map. Figure shows theand thermal microscope view operating device at a 80 mW mW heating heating power. power. Figureis4 measured Figure 4 shows the thermal microscope view of the operating device at a 80 mW heating power. As As can can be be seen seen from from the the line line scan scan through through the the center center of of the the heated heated zone, zone, the the temperature temperature difference difference As can bethe seen fromresistor the linecontacts scan through the30% center heated zone, the temperature difference between planar is about of of thethe maximum temperature at the center of the between the planar resistor contacts is about 30% of the maximum temperature at the center of the
Sensors 2016, 16, 1612 Sensors 2016, 16, 1612
5 of 7 5 of 7
heater. However, the temperature change in the SnO2 active layer area was measured to be lower thanSensors 7% with a maximum 17 C temperature difference between the hottest and coldest points. 2016, 16, 1612 resistor contacts is about 30% of the maximum temperature at the center 5 of 7of the between the planar Sudden temperature jumps on the sensor electrodes and heater electrodes were probably linked to heater. However, the temperature change in the SnO2 active layer area was measured to be lower heater. However, the temperature changeemissivities, in the SnO2 whereas active layer was measured to be lower the temperature dependence of the metal thearea thermal microscope uses a single than than 7% withwith a maximum 17 ◦ C temperature differencebetween between hottest coldest points. a maximum difference thethe hottest and and coldest points. emissivity7% value for each pixel17atC thetemperature reference radiance measurement temperature. Sudden temperature jumps on the sensor electrodes and heater electrodes were probably linked to Sudden temperature jumps on the sensor electrodes and heater electrodes were probably linked to the temperature dependence of the metal emissivities, whereas the thermal microscope uses a single the temperature dependence of the metal emissivities, whereas the thermal microscope uses a single emissivity value for for each pixel atatthe measurement temperature. emissivity value each pixel thereference reference radiance radiance measurement temperature.
Figure 4. Temperaturedistribution distribution image image taken (a) (a) and and temperature Figure 4. Temperature distribution image taken by bya athermal thermalmicroscope microscope temperature 4. Temperature taken by and temperature distribution line scan through center of the device at 80 mW (b). distribution distribution line line scan scan through through center center of of the the device device at at 80 80 mW mW (b). (b).
Figure 5 shows the results of transient measurements taken with the thermal microscope. Figure 55 shows shows the the results results of of transient measurements measurements taken with the thermal microscope. Figure Measurements were performed at 2 transient Hz with a pulse width of 200taken ms. with the thermal microscope.
Measurements were were performed performed at at 22 Hz Hz with with aa pulse pulse width width of of 200 200ms. ms. Measurements
Figure 5. Heat-up and cool-down times as measured at 100 C, 150 C, and 200 C average active area temperature. ◦ C average Figure 5. 5. Heat-up at at 100 C,◦ C, 150150 C,◦and 200 200 C average active area Figure Heat-upand andcool-down cool-downtimes timesasasmeasured measured 100 C, and active temperature. area temperature.
Sensors 2016, 16, 1612
6 of 7
As is commonly used, the rise time can be defined as the time for the sensor to heat up from 10% to 90% of the maximum stabilized temperature. Similarly, the fall time is defined as the time for the sensor to cold down from 90% to 10% of its maximum temperature. Transient measurements were made for three different working temperatures. Heat-up and cool-down times of 40 ms and 20 ms were obtained, respectively. These values can be improved by reducing the size of the heater areas inside the thermally insulating Si3 N4 membranes to reduce heat leakages. Also, by using a low-pressure PECVD (LPECVD) membrane and passivation layers with low intrinsic stress, the film thicknesses can be further reduced. In this way the thermal mass can be reduced and the rise-fall time values can be improved. 4. Conclusions This work describes the design and fabrication of a novel micro-hotplate structure for thin-film metal oxide gas sensors, using silicon micromachining methods. ITO is used as the resistor material with a square planar geometry. The structure and fabrication techniques are simple but they meet the required performance specifications of low power consumption and linearity of heating efficiency for such devices. Further, the demonstrated heating element achieves high temperature uniformity with a 17 ◦ C gradient at a 300 ◦ C working temperature on the active film area with 80 mW power consumption. The rise and fall times are measured to be 40 ms and 20 ms, respectively. The hotplate structure with the square planar ITO geometry that we introduced solves the temperature nonuniformity problem. Cold spots at elevated temperatures, which arise from the non-continuous nature of conventional resistor meanders [14], are avoided. Key advantages of the planar ITO resistors are their continuity underneath the active metal oxide layers and the lacking necessity of extra microstructures, such as heat-spreading plates. The micro-hotplate with a planar ITO resistor can serve as the platform for integrated thin-film gas sensors. It brings simplicity to sensor design and fabrication. Both the membrane structure and the resistor are transparent; it could be a promising technology for sensor applications based on both of the electrical and optical measurements. Acknowledgments: This work is supported by the project DPT-HAMIT and the Turkish Science Industry and Technology Ministry SANTEZ Program. One of the authors (E.O.) also acknowledges partial support from the Turkish Academy of Sciences. Author Contributions: M.C.Ç. and D.C. conceived, designed the experiments and made the measurements. M.C.Ç. made the fabrication. M.C.Ç. and B.B. prepared the manuscript and figures. E.Ö. provided consumables, processing equipment and characterization setup, followed all the work from the beginning, including the design, fabrication and measurements, read the manuscript, commented on the design and manuscript, and made corrections on the manuscript. Conflicts of Interest: The authors declare no conflict of interest.
References 1. 2.
3. 4. 5. 6.
Wang, C. Metal Oxide Gas Sensors: Sensitivity and Influencing Factors. Sensors 2010, 10, 2088–2106. [CrossRef] [PubMed] Çalı¸skan, D.; Bütün, B.; Çakır, M.C.; Özcan, S¸ .; Özbay, E. Low dark current and high speed ZnO metal–semiconductor–metal photodetector on SiO2 /Si substrate. Appl. Phys. Lett. 2014, 105, 161108. [CrossRef] Razeghi, M.; Rogalski, A. Semiconductor ultraviolet detectors. J. Appl. Phys. 1996, 79, 7433–7473. [CrossRef] Hwang, W.; Shin, K.; Roh, J.; Lee, D.; Choa, S. Development of Micro-Heaters with Optimized Temperature Compensation Design for Gas Sensors. Sensors 2010, 11, 2580–2591. [CrossRef] [PubMed] Manginell, R.P.; Smith, J.H.; Ricco, A.J. An overview of micromachined platforms for thermal sensing and gas detection. Proc. SPIE 1997, 3046. [CrossRef] Semancik, S.; Cavicchia, R.E.; Wheelera, M.C.; Tiffanya, J.E.; Poiriera, G.E.; Walton, R.M.; Suehlea, J.S.; Panchapakesanb, B.; DeVoeb, D.L. Microhotplate platforms for chemical sensor research. Sens. Actuators B Chem. 2010, 77, 579–591. [CrossRef]
Sensors 2016, 16, 1612
7. 8. 9. 10.
11. 12.
13. 14. 15.
7 of 7
Min, Y. Properties and Sensor Performance of Zinc Oxide Thin Films. Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, MA, USA, 2003. Semancik, S.; Cavicchi, R. Kinetically controlled chemical sensing using micromachined stuctures. Acc. Chem. Res. 1998, 31, 279–287. [CrossRef] Steffes, H.; Imawan, C.; Solzbacher, F.; Obermeier, E. Enhancement of NO2 sensing properties of In2 O3 -based thin films using an Au or Ti surface modification. Sens. Actuators B Chem. 2001, 78, 106–112. [CrossRef] Kolmakov, A.; Klenov, D.O.; Lilach, Y.; Stemmer, S.; Moskovits, M. Enhanced Gas Sensing by Individual SnO2 Nanowires and Nanobelts Functionalized with Pd Catalyst Particles. Nano Lett. 2005, 5, 667–673. [CrossRef] [PubMed] Jing, Z.; Zhan, J. Fabrication and Gas-Sensing Properties of Porous ZnO Nanoplates. Adv. Mater. 2008, 20, 4547–4551. [CrossRef] Sidek, O.; Ishak, M.Z.; Khalid, M.A.; Abu Bakar, M.Z.; Miskam, M.A. Effect of heater geometry on the high temperature distribution on a MEMS micro-hotplate. In Proceedings of the 3rd Asia Symposium on Quality Electronic Design (ASQED), Kuala Lumpur, Malaysia, 19–20 July 2011; pp. 100–104. Martinelli, E. Self-adapted temperature modulation in metal-oxide semiconductor gas sensors. Sens. Actuators B Chem. 2010, 161, 534–541. [CrossRef] Briand, D.; Heimgartner, S.; Gétillat, M.-A.; van der Schoot, B.; de Rooij, N.F. Thermal optimization of micro-hotplates that have a silicon island. J. Micromech. Microeng. 2002, 12, 971–978. [CrossRef] Guha, P.K.; Ali, S.Z.; Lee, C.C.C.; Udrea, F.; Milne, W.I.; Iwaki, T.; Covington, J.A.; Gardner, J.W. Novel design and characterization of SOI CMOS micro-hotplates for high temperature gas sensors. Sens. Actuators B Chem. 2007, 127, 260–266. [CrossRef] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).
