A Micromachined Calorimetric Gas Sensor: an ... - ACS Publications

A pellistor is a type of calorimetric combustible gas sensor that operates on the ... when exposed to a mixture of combustible gas (typically methane)...
0 downloads 0 Views 233KB Size
Anal. Chem. 2003, 75, 126-132

A Micromachined Calorimetric Gas Sensor: an Application of Electrodeposited Nanostructured Palladium for the Detection of Combustible Gases Philip N. Bartlett* and Samuel Guerin

Department of Chemistry, University of Southampton, Highfield, Southampton, SO17 1BJ, U.K.

Palladium films with regular nanoarchitectures were electrochemically deposited from the hexagonal (H1) lyotropic liquid crystalline phase of the nonionic surfactant octaethyleneglycol monohexadecyl ether (C16EO8) onto micromachined silicon hotplate structures. The H1-e Pd films were shown to have high surface areas (∼28 m2 g-1) and to act as effective and stable catalysts for the detection of methane in air on heating to 500 °C. The response of the H1-e Pd-coated planar pellistors was found to be linearly proportional to the concentration of methane between 0 and 2.5% in air with a detection limit below 0.125%. Our results show that the electrochemical deposition of nanostructured metal films offers a promising approach to the fabrication of micromachined calorimetric gas sensors for combustible gases. A pellistor is a type of calorimetric combustible gas sensor that operates on the principle of detecting a change in the temperature of a heated catalytic element when exposed to a mixture of combustible gas (typically methane) and air. The first example of a pellistor was described in the patent literature over 40 years ago.1 Early devices had poor sensitivity and were easily poisoned, because they consisted of a bare coil of fine platinum wire.2,3 Over the subsequent years, a number of changes and improvements have been made to the basic design, and today, pellistors are commercially available from of a number of suppliers around the world and are widely used to monitor and protect against the buildup of combustible gases in industrial applications.2,3 Modern commercial pellistors consist of a coil of fine precious metal wire (usually Pt and typically 10-50 µm in diameter) embedded within a refractory bead loaded with a precious metal catalyst (usually Pd). The central metal coil is used to electrically heat the surrounding catalyst pellet to its operating temperature of typically ∼550 °C. At the same time, the metal coil is used to detect the additional heat produced by the combustion of the gases on the catalyst pellet through the change in the resistance of the coil as the temperature of the catalyst pellet increases. To reduce heat losses by conduction along the metal wire to the bonding posts used to support the device and make electrical contact to it, and thus to achieve reasonable (1) Baker, A.; U.K. Patent 892530, 1959. (2) Hilger, A. Solid State Gas Sensors; Moseley, P. T., Tofield, B. C., Eds.; Bristol: Philadelphia, 1987; Chapter 2. (3) Moseley, P. T. Meas. Sci. Technol. 1997, 8, 223-237.

126 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

sensitivity, it is necessary to use a very fine wire to make the central heater/resistance thermometer coil. Even so, the power consumption of this type of wire-wound pellistor structure remains quite high, typically 120-500 mW, depending on the particular commercial device and its characteristics. In addition, the use of a thin wire coil to support the catalyst pellet makes the devices fragile. A further significant disadvantage of present commercial pellistor designs is that they are not amenable to automated mass production, and many of the of the steps in their fabrication are carried out by hand. This both increases fabrication costs and reduces the reproducibility of response between devices. Over the last 10 years, there have been significant advances in the routine use of silicon micromachining to make a variety of sensor structures, including thermal sensor structures and microhotplates. In a typical silicon microhotplate structure, a thin silicon dioxide or silicon nitride back-etched membrane (one or two millimeters on a side and a few hundred nanometers thick) containing or supporting a microfabricated heater structure (typically, a serpentine Pt track a few micrometers wide) is supported from a thicker silicon frame.4 By making the membrane that supports the heater thin or by reducing the cross sectional area between the membrane and the supporting silicon frame, the power consumption of the microhotplate can be minimized while still allowing the central region of the hotplate to reach a high temperature. The use of micromachined hotplate structures to replace the fine metal coils used in the present generation of commercial pellistors is attractive for several reasons. First, the use of a micromachined hotplate structure would allow parallel fabrication of devices on the wafer level. Since a wafer might typically comprise 240 or more individual hotplates, this would represent a potential saving in fabrication costs and offer the possibility of improvements in the reproducibility of fabrication. Second, by using micromachined hotplate structures, it should be possible to reduce the size of the sensor and to reduce its power consumption. This could open up the opportunity for new applications, for example, in hand-held or portable devices. As a result, a number of groups have been active in trying to fabricate planar pellistor structures.5-17 (4) Gardner, J. W.; Varadan, V. K.; Awadelkarim, O. O.; Microsensors MEMS and Smart Devices; John Wiley & Sons Ltd: Chichester, 2001. (5) Gall, M. Sens. Actuators, B 1993, 15-16, 260-264. (6) Lee, D.-D.; Chung, W.-Y.; Sohn, B.-K. Sens. Actuators, B 1993, 13-14, 252255. 10.1021/ac026141w CCC: $25.00

© 2003 American Chemical Society Published on Web 11/28/2002

In trying to develop a planar pellistor, there are two aspects of the problem that need to be considered: the design and fabrication of the micromachined substrate and the development and deposition of a catalyst that is compatible with this micromachined substrate. Numerous papers describe new substrates used to make planar pellistors for the detection of combustible gases.7-9,11,12,14,15,17-21 Some focus on the selective detection of different combustible gases;5,7,10,16,22 some, on reducing power consumption;9,13,15,23 some, on the development of new catalysts for the detection of combustible gases;6,24,25 and some present new methods to either acquire or treat the sensor data.5,10,12,14,16,22 Nevertheless, despite these efforts, commercial planar pellistor devices are not available, because the catalytic layers used have generally performed poorly in comparison with materials used in traditional beads. In this paper, we report a unique approach to this problem that uses electrochemical deposition from a lyotropic liquid crystalline phase to deposit a high-surface-area, nanostructured catalyst layer onto the center of a microhotplate. This approach has several important advantages. First, electrochemical deposition of the catalyst can be carried out after all the silicon microfabrication steps are complete. Second, the electrochemical deposition of the catalyst occurs only on the electrode surface. Thus, by using photolithography to locate and pattern the electrode region on the microhotplate structure, the catalyst area can be readily defined. Third, the thickness of the catalyst layer is directly controlled by the total charge passed during the electrochemical deposition. Fourth, the electrochemical deposition can, in principle, be carried out once processing of the silicon wafer is complete, either before the wafer is diced and individual devices are mounted and bonded or, as was done here, after dicing, mounting, and bonding individual devices. Fifth, the use of (7) Menil, F.; Lucat, C.; Debeda, H. Sens. Actuators, B 1995, 24-25, 415420. (8) Vauchier, C.; Charlot, D.; Delapierre, D. Sens. Actuators, B 1991, 5, 3336. (9) Gall, M. Sens. Actuators, B 1991, 4, 533-538. (10) Debeda, H.; Rebiere, D.; Pistre, J.; Menil, F. Sens. Actuators, B 1995, 2627, 297-300. (11) Krebs, P.; Grisel, A. Sens. Actuators, B 1993, 13-14, 155-158. (12) Cavicchi, R. E.; Suehle, S.; Kreider, K. G.; Gaitan, M.; Chaparala, P. Sens. Actuators, B 1996, 33, 142-146. (13) Lee, D.-D.; Chung, W.-Y.; Choi, M.-S.; Baek, J.-M. Sens. Actuators, B 1996, 33, 147-150. (14) Aigner, R.; Dietl, M.; Katterloher, R.; Klee, V. Sens. Actuators, B 1996, 33, 151-155. (15) Qiu, L.; Obermeier, E.; Schubert, A. Transducers ’95 - Eurosensors IX 1995, 520-523. (16) Aigner, R.; Auerbach, F.; Huber, P.; Muller, R.; Scheller, G. Sens. Actuators, B 1994, 18-19, 143-147. (17) Zanini, M.; Visser, J. H.; Rimai, L.; Soltis, R. E.; Kovalchuk, A.; Hoffman, D. W.; Logothetis, E. M.; Bonne, U.; Brewer, L.; Bynum, O. W.; Richard, M. A. Sens, Actuators, A 1995, 48, 187-192. (18) Moller, S.; Lin, J.; Obermeier, E. Sens. Actuators, B 1995, 24-25, 343346. (19) Kimura, M.; Manaka, J.; Satoh, S.; Takano, S.; Igarashi, N.; Nagai, K. Sens. Actuators, A 1995, 24-25, 857-860. (20) Mustchall, D.; Sheibe, C.; Obermeier, E. 8th International Conference on Solid State Sensors and Actuators 1995, 57. (21) Fung, S. K. H.; Tang, Z.; Chan, P. C. H.; Sin, J. K. O.; Cheung, P. W. 8th International Conference on Solid State Sensors and Actuators 1995, 207. (22) Sommer, V.; Tobias, P.; Kohl, D. Sens. Actuators, B 1993, 12, 147-152. (23) Lei, Y.; Jiang, Z.; Tao, Y.; Wu, Z. Sensor Review 1995, 15, 22-24. (24) Ehrhardt, J. J.; Colin, L.; Accorsi, A.; Kamierczak, M.; Zdanevitch, I. Sens. Actuators, B 1992, 7, 656-660. (25) Wilson, A.; Wright, J. D.; Murphy, J. J.; Stroud, M. A. M.; Thorpe, S. C. Sens. Actuators, 1994, 18-19, 506-510.

deposition from a lyotropic liquid crystalline phase allows a unique degree of control over the nanostructure of the catalyst layer. Over the past 5 years, together with colleagues in Southampton, we have demonstrated that by carrying out metal deposition from the lyotropic phases of nonionic surfactants, such as octaethylene glycol monohexadecyl ether (C16EO8, where C16 denotes the 16carbon-chain hydrophobic tail and EO8 denotes the eight ethylene oxide units in the hydrophilic headgroup of the molecule), we can form metal or alloy powders and films with regular nanoarchitectures on the 2-3-nanometer scale that are directly determined by the structure of the lyotropic liquid crystalline phase used to template the metal deposition.26-32 Not only does this approach give regular nanostructures with very high specific surface areas (in excess of 106 cm2/cm3) but it also allows us to choose the type of structure, through the choice of the particular lyotropic phase used at the template, and to systematically vary the size of the pores and thickness of the walls within the structure through the choice of the particular surfactant used and by adding hydrocarbons, such as heptane, to the templating mixture. EXPERIMENTAL SECTION Micromachined Pellistor Structure. The micromachined planar pellistor substrates used in this work were specially designed by J. W. Gardner at the Sensor Research Laboratory, Department of Engineering, University of Warwick, and fabricated at IMT, Neuchatel, Switzerland. They were prepared using standard silicon microfabrication techniques using a four mask process.33 The base material was a 380-µm-thick (100) silicon wafer. The top surface of the wafer was first coated with an 80nm-thick layer of thermal oxide designed to help minimize any residual stress in the final microhotplate membrane. This was followed by the deposition of a 250-nm-thick layer of low-stress silicon-rich silicon nitride (Si3N4) deposited by low-pressure chemical vapor deposition (LPCVD). On the top of this was deposited a 10-nm-thick seed layer of tantalum, followed by a 200nm-thick layer of platinum, which was then patterned to form a serpentine resistance thermometer/heater structure. This was then covered by a further 250-nm-thick low-stress layer of siliconrich silicon nitride using LPCVD to totally encapsulate the platinum heater track. On the top of this second silicon nitride layer, and directly above the platinum heater, a 300-nm-thick 1 × 1 mm square gold electrode structure was patterned by a lift-off process. A 10-nm-thick layer of titanium was used as a seed layer to improve the adhesion of the gold. The two gold contact pads for the platinum heater and the contact pad for the gold electrode were deposited and patterned at the same time. A KOH anisotropic (26) Attard, G. S.; Bartlett, P. N.; Coleman, N. R. B.; Elliot, J. M.; Owen, J. R.; Wang, J. H. Science 1997, 278, 838-840. (27) Attard, G. S.; Goltner, C. G.; Corker, J. M.; Henke, S.; Templer, R. H. Angew. Chem., Int. Ed. Engl. 1997, 36, 1315. (28) Attard, G. S.; Leclerc, S. A. A.; Maniguet, S.; Russell, A. E.; Nandhakumar, I.; Bartlett, P. N. Chem. Mater. 2001, 13 (5), 1444-1446. (29) Bartlett, P. N.; Birkin, P. R.; Ghanem, M. A.; de Groot, P.; Sawicki, M. J. Electrochem. Soc. 2001, 148 (2), C119-C123. (30) Bartlett, P. N.; Gollas, B.; Guerin, S.; Marwan, J. Phys. Chem. Chem. Phys. 2002, 4, 3835-3843. (31) Elliott, J. M.; Attard, G. S.; Bartlett, P. N.; Coleman, N. R. B.; Merckel, D. A. S.; Owen J. R. Chem. Mater. 1999, 11, 3602-3609. (32) Guerin, S.; Attard, G. S. Electrochem. Comm. 2001, 3, 544-548. (33) Gardner, J. W.; Lee, S. M.; Bartlett, P. N.; Guerin, S.; Briand D.; de Rooij, N. J. Transducers ’01: Eurosensors XV, Digest of Technical Papers 2001, 1-2, 820-823.

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

127

Figure 1. Schematic representation of the top view (left) and the cross section (right) of the micromachined planar pellistor substrate.

back-etch was then used to create the thin silicon nitride micromachined membrane structure (2 × 2 mm) in the center of the 4 × 4 mm silicon die. Before use, the devices were diced from the wafer and mounted on TO5 headers. The Pt heater had a nominal resistance of 100 Ω, and the power consumption of the mounted device was around 175 mW at 500 °C. Full details of the thermal characterization of this type of device have been given elsewhere.33,34 Nanostructured Palladium Deposition. Nanostructured palladium films were electrochemically deposited from a mixture of 45 wt % 1 mol dm-3 ammonium tetrachloropalladate, (NH4)2PdCl4 (Johnson Matthey or Aldrich), dissolved in water and 55 wt % octaethyleneglycol monohexadecyl ether, C16EO8 (Fluka), to which was added heptane (Fisher) in a ratio of 1 mol of heptane for every 4 mol of C16EO8. The solutions were thoroughly mixed using a wooden stick for several minutes in order to obtain a homogeneous solution. The formation of the homogeneous H1 hexagonal liquid crystalline phase was confirmed using a polarizing optical microscope (Olympus BH-2) fitted with a cooling/ heating stage (Linkam TMS-90). All solutions were prepared using reagent grade water from a Whatman StillPlus purification system. A conventional three-electrode system was used to perform all electrochemical experiments using a large-area platinum counterelectrode together with a homemade saturated calomel reference electrode (SCE). The potential of the reference electrode was regularly checked against a commercial SCE (Russell) and was always within (4 mV of the commercial electrode value. Nanostructured palladium films (referred to as H1-e films to denote films electrochemically deposited from the H1 hexagonal lyotropic liquid crystalline phase) were produced by electrochemical deposition from the plating mixture at +0.1 V vs SCE. The thickness of the palladium films was controlled by recording the total charge passed during deposition. All planar pellistors reported in this paper were produced by depositing an amount of palladium corresponding to a total charge of 20 mC (corresponding to a charge density of 1.7 C cm-2). Full details of the deposition and characterization of the nanostructured H1-e palladium films is given elsewhere.30 Following deposition, the H1-e palladium films were rinsed in deionized water for at least 1 h and then cycled between 0.2 and 1.1 V at 200 mV s-1 in 2 mol dm-3 H2SO4 (Analar, Aldrich) until the oxide stripping peak at 0.5 V reached its maximum size (typically after 50 cycles). (34) Pike, A.; Gardner, J. W. Sens. Actuators, B 1997, 45, 19-26.

128

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

Scanning electron microscopy (SEM) was performed using a JEOL-JFM 6400 fitted with an energy dispersive spectroscopy detector (EDS). Transmission electron microscopy (TEM) was performed using a JEOL-JEM 2000 FX. Gas Test Measurements. The response to methane of the completed planar pellistor structures was measured using a purpose-built gas test system in which the planar pellistor formed one arm of a Wheatstone bridge. The pellistors were fitted in a 150-cm3 sample chamber. Nylon tubing (Phase Separation Ltd.) was used to make all the connections in the gas flow system. The gas flow was controlled using two mass flow controllers (Unit Instrument Ltd, model UFC 1100) set so that the total flow through the gas test chamber was always 500 sccm (standard cubic centimeter per minute). Varying concentrations of methane in air were generated by blending the flows from bottled pure synthetic air (BOC) and 2.5% CH4 in synthetic air (BOC Special Gases). The gas test system was tested by using a commercial pellistor (50 N, City Technology Ltd). The output recorded for the commercial pellistor using our gas test system matched that reported by the manufacturer to within (2 mV. Before each measurement, the Wheatstone bridge circuit was balanced (i.e., the bridge output was set to 0 mV) under a constant flow of pure synthetic air with the planar pellistor held at its operating temperature (∼500 °C). RESULTS AND DISCUSSION Micromachined Pellistor Structure. For the experiments described in this paper, a purpose-designed and -fabricated micromachined hotplate structure was used. This had a platinum resistance heater track incorporated within a 500-nm-thick, 2 mm × 2 mm square, low-stress silicon nitride membrane supported on a 4 mm × 4 mm square silicon frame. On top of the silicon nitride membrane and directly above but electrically insulated from the platinum resistance heater track, there was a 1 mm × 1 mm square gold electrode that was used for the electrochemical deposition of the catalyst layer. The microhotplate design used here was neither especially small nor designed to have especially low power consumption; rather, it was designed as a robust test structure to be used to investigate the feasibility of using an electrochemically deposited catalyst in a planar pellistor configuration. A schematic representation of the structure of the microhotplate is shown in Figure 1. The nanostructured palladium catalyst was electrochemically deposited onto the gold electrode on top of the hotplate from the hexagonal, H1, phase of the lyotropic liquid crystalline plating

Figure 2. Transmission electron micrograph of part of an H1-e palladium film electrochemically deposited at + 0.1 V vs SCE. The film shows the nanostructure at the edge of the sample running into the bulk. In the dark region at the top, the sample is too thick to allow the electron beam to pass through. The pore-to-pore spacing calculated from the micrograph is 6.1 nm. The scale bar is 50 nm.

mixture. The deposition was carried out by bringing the device into contact with a small amount (typically ∼0.2 cm3) of the viscous deposition mixture attached to a coiled platinum wire counterelectrode and saturated calomel reference electrode. Direct evidence for the regular nanostructure of the palladium electrochemically deposited from the lyotropic liquid crystalline mixture was obtained using transmission electron microscopy. The micrograph in Figure 2 shows the regular orientation of pores which run into the bulk. The diameter of these pores was found to be ∼2.5 ( 0.5 nm, with the thickness of the walls separating the pores having the same value. In some cases, a less-wellordered structure was observed. In those cases, the diameter of the pores was found to be ∼3.3 ( 0.5 nm, with the thickness of the walls ∼1.7 ( 0.5 nm. In the remainder of this paper, we use the terminology H1-e Pd film to denote a nanostructured palladium film electrochemically deposited from the H1 hexagonal lyotropic liquid crystalline phase of the plating mixture. Full details of the deposition of H1-e Pd films and their characterization have been given elsewhere.30 Cyclic voltammetry in sulfuric acid of H1-e Pd films deposited on the planar pellistor substrates provides a simple and very sensitive method to estimate the accessible electroactive surface area of the metal. Following deposition, the H1-e Pd films were cycled in 2 mol dm-3 H2SO4 at 200 mV s-1 between 0.2 and 1.1 V for ∼50 cycles. Over this time, the electroactive surface area, as judged by the size of the oxide stripping peak centered at 0.5 V on the cathodic scan, increases steadily and reaches a maximum value. Figure 3 shows a typical cyclic voltammogram of a H1-e Pd film after cycling in sulfuric acid. On the basis of the total charge corresponding to the oxide stripping peak and using the conversion factor of 424 µC cm-2 given by Rand and Woods35 for oxide stripping on palladium in sulfuric acid, we can estimate the electroactive surface area. Furthermore, the charge passed during deposition of the films enables us to estimate the total mass of palladium deposited on the electrode, provided that the faradaic efficiency of the process is known. Separate experiments carried out using the electrochemical quartz crystal microbalance and reported elsewhere30 gave a value of the faradaic efficiency under these conditions of 96%. Thus, by combining the charge passed in deposition and the charge passed in the acid voltammetry to strip the oxide layer from the palladium surface, we obtain an (35) Rand, D. A. J.; Woods, R. J. Electroanal. Chem. 1972, 35, 209-218.

Figure 3. Cyclic voltammogram of an H1-e palladium film (electrodeposited at +0.1 V vs SCE on a 0.78-mm2 gold electrode with a total charge passed of 20 mC) recorded at 200 mV s-1 in 2 mol dm-3 H2SO4.

estimate for the specific surface area of the nanostructured palladium film. Eleven planar pellistors coated with H1-e Pd were produced, and after excluding one device that showed anomalous behavior, the electroactive specific surface areas for the H1-e Pd films deposited on the remaining 10 planar pellistors all lay between 26.8 and 34.9 m2 g-1, giving an average of 29.2 ( 4.2 m2 g-1. This can be compared to the ideal value, calculated on the basis of a regular hexagonal pore structure with 2.5-nm-diameter pores separated by palladium walls 2.5 nm thick of 39 m2 g-1. Scanning electron microscopy was used to examine the H1-e Pd films deposited onto evaporated gold on glass electrodes, Figure 4a, and onto a planar pellistor, Figure 4b. The micrographs show that the H1-e Pd films are relatively smooth on the micrometer scale, dense, uniform, and well-adhered to the gold electrode surfaces. Further details of the TEM and X-ray analysis of nanostructured palladium films and their electrochemistry are given elsewhere.30 Response to Methane. Once the H1-e Pd-coated planar pellistors had been made and characterized by cyclic voltammetry in sulfuric acid, their responses to methane in air were measured. The planar pellistors were connected in one arm of a Wheatstone bridge and an electrical potential was applied across the bridge in order to bring the sensor to its operating temperature (∼500 °C). The device was then left, heated to its operating temperature, for ∼1.5 h in order to obtain a stable baseline. This initial drift of the baseline is believed to be due to changes in the resistance of the platinum heater when it is first heated to 500 °C. Similarly, when exposed to 2.5% methane in air for the first time, the response of the devices passed through a maximum and then decayed exponentially to a plateau so that it was necessary to leave the devices for at least 2 h in methane for the response to stabilize. It is known from the literature that on heating in air, formation of bulk palladium oxide starts around 200 °C.36,37 Work on palladium single crystals suggests that oxide formation occurs via a suboxide, which acts as a precursor to bulk oxidation. This process has been described by Su et al.38 as formation of an amorphous (36) Farrauto, R. J.; Hobson, M. C.; Kennelly, K.; Waterman, E. M. Appl. Catal., A 1992, 81, 227-237. (37) Kohl, D. Sens. Actuators, B 1990, 1, 158-165. (38) Su, S. C.; Carstens, J. N.; Bell, A. T. J. Catal. 1998, 176, 125-135.

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

129

Figure 4. Scanning electron micrographs of H1-e palladium films deposited on (a) an evaporated gold on glass electrode and (b) a micromachined planar pellistor substrate. Part (a) shows a top view of the film showing the dense and uniform layer; part (b) shows a side view of the edge of the film (tilt angle of 70°); again the film appears dense and uniform. The scale bars are both 1 µm.

layer which transforms into crystalline PdO. In the same paper, they found that methane does not readily dissociate on PdO, and hence, an induction period is observed before rapid reduction occurs. During the induction period, small particles of metallic palladium form and segregate from the PdO surface. They concluded that the reduction of PdO by methane occurs via a nucleation mechanism and requires the presence of metallic Pd. Other researchers have reported similar behavior,39 and some have suggested it.36,40 It was also reported that fully oxidized Pd has been found to be inactive for methane oxidation.41 Thus, it appears that the catalytic oxidation of methane requires the presence of both the palladium metal and its oxide.36,37,39-49 We speculate that the initial transient response is due to changes in the film composition when brought to these temperatures in the methane-air mixture. Using SEM, we find no evidence for any significant morphological changes during this conditioning. Calibration Curves. Figure 5a shows the response of a planar pellistor when subjected to concentrations of methane in air varying between 0 and 2.5% increasing in steps of 0.125%. The response shows excellent linearity over the range 0 to 2.5% methane in air with a detection limit, under these conditions, of 700 °C), the catalyst structure breaks down. This breakdown leads to a loss of sensitivity toward methane.

Figure 8. (a-c) Scanning electron micrographs of a H1-e Pd planar pellistor after heating in 2.5% methane in air to 750 °C. Part (c) is a general view of the microhotplate showing the existence of two distinct regions. Parts (a) and (b) are closer views of those two catalyst regions, as indicated by the arrows. The scale bars in (a) and (b) are 1 µm, and in (c), are 100 µm.

of the hotplate is the hottest area. Figure 8b shows an SEM image of the inner region of the catalyst, and it is clear that the structure of the film has been significantly altered by the high temperature (between 700 and 800 °C) and is now made up of agglomerated palladium/palladium oxide particles with sizes between 1 and several micrometers (evidence that the particles contain Pd was obtained by EDAX). In contrast, Figure 8a shows a SEM image at the same magnification as Figure 8b for the outer region of the film. In this case, there appears to be few structural changes (compare to Figure 6). These experiments indicate that the temperature across the catalyst layer on our hotplate is non-

132 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

CONCLUSIONS In this paper, we have demonstrated the first application of an electrochemically deposited catalyst film with a regular nanoarchitecture in a gas sensor. Our results demonstrate that these films can be deposited onto micromachined silicon structures and that this offers a very attractive approach because of the high degree of control over location, thickness, nanostructure, and composition of the catalyst that can be achieved in this way. Our results show that the H1-e films deposited on the microhotplates have high surface areas (∼28 m2 g-1); exhibit excellent adhesion to the device; and on heating in air to 500 °C, are converted to an active catalyst for methane oxidation. The present devices have a power consumption of ∼175 mW and show a linear response to methane between 0 and 2.5% in air with a sensitivity of ∼35 mV/% methane and good stability. The detection limit for the devices is below 0.125% methane in air. These preliminary results are extremely encouraging and strongly suggest that the combination of a micromachined hotplate structure with the electrochemically deposited catalyst can be developed as a practical combustible gas sensor. Further work needs to be done on the design and optimization of the hotplate structure, in improving the reproducibility of device fabrication, and in studies of poison resistance. However, given the unprecedented degree of control over and the ability to design both the hotplate and the catalyst that this approach offers, we are optimistic that practical commercial devices will result from this work. ACKNOWLEDGMENT The authors acknowledge G. Attard, J. Elliott, and M. Willett for helpful discussions on templated deposition and pellistors, J. Elliott for TEM images, J. W. Gardner for supplying the microhotplates, and City Technology Ltd. for financial support. Received for review September 17, 2002. Accepted October 24, 2002. AC026141W