Ag doped WO 3 -based powder sensor for the detection of NO gas in air Ling Chen, Shik Chi Tsang * Surface and Catalysis Research Centre, University of Reading, Whiteknights, Reading RG6 6AD, UK Received 21 August 2002; received in revised form 3 November 2002; accepted 11 November 2002 Abstract WO 3 -based materials as sensors for the monitor of environmental gases such as NO x (NO  NO 2 ) have been rapidly developed for various potential applications (stationary and mobile uses). It has been reported that these materials are highly sensitive to NO x with the sensitivity further enhanced by adding precious group metals (PGM such as Pt, Pd, Au, etc.). However, there has been limited work in revealing the sensing mechanism for these gases over the WO 3 -based sensors. In particular, the role of promoter is not yet clear though speculations on their catalytic, electronic and structural effects have been made in the past. In parallel to these PGM promoters here we report, for the ®rst time, that Ag promotion can also enhance WO 3 sensitivity signi®cantly. In addition, this promotion decreases the optimum sensor temperature of 300 8C for most WO 3 -based sensors, to below 200 8C. Characterizations (XRD, TEM, and impedance measurement) reveal that there is no signi®cant bulk structure change nor particle size alteration in the WO 3 phases during the NO exposure. However, it is found that the Ag doping creates a high concentration of oxygen vacancies in form of coordinated crystallographic shear (CS) planes onto the underneath WO 3 . It is thus proposed that the Ag particle facilitates the oxidative conversion of NO to NO 2 followed by a subsequent NO 2 adsorption on the defective WO x sites created at the Ag±WO 3 interface; hence, accounting for the high molecular sensitivity. # 2002 Elsevier Science B.V. All rights reserved. Keywords: NO; Gas sensor; Sliver; WO 3 -based powder 1. Introduction In recent years, the demand for gas sensors based on safety and process control requirements has been expanding [1]. Much interest is centered on studies of tin oxide as gas sensitive resistors since it is inexpensive, robust, highly sensitive to ¯ammable hydrocarbon gases [2] and can be tuned to differentiate gases with good selectivity [3]. However, this sensor material is not suitable for detecting some of the most dangerous of air pollutants, namely nitric oxide (NO) and nitrogen dioxide (NO 2 ) (collectively referred to as NO x ), because of its poor sensitivity. Currently, approximately one-half of all NO x emissions into the envir- onment are due to power plants and industrial boilers [4]. NO x gas, which is the precursor to nitric and nitrous acid, causes acid rain and photochemical smog; and hence, con- stitutes the critical factor for the destruction of ozone in the troposphere. In fossil fuel combustion, NO x is formed by high temperature chemical processes from both nitrogen present in the fuel and oxidation of nitrogen in the air. Typically, the NO x emissions consist of 90±95% NO with the remainder being N 2 O and NO 2 ranging 0±4000 ppm [4,5]. In case of environmental monitoring, according to the American Conference of Governmental Industrial Hygienist (ACGIH), the threshold limit values (TLV) for NO 2 and NO are 3 and 25 ppm, respectively [5]. Therefore, a NO x sensor requires high sensitivity for detecting low concentration of gases. Another enormous driving force for developing high sensitivity NO x sensor comes from automotive industry. In order to decrease both fuel consumption and carbon dioxide production, new engines with excess of air versus stoichio- metry have been developed. Unfortunately, the conventional catalytic converter of exhaust gas does not work without careful control of NO x [6]. Tungsten trioxide (WO 3 ) is a wide band-gap n-type semiconductor that has attracted much recent interest as a promising sensor because of its excellent sensitivity and selectivity [7].WO 3 thin ®lms were used initially for detecting H 2 S and H 2 . Yamazoe and Miura [8] were amongst the ®rst to report that tungsten trioxide sintered ®lms are selective sensors for low concentrations of nitrogen oxides. WO 3 thin ®lms activated by noble metals (Pd, Pt, Au) layers have later been found to be more sensitive, selective and shown to give a quicker response to NO x [5]. The sensing Sensors and Actuators B 89 (2003) 68±75 * Corresponding author. Tel.: 44-1189-316346; fax: 44-1189-316632. E-mail address: s.c.e.tsang@reading.ac.uk (S.C. Tsang). 0925-4005/02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 4005(02)00430-6 materials of WO 3 promoted with 1 wt.% metal or metal oxides for NO detection were then systematically screened. It was found that some of these promoters were critical for improving the sensitivity [9]. The sensing mechanism of semiconductor oxides lie, in general, on the changes in the resistance resulting from physisorption, chemisorption, catalytic reactions and spe- cies migration (spillover) on the promoter±sensor particles upon analytical gas exposure [10]. It has been acknowledged that the thin ®lm microstructure plays a major role in this behavior [11,12]. This is mainly because grains and grain boundary regions are expected to have signi®cantly different electronic properties and, accordingly, to show a different electronic response after the interaction with gases [11]. However, there is limited work reported in the literature on addressing the precise sensing mechanism for the NO x gases detection over WO 3 -based sensors and there is no work to reveal the roles of the promoter. It is known that sensitivity of NO 2 is many times higher than NO over the WO 3 -based sensors and most of the best-reported promoters are also known to be good oxidation catalysts. As a result, one of the roles of promoter may be to provide a surface for the catalytic conversion of NO to NO 2 that are responsible for the high sensitivity associated with the promotion. However, whether a speci®c interaction(s) between the promoter and the WO 3 exists is not yet clear. In addition, WO 3 phases generally regarded as distortions of the cubic ReO 3 structure [13] are known to easily form non-stoichio- metric oxides in which the oxidation state of the W is less than 6 [14]. Their stability, relationship with promoter and to the overall NO x sensing are still unclear. In this paper, we report the screening of WO 3 powder doped with different promoters as sensor candidates. Amongst all the promoters tested, Ag/WO 3 shows the best response to 50 ppm NO in air at the temperature range of 150±350 8C. It is found that the intrinsic high sensitivity of Ag/WO 3 could not be mimicked via passing the same gas mixture over a separate Ag/Al 2 O 3 bed (a good catalyst for NO to NO 2 ) prior to a pure WO 3 sensor arranged in series under identical conditions. This suggests that Ag particles supported on WO 3 not only provide NO 2 but also give some kinds of bene®cial interfacial effects with the underneath WO 3 . It is evident from the XRD and the TEM that the Ag doping creates oxygen-de®cient sites that are believed to be crucial for the enhanced sensitivity obtained over this material. 2. Experimental 2.1. Sensor material preparation Analytical grade WO 3 powder (Aldrich, 99.9%) was purchased and used without further puri®cation. Doping of WO 3 was achieved by mixing 1 wt.% precursors (soluble salts) dissolved in a minimum amount of deionized water (DI) with a dried WO 3 powder. The resulting mixture was kept stirring at 80 8C until it was completely dried. All solids were calcined at 600 8C for 2 h in air. The calcined powder, 0.6 g, was cold pressed at 10 tonnes into a pellet (area  1cm 2 ) using an infra-red press. 2.2. Testing conditions The sensor pellet was then sandwiched between two platinum electrodes (Fig. 1), which were in turn connected to an impedance analyzer (Solartron SI 1260) via two jacketed co-axial cables. The electrical responses of the sensor materials were measured in a ¯owing stream of pure air or in 40 ppm NO in air (60:40 mixture from an air cylinder with another cylinder (BOC special gases) contain- ing ca. 100 ppm NO balanced with air) by the ac impedance analyzer (Fig. 1). The ¯ow rate of the gases was kept at 100 ml/min throughout the measurement using mass ¯ow controllers. The schematic diagram is presented below. 3. Results and discussion 3.1. Pure WO 3 Before conducting the screening 1% doped WO 3 sensors, pure tungsten trioxide sensor pellet was initially tested for the detection of NO in air in order to optimize acquisition parameters. It was found that the resistance (collected from the impedance at 0.5 Hz) of the material decreased at increasing temperature, thus the observation is consistent with the semiconducting behavior of the material. The resistance of our WO 3 increased when the sensor was exposed to NO gas in a similar way to WO 3 sensors reported in [5,9,15].WO 3 is an n-type semiconductor, which means that NO forms an anionic adsorbate on the surface of WO 3 particles under the present conditions. Although the material responded almost instantaneously to the change of air to NO gas, it required 20±30 min for the resistance to settle (response time to reach steady state). This presumably is the minimum time for the gas molecules to diffuse through the compressed WO 3 powder with reference to this parti- cular sensor con®guration (the gaseous molecules likely Fig. 1. The apparatus for the sensing measurements. L. Chen, S.C. Tsang / Sensors and Actuators B 89 (2003) 68±75 69 diffuse into the compressed powder through the edges of the pellets). As a result, all data presented in this paper were then collected at least 30 min after the gas exposure. It is inter- esting to note from Table 1 that the sensitivity, S (S  resistance in NO/resistance in air) peaked at around 300 8C. This is in a good agreement with the optimum temperature of around 300 8C reported in the literature using WO 3 -based sensors [5,15]. It is accepted that lower sensi- tivity is obtained at higher temperatures since the thermal energy of the carrier (electrons) can overcome the charge depletion layer in WO 3 . On the other hand, the poor sensi- tivity at low temperature is related to the high activation barrier for NO to form adsorbate on the oxide surface. As typical of compressed powder, there are some variations in the absolute response to the same gas mixture over different sensors, but comparisons with different temperature, gas composition, ¯ow, etc. can be reliably made by using the same sensor [16]. As seen from Table 2, the optimum sensitivity for 40 ppm NO in air using our pure WO 3 is about S  9 Æ 2 collected at 300 8C. This value is slightly higher than but comparable with the results presented by Yamazoe and co-workers [15], who obtained S  4:6 using a milled WO 3 powder sensor. It is noted that Penza et al. [5] achieved S  4 using a WO 3 thin ®lm sensor and Tom- chenko et al. [17] achieved S  1:5 using a WO 3 (Bi 2 O 3 ) thick ®lm sensor at 300 8C at atmosphere of 40 ppm NO in air. 3.2. Screening 1% doped WO 3 sensors After the ®rst report of using WO 3 -based sensor for NO analysis, Yamazoe and co-workers [9] carried out extensive screening of the WO 3 sensors doped with each of the 48 elements at 1% level. Elements such as Pb and Re were shown to be exceptionally effective in enhancing the sensi- tivity for NO gas. Penza et al. [5] reported the bene®cial effect of doping (Pd, Pt, Au). It is noted that these dopants are also known to be effective as NO oxidation catalysts. In light of these works, here we investigated only doping with a few elements to see the promotion effect (Pb, Ag, Sr, La). Particular emphasis was on the effect of Ag since Ag/Al 2 O 3 is a well-accepted catalyst in the catalysis community for NO conversion to NO 2 in air. As seen from Table 3, the sensitivity is indeed enormously improved by doping the WO 3 with silver giving S  21: 5. The order of effectiveness of dopants is Ag @ Sr > Pb > La. 3.3. Ag doped WO 3 With regards to the signi®cant improvement in sensitivity at 300 8C as the WO 3 is promoted with silver, it is interesting to investigate whether this doped sensor gives the same temperature response as the pure WO 3 (Table 1). Table 4 shows that the sensitivity of this sensor increases with decreasing temperature (250 8C versus 300 and 350 8C). This is interesting and may be important since this new sensor offers lower operation temperature than most of the reported WO 3 -based sensors. A sensitivity as high as 38.3 can be obtained at 250 8C over this Ag/WO 3 sensor as compared to 2.8 without the Ag promotion (13.7 times higher in sensitivity). There was no apparent change in the time ($20±30 min) required to reach steady conditions Table 1 Temperature effect for NO sensitivity Temperature (8C) Sensitivity (5 min) Sensitivity (30 min) 250 1.8 2.8 300 7.1 10.0 350 3.0 5.0 Typical conditions: 0.6 g WO 3 compressed pellet in a flowing stream of 40 ppm NO in air at 100 ml/min. Table 2 Variations over different WO 3 sensors using same experimental conditions Temperature (8C) Sensitivity 300 11.1 300 10.0 300 9.6 300 9.1 300 6.8 Typical conditions: 0.6 g WO 3 compressed pellet in a flowing stream of 40 ppm NO in air at 100 ml/min. Table 3 The sensitivity of doped materials Entry Material Sensitivity 1WO 3 a 7.1 2 WO 3 b 10.6 3 Pb/WO 3 8.0 4 Ag/WO 3 21.5 5 Sr/WO 3 12.8 6 La/WO 3 3.9 Typical conditions: 0.6 g WO 3 -based compressed pellet in a flowing stream of 40 ppm NO in air at 100 ml/min 8C at 300 8C. The powder mixture was filtered, washed, dried and calcined at 600 8C for 2 h. Characteristic FTIR bands confirmed the nature of theWO 3 . a Commercial WO 3 . b WO 3 prepared by a spray method using 4 g Na 2 WO 4 (Aldrich, 99.8%) pre-dissolved in 100 ml DM water, followed by spraying into a 6 M HCl solution with stirring at 80 8C. Table 4 Sensitivity change of Ag doped WO 3 at different temperature Temperature (8C) Sensitivity 250 38.3 300 21.5 350 8.8 Typical conditions: 0.6 g WO 3 compressed pellet in a flowing stream of 40 ppm NO in air at 100 ml/min. 70 L. Chen, S.C. Tsang / Sensors and Actuators B 89 (2003) 68±75 (rate of response) at different temperatures, which suggests that the gas diffusion into the porous powder is not critically dependent on the operation temperature but perhaps on the ¯ow characteristics/porosity within the powder. The sensi- tivity enhancement with the decrease of optimum tempera- ture for NO detection in air associated with the Ag doping on WO 3 has not been reported before this present study. However, Penza et al. [5] reported that noble metals (Pd, Pt, Au) promotions to WO 3 give similar observations. Typically, WO 3 promoted with Pd shifts the optimum tem- perature from 300 to 200 8C with a four-fold sensitivity increase (from S  4 to 16 for 40 ppm NO). It is noted that these metal promoters including Ag are also well known to be capable of catalyzing NO to NO 2 and NO 2 is known to display a much higher molecular sensitivity than NO at the same concentration. Experiments were designed to investigate the role of Ag particles on WO 3 . A catalyst of 2% Ag/Al 2 O 3 was synthe- sized according to [18], which is ef®cient for the conversion of NO to NO 2 in air. The material was sandwiched between two silica wool plugs in a 5 cm in length, 4 mm i.d. vertical Pyrex tube housed in a temperature controlled furnace. Thus, this bed of 5 g of 2% Ag/Al 2 O 3 catalyst was inserted between the gas supply and the pure WO 3 sensor. This setup allowed the physical separation of the Ag from WO 3 with different temperature controls. Fig. 2A shows clearly that when the temperature of the Ag catalyst is increased to 250 8C, the sensitivity of the WO 3 sensor kept at 150 8C increased by three times. Without heating the Ag catalyst or bypassing the gas stream from the Ag catalyst bed no sensitivity enhancement was encountered. Similar observa- tion was obtained when the sensor was kept at 250 8C (Fig. 2B). Although no attempt was carried out to monitor the concentration of NO and NO 2 gases because of the low concentration, one of the roles of the Ag is apparent to provide surface for the conversion of NO to NO 2 according to the literature [18]. It is worth noting that the sensitivity enhancement (three to four times) for separate beds (Ag/ Al 2 O 3  WO 3 ) is still far less than those observed using the Ag doped WO 3 sensor (13.7 times) despite using a long catalyst bed and/or a higher silver content. This indicates that Ag doping onto the WO 3 may also provide a unique interface for the interactions with NO x molecules. 3.4. Characterization of the sensor materials Impedance analysis has been increasingly used to char- acterize the inter-granular particles and interface phenom- ena. It has been found that the simultaneous measurement of resistance and capacitance of a specimen over a wide range of frequency gives much more information than dc or a single-frequency ac measurement [19,20]. Fig. 3 shows the complex impedance spectra of 1% Ag/WO 3 at 350 8C in air and in 40 ppm NO in air mixture, respectively. Both of the impedance values (resistance at variable frequencies) are found to depend on the input frequency, giving typical Fig. 2. Histogram of NO sensitivity vs. Ag/Al 2 O 3 bed temperature. An increase in temperature of the Ag/Al 2 O 3 bed (production of NO 2 ) results in the increase in NO sensitivity over the WO 3 sensor. Fig. 3. Complex impedance plots of Ag/WO 3 (with simulated RC elements) in air and in 40 ppm NO. L. Chen, S.C. Tsang / Sensors and Actuators B 89 (2003) 68±75 71 semi-circle impedance spectra. Their simulated values using equivalent circuitry with RC elements (C to mimic the surface charge depletion layers on particle and between particles, R  granular and inter-granular resistance of the sensor material in the parallel mode) are also presented in Fig. 3. It is interesting to note that when 40 ppm NO in air mixture was passed over the sensor, the impedance of the material was dramatically increased. In general, resistance of WO 3 -based sensors reported in the literature increases upon exposure to NO in air using dc measurement. It is ascribed to the NO or NO 2 (product of NO oxidation over metal doper) adsorption on WO 3 surface generating strongly anionic adsorbates which create a large charge depletion layer (increase in capacitance) to restrict the dc current ¯ow [15]. It is noted however, our simulation results indicate that the capacitance and the contact resistance (Ra) of the Ag/ WO 3 were only marginally altered to a small extent but the main change was in its granular/inter-granular resistance. Thus, it is likely that adsorption of these gases may somehow interfere with the main pathway for a current ¯ow through the sensor particles in this particular sensor. Studies of electrical conductivity of WO 3 have been extensively carried out in the past. It is generally accepted that the electrical conductivity of WO 3 depends critically on its stoichiometry and particularly the presence of oxygen vacancies in WO 3 . Electrical conductivity increases at higher defect concentra- tion [21,22]. Thus, NO x adsorption on these oxygen-de®- cient W sites blocking the current passage through the material is therefore envisaged. 3.4.1. X-ray diffraction (XRD) XRD analyses of this sensor material were carried out using a Spectrolab Series 1300 CPS 120 X-ray powder diffractometer equipped with a capillary sample holder. The data were collected in Debye±Scherrer geometry using a monochromated X-ray beam of nickel ®ltered Cu Ka radiation (l  0:154 nm). This was obtained from a curved Fig. 4. XRD spectra of WO 3 before and after the NO exposure. Fig. 5. XRD spectra of 1% Ag/WO 3 before and after the NO exposure. 72 L. Chen, S.C. Tsang / Sensors and Actuators B 89 (2003) 68±75 quartz monochromatic crystal. The X-ray beam was passed through the capillary, which contained the sample and was then diffracted onto the detector. The detector collected data for all angles in the range 4±648,2y, simultaneously. Fig. 4 gives the diffraction spectra of the WO 3 before and after the NO exposure, which shows a triclinic polycrystal- line structure matching most of the major peaks collected. From Fig. 4, we can see that the (0 0 1) peak has the highest intensity. The lattice constants of the WO 3 were computed to be a  7:311(2) A Ê , b  7:512(3) A Ê , c  3:846(2) A Ê . These values and their relative intensities match very well with those given by the JCPDS ®le no. 20±1323 (a  7:300 A Ê , b  7:520 A Ê , c  3:845 A Ê ) for triclinic WO 3 . It was observed that there is no new peak or any peak shift before or after the NO treatment. This suggests that the crystal structures were identical throughout the sensing. It is noted that this observation is in an agreement with the previous report [19] that revealed no bulk structural change in the WO 3 after its exposure to NO gas in air. In contrast, it is interesting to ®nd that there is a very low but clearly observable diffraction hump at the 2y angle between 10 and 188 in the 1% Ag/WO 3 sample (Fig. 5). Such low angles do not match with any possible diffrac- tion peaks from silver or possible compounds (in fact, no peaks can be assigned to Ag at this doping level) but merely re¯ect some local orders with relatively large lattice para- meters. WO 3 is easy to form local non-stochiometric regions, especially oxygen de®ciency in WO 3 can generate crystallographic shear (CS) structures giving a wide range of local non-stoichiometry of WO 3Àx [23]. Detailed XRD char- acterizations of intermediate phases of W 5 O 14 ,W 18 O 49 and W 20 O 58 derived from the parent WO 3 with long-range order structures have been reported [24]. It is therefore speculated that silver doping might have introduced some degree of local order non-stoichiometry structure in the sensor material. However, identi®cation of this phase(s) by XRD in the background of WO 3 was proved to be very dif®cult because of the low concentration and the poor diffraction region. It is noted however, that the well-characterized W 5 O 14 phases Fig. 6. A low-resolution TEM micrograph showing the 1 wt.% doped WO 3 material after the NO exposure (magnification  152,000 showing average particle size of 300 nm). Fig. 7. A high-resolution TEM micrograph and a model showing that the 1 wt.% doped WO 3 material after the NO exposure generates a high concentration of local ordered CS structures. L. Chen, S.C. Tsang / Sensors and Actuators B 89 (2003) 68±75 73 show a maximum peak value at 2y of 148 [24] which ®ts coincidentally into the hump region of our XRD. 3.4.2. Transmission electron microscopy (TEM) TEM experiments were carried out using a Phillips CM 20, 200 kV (0.26 nm point resolution) under bright ®eld conditions. It was shown from Fig. 6 that the average particle size of the 1% Ag/WO 3 -based material before or after test is about 300 nm in diameter. There was no obvious change in particle size and morphology after this material was quenched from 40 ppm NO in air during the sensing. Fig. 7 shows the high-resolution TEM images of 1% Ag/ WO 3 which revealed high concentration of the defective regions (enriched with CS structure at [0 1 0] with respect to [0 0 1] WO 3 ). It is found that the separation of these defective lines is not always identical re¯ecting their lack of a long-range order (non-stoichiometry region of disorder nature). However, some local regions show ordered patterns (the lack of their long-range order and their heterogeneity may explain the poor re¯ection hump obtained in XRD). Typically, as seen in the ®gure, a region containing ca. 6.93 A Ê inter-planar separation can be visualized. It is inter- esting to note that these defective lines in WO 3 are much easily found in the regions at a close proximity to the Ag particles (at or near to surface). The reason for the creation of CS defects associated with Ag doping is not yet known. So far, there is no evidence that Ag atoms are incorporated into these local ordered defective structures (from XRD, HRTEM). It is however, interesting to point out that silver nanoparticle, upon melting, can absorb a large amount of oxygen from environment (reaching about 10 times its volume, or 0.3% of its weight in oxygen). On cooling to a few degrees above solidi®cation, it abruptly releases much of its oxygen adsorbed in a dramatic phenomenon known as `spit' [25]. There are many recent studies on catalytic oxidation using supported silver nanoparticles based upon these interesting properties. Thus, when the Ag particles are in a close contact with WO 3 , it is likely that oxygen migration between the silver nanoparticles and the WO 3 support is involved. The phenomenon of metal±oxide sup- port interaction and the implications on catalysis are cur- rently being explored in copper, silver or gold on oxide supports [26]. 4. Conclusion Initial development of NO sensor to monitor NO gas in air based on WO 3 is presented here. As far as we are aware, 1% Ag promotion to WO 3 is for the ®rst time disclosed in the open literature. Thus, the Ag promotion can dramatically enhance sensitivity and decrease the optimum sensor tem- perature of the compressed WO 3 powder sensor. Attempts to understand the bene®cial effects of Ag have been carried out. It is evident that Ag surface can oxidize NO to NO 2 that shows a high molecular response (sensitivity) to WO 3 -based materials. Thus, one of the roles of Ag promotion is to provide metal surface for the NO conversion to NO 2 .Itis however, also found that small silver particles residented on WO 3 produce a much higher sensitivity than those using two separate beds (Ag/Al 2 O 3 and WO 3 ). This result indicates a structural synergy between silver and WO 3 (metal±support interaction) for optimum sensitivity. Impedance measure- ment suggests that adsorption of NO x , possibly takes place on the oxygen defective WO 3Àx sites; hence, interrupting this main electrical conductive pathway of the material. Thus, other roles of Ag promotion may include creation of these defective sites on the WO 3 at and near the interface between the small Ag particles and the bulk WO 3 particles as indicated by the present XRD and the TEM results. A detailed study on the WO 3 promoted at different Ag loadings will be very useful in correlating the active sites (created at the Ag±WO 3 interface) with the sensing sensitivity, which is being carried out at the present laboratory. References [1] S.C. Tsang, C. Bulpitt, Rare earth oxide sensors for ethanol analysis, Sens. Actuators, B, Chem. 52 (1998) 226±235. [2] T. Seiyama, A. Kato, K. Fujiishi, M. Nagatani, Anal. Chem. 34 (1962) 1502. [3] C. Bulpitt, S.C. Tsang, Detection and differentiation of C 4 hydrocarbon isomers over the Pd±SnO 2 compressed powder sensor, Sens. Actuators, B, Chem. 69 (2000) 100±107. [4] B.T. Marquis, J.F. Vetelino, A semiconducting metal oxide sensor array for the detection of NO x and NH 3 , Sens. Actuators, B, Chem. 77 (2001) 100±110. [5] M. Penza, C. Martucci, G. Cassano, NO x gas sensing characteristics of WO 3 thin films activated by noble metal (Pd, Pt, Au) layers, Sens. Actuators, B, Chem. 50 (1998) 52±59. [6] F. Me  nil, V. Coillard, C. Lucat, Critical review of nitrogen monoxide sensors for exhaust gases of lean burn engines, Sens. Actuators, B, Chem. 67 (2000) 1±23. [7] S.C. Moulzolf, S. Ding, R.J. Lad, Stoichiometry and microstructure effects on tungsten oxide chemiresistive films, Sens. Actuators, B, Chem. 77 (2001) 375±382. [8] N. Yamazoe, N. Miura, New approaches in the design of gas sensors, in: G. Sberveglieri (Ed.), Gas Sensor, Kluwer, Dordrecht, The Netherlands, 1992, Chapter 1, pp. 1±42. [9] X. Wang, N. Miura, N. Yamazoe, Study of WO 3 -based sensing materials for NH 3 and NO detection, Sens. Actuators, B, Chem. 66 (2000) 74±76. [10] S.C. Tsang, C.D.A. Bulpitt, P.C.H. Mitchell, A.J. Ramirez-Cuesta, Some new insights into the sensing mechanism of palladium promoted tin(IV) oxide, J. Phys. Chem. B 105 (2001) 5737±5742. [11] W. Go È pel, K.D. Schierbaum, NO 2 sensors-current status and future- prospects, Sens. Actuators, B, Chem. 26 (1995) 1±12. [12] C.N. Xu, J. Tamaki, N. Miura, N. Yamazoe, Grain-size effects on gas sensitivity of porous SNO 2 -based elements, Sens. Actuators, B, Chem. 3 (1991) 147±155. [13] A.S. Wells, Structural Inorganic Chemistry, Oxford Science Publications, Oxford, 1987, pp. 569±574. [14] G. Sberveglieri, L. Depero, S. Groppelli, P. Nelli, WO 3 sputtered thin films for NO x monitoring, Sens. Actuators, B, Chem. 26/27 (1995) 89±92. [15] M. Akiyama, J. Tamaki, N. Miura, N. Yamazoe, Tungsten oxide- based semiconductor sensor highly sensitive to NO and NO 2 , Chem. Lett. (1991) 1611±1614. 74 L. Chen, S.C. Tsang / Sensors and Actuators B 89 (2003) 68±75 [16] S. Morrison, M. Madou, Chemical Sensing with Solid State Devices, Academic Press, Boston, 1989. [17] A.A. Tomchenko, V.V. Khatko, I.L. Emelianov, WO 3 thick-film gas sensors, Sens. Actuators, B, Chem. 46 (1998) 8±14. [18] N. Bogdanchilova, FC. Meunier, M. Avalos-Borja, J.P. Breen, A. Pestryakov, On the nature of the silver phases of Ag/Al 2 O 3 catalysts for reactions involving nitric oxide, Appl. Catal., B, Environ. 36 (4) (2002) 287±297. [19] T. Ishihara, H. Fujita, Y. Takita, Effects of Pt addition for SrSnO 3 - WO 3 capacitive type sensor on NO detection at high temperature, Sens. Actuators, B, Chem. 52 (1998) 100±106. [20] W.B. Johnson, W.L. Worrell, in: J.R. Macdonald (Ed.), Impedance Spectroscopy, Wiley, New York, 1987, 238 pp. [21] S.C. Moulzolf, L.J. Legore, R.J. Lad, Heteroepitaxial growth of tungsten oxide films on sapphire for chemical gas sensors, Thin Solid Films 400 (1/2) (2001) 56±63. [22] H. Hosono, M. Miyakawa, H. Kawazoe, K. Shimizu, Formation of electronic conducting amorphous WO 3 thin films by ion implanta- tion, J. Non-Cryst. Solids 241 (2/3) (1998) 190. [23] J.S. Anderson, R.J.D. Tilley, in: M.W. Roberts, J.M. Thomas (Eds.), Surface and Defect Properties of Solid, vol. 3, The Chemical Society, London, 1974. [24] G.L.F. Rey, A. Rothschild, J. Sloan, R. Rosentsveig, R. Popovitz- Biro, R. Tenne, Investigations of nonstoichiometric tungsten oxide nanoparticles, J. Solid State Chem. 162 (2001) 300±314. [25] R.L. Davies, S.F. Etris, The development and functions of silver in water purification and disease control, Catal. Today 36 (1997) 107± 114. [26] M. Haruta, Y. Souma (Eds.), Copper, silver and gold in catalysis, Catal. Today, vol. 36, no.1, Elsevier, 1997. Biographies Ling Chen received a BSc degree in chemistry at Guangxi Teachers College, China in 1995. She is currently completing her postgraduate course under the supervision of Dr. Tsang in the Catalysis Research Centre at Reading. Her research was concerned with solid state sensors and catalysis. Shik Chi Tsang obtained a first class degree in chemistry from Birbeck College, London University in 1987. He received his PhD from Reading University in 1991. After spending 4 years as a departmental research fellow in the Inorganic Chemistry Laboratory at Oxford University, he returned to Reading in 1995. Dr. Tsang currently holds a readership at Reading University and a Royal Society University Fellowship for catalysis research. His main research interests are catalysis, mesoporous materials, supercritical CO 2 and sensors. L. Chen, S.C. Tsang / Sensors and Actuators B 89 (2003) 68±75 75 . responses of the sensor materials were measured in a ¯owing stream of pure air or in 40 ppm NO in air (60:40 mixture from an air cylinder with another cylinder. sensitivity for detecting low concentration of gases. Another enormous driving force for developing high sensitivity NO x sensor comes from automotive industry. In order