Sensors and Actuators B 109 (2005) 128–134 New technology of metal oxide thin film preparation for chemical sensor application Viacheslav Khatko a,∗ , Josep Calderer b , Eduard Llobet a , Xavier Correig a a Departament d’Enginyeria Electronica, Universitat Rovira i Virgili, Campus Sescelades, 43007 Tarragona, Spain b Departament d’Enginyeria Electronica, Universitat Politecnica de Catalunya, Campus Nord, 08034 Barcelona, Spain Available online 3 May 2005 Abstract The reduction of grain size in metal oxide films is one of the key factors to enhance the gas sensing properties of semiconductor layers. The basic idea introduced here is to create thin metal oxide films with small grain size by using a special regime of rf sputtering from either metallic or metaloxide targets. The regime includes the deposition of thin films with one or several interruptions of the sputtering process. The idea has been checked by preparing WO 3 thin films using reactive rf sputtering from a pure tungsten target. Four types of films were prepared. For the first type a non-interrupted sputtering was used. In the deposition of films type 2, 3 and 4, the sputtering process was interrupted once, two and three times, respectively. It was found that the thickness of the WO 3 films and the sensing properties of WO 3 based sensors heavily depend on the number of interruptions during the deposition process. © 2005 Elsevier B.V. All rights reserved. Keywords: WO 3 thin films; Deposition with interruptions; WO 3 -based sensors; Rf sputtering 1. Introduction According to Morrison [1], there exist four general ways to increase the selectivity of gas sensors. These comprise usingcatalysts andpromoters[2–6], controllingtheoperating temperature of the sensors [7–9], including special surface additives for specific surface adsorption [10] and applying differential filters [11–14]. Nowadays, all these approaches are being developed very intensively. In our opinion, in the last few years a new way to preparegas sensors has appeared. This way is connected with the attempt to find methods that increasethe surfacearea ofactivelayersfor chemicalsensing. Since the sensor sensitivity is related to the surface–volume ratio of its sensing film, research can be carried out in three directions: • The first one is related to the investigation of active layers prepared by using nanopowder materials, where parti- cle size is reduced to nanometers. Nanostructure is ex- pectedto havea dramaticinfluenceon sensorperformance. ∗ Corresponding author. Tel.: +34 977558653; fax: +34 977559605. E-mail address: vkhatko@etse.urv.es (V. Khatko). Many different nano-sized powders and technologies for the preparation of active layers are used to investigate the effect of the nanostructure on their sensing characteristics to different toxic gases [15–18]. • The second direction consists of using special methods of preparation for the surface patterning of active layers. One of suchmethods impliesa processof anisotropic etchingto provide a sensor substrate with a much higher surface area [19]. A second method consists ofusing a porous structure on the base of a highly ordered nanoporous alumina layer [20,21]. • The third direction includes the technologies via with thin films of nanometer grain size can be deposited. As a rule, these technologies are used to obtain thin film gas sensors. For example, active layers with grain size of 1–2nm could be deposited using rf sputtering or dc magnetron methods [22,23]. Oneofthe basicideastocreatemetalfilmswithsmall grain size is to use successive step-by-step deposition of ultra-thin films resulting in an island structure of two different mate- rials. In this case, at a particular stage of pure metal cluster 0925-4005/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2005.03.073 V. Khatko et al. / Sensors and Actuators B 109 (2005) 128–134 129 development, clusters of another material can be formed to restrict the coalescence of the metal clusters and the forma- tion of a continuous layer. This technology has been applied with success inthe development of catalyticlayers for silicon micromachined calorimetric gas sensors [22,23]. A second basic idea consists of using a special regime of thin film deposition by dc magnetron, ion-beam or rf sputter- ingfrommetallicor metaloxidetargets. Two differentspecial regimes can be used. In the first one the formation of “extra” interfaces in the body of the metal film occurs as a result of the sputteringpowerdensity beingchanged duringfilmdepo- sition. As a rule a low deposition rate is set during the initial stage of film deposition and a high deposition rate is used during the final stage of the deposition. The second regime implies growing the film with one or several interruptions of the deposition process. In this case the “extra” interfaces are introduced into the body of the thin film. During the inter- ruption of the sputtering process or when a sudden change in the deposition rate occurs, an equilibrium surface is formed due to the free surface bond saturation by the atoms from residual atmosphere and/or the structural relaxation of the interface. For the subsequent prolongation of the deposition process, film growth begins over again on the new “extra” equilibrium surface and the average grain size of the film at the surface is smaller than in the original film. It has been shown that this leads to metal films with a decreased average grain size. The first regime has been used for the deposition of Schottky barriers and MOS transistor gates [24,25]. The use of this regime has allowed obtaining the enhancement of catalytic properties of Pt-SiO 2 thin films, as well [26]. Among metal oxide semiconductors, tungsten oxide is a promising material for gas sensing. Several studies have shown that it can be used for the detection of nitrogen ox- ide (NO and NO 2 ), carbon monoxide, ammonia vapours, and hydrocarbons. Tungsten oxide films can be deposited by re- active rf sputtering, thermal evaporation and other methods. The results obtained indicate that the characteristics of the sensors heavily depend on the conditions and methods used in their preparation[27–31]. Since grainsize is one ofthe key factors to enhance the gas sensing properties of metal oxide sensors, the aim of this paper is to study the influence of in- terrupting the deposition process on the sensing properties of WO 3 thin films prepared by rf sputtering. 2. Experimental The tungsten oxide films were deposited on top of silicon wafers by reactive rf magnetron sputtering using a ESM100 Edwards sputtering system. A metal target of 99.95% purity with a diameter of 100 mm and thickness of 3.175mm was used. The target to substrate distance was set at 70 mm. The silicon wafers, oxidised in dry oxygen, were held in thermal contact with a holder during the deposition process. The sub- strate temperature was kept constant during film deposition at room temperature. The sputtering atmosphere consisted of Ar–O 2 mixed gas and its flow rate was controlled by separate gas flowmeters to provide an Ar:O 2 flow ratio of 1:1. The pressure in the deposition chamber during sputtering was 5 × 10 −3 mbar. The rf sputtering power was 200W. These conditions of deposition gave an average deposition rate up to 2.12 nm per min. Four types of tungsten oxide films were prepared. For the first type a non-interrupted sputtering was used. In the de- position of films type 2, 3 and 4, the sputtering process was interrupted once, two and three times, respectively. A shutter was used to interrupt the deposition process. The total de- position time was 40, 40.5, 41 and 41.5 min, for films type 1, 2, 3 and 4, respectively. The interruption time was set to 30 s. Film thickness was controlled by ellipsometry (PLAS- MOS 2000) and stylus profilomentry (DEKTAK 3030) and was calculated from AES profiles as well. The samples used to investigate the gas sensing properties ofthe filmsasafunction ofthenumberof interruptionsduring the deposition process were based on silicon substrates. The top contacts to the sensing layers were formed using air dry silver paste (Heraeus AD1688-06). Using this paste the test samples were fixed on the ceramic heater prepared according to the method reported in [32]. Theresponse ofthedifferentfilmsto nitrogendioxide,car- bon monoxide, ethanol and ammonia was investigated. The sensors were kept in a temperature and moisture controlled test chamber (27 ◦ C, ±1 ◦ C and 41–43% RH). The sensors were operated at the temperaturerange from 150 to 300 ◦ Cto analyse the effect of working temperature on their response. The resistance of the sensing layers in the presence of either pure air (R air ) or the different pollutants (R gas ) at the different concentrations was monitored and stored in a PC. The morphology of the sensing layers was determined by AFM. The sensing layer surface and the chemical element distributions in the samples were examined with a PHI-660 Auger spectrometer operating at 3 kV and using a probe di- ameter up to 1␮m. Auger electron collection depth was up to 2.0 nm. 3. Results and discussion 3.1. Measurement of film thickness Table 1 shows the measured thickness of the WO 3 thin films. Two basic tendencies in the thickness as a function of Table 1 Thickness (nm) ofthe WO 3 thin film asa function of thenumber of interrup- tions of the deposition process as estimated by profilometry andellipsometry Number of interruptions Measurement method Profilometry Ellipsometry 0 83.2 ± 1.0 81.51 ± 0.11 1 80.0 ± 1.0 65.9 ± 0.61 2 74.0 ± 1.0 55.6 ± 0.33 3 69.5 ± 1.0 – 130 V. Khatko et al. / Sensors and Actuators B 109 (2005) 128–134 Fig. 1. AES profiles of rf sputtered WO 3 thin films without interruption (a), one (b) and two (c) interruptions of the deposition process. V. Khatko et al. / Sensors and Actuators B 109 (2005) 128–134 131 the numberof interruptionsduringthe depositionprocess can be derived from Table 1: • Differences arisein thethickness measuredby profilomen- try and ellipsometry forthose tungsten oxide thin films de- posited with interruptions during the deposition process. The formation into the body of the thin film of several “extra” interfaces made difficult the measurement of film thickness by ellipsometry. Particularly, the method could not be applied to measure films deposited with three inter- ruptions. • The total thickness of WO 3 thin films decreases when the number of interruptions during the deposition process in- creases. Fig. 1 shows AUG depth profiles of chemical el- ements into the WO 3 thin films. It can be seen decreasing the film total thickness that is proportionate to total sputter time. 3.2. Gas sensitivity studies The responses of the different films to nitrogen dioxide, carbon monoxide,ammonia and ethanolwere analysed atop- erating temperatures 150, 200, 250 and 300 ◦ C. Fig. 2 shows Fig. 2. Sensor responseofrf sputtered WO 3 sensing layers toNO 2 at 150 ◦ C (a) and 200 ◦ C (b). WO 3 (0), WO 3 (1) and WO 3 (3) are sensing layers prepared without, with one and three interruptions of thedeposition process, respectively. Fig. 3. Sensor response of rf sputtered WO 3 sensing layers to ammonia at 250 ◦ C. WO 3 (0), WO 3 (1) and WO 3 (3) are sensing layers prepared without, with one and three interruptions of the deposition process, respectively. the response of the different tungsten oxide thin films to NO 2 at 150 ◦ C(Fig. 2a) and 200 ◦ C(Fig. 2b). No matter the op- erating temperature set, sensors based on tungsten trioxide films prepared with the maximum number of interruptions (i.e. three) during their growth show the highest sensitivity to NO 2 . For example, at 1 ppm of NO 2 , the sensitivity for these sensors calculated using the relation S=(R gas − R air )/R air is 2.156, 0.468, 0.255 and 0.167 at 150, 200, 250 and 300 ◦ C, respectively. None of the WO 3 sensing layers responded to CO in the temperature range investigated. The sensors responded to ammonia when operated be- tween 200 and 300 ◦ C. Fig. 3 shows the response of the dif- ferent sensing layers to 10ppm of ammonia at the working temperature of250 ◦ C. Onceagain, the sensorsprepared with three interruptions during the deposition process show the highest sensitivity. Table 2 summarises the sensitivity results to ammonia for the different films studied. It can be seen that 250 ◦ C is the optimal temperature for ammonia sensing (sensitivity is higher at this operating temperature). Fig. 4 shows the response of the different WO 3 sens- ing layers to ethanol at the working temperatures of 250 ◦ C (Fig. 4a) and 300 ◦ C(Fig. 4b). The sensors responded to ethanol when operated between 200 and 300 ◦ C. At 200 and 250 ◦ C theyprepared withoutinterruptions duringtheir depo- sition process show the highest sensitivities. However, when operated at 300 ◦ C, sensors prepared with three interruptions during their deposition process show the highest sensitivity Table 2 Sensor sensitivity (S) in the presence of 10 ppm of ammonia as a function of the working temperature and number of interruptions of the deposition process Number of interruptions Working temperature ( ◦ C) 200 250 300 00.501 0.527 0.0 10.19 0.372 0.209 30.616 1.189 0.22 132 V. Khatko et al. / Sensors and Actuators B 109 (2005) 128–134 Fig. 4. Sensor response of rf sputtered WO 3 sensing layers to ethanol at 250 ◦ C (a) and 300 ◦ C (b). WO 3 (0), WO 3 (1) and WO 3 (3) are sensing layers prepared without, with one and three interruptions of the deposition process, respectively. to ethanol. Table 3 summarises the sensitivity to ethanol for the different films and operating temperatures studied. 3.3. Morphology of the WO 3 films The surface ofthe tungsten trioxide films was investigated by AFM (Nanoscope III). Fig. 5 clearly points out the in- fluence of the number of interruptions during the sputtering process on the morphology of deposited metal oxide films. It can be seen that increasing the number of interruptions leads to a decrease in the roughness of the film surface. Table 3 Sensor sensitivity (S) in the presence of 10 ppm of ethanol as a function of the working temperature and number of interruptions of the deposition process Number of interruptions Working temperature ( ◦ C) 200 250 300 00.103 0.463 0.332 10.00.303 0.384 30.00.437 0.449 Fig. 5. AFM surface morphology of WO 3 thin films without interruption (a), one (b) and three (c) interruptions of the deposition process. 3.4. Discussion As indicatedabove,thechanges inthe thickness ofthe dif- ferent WO 3 films allow drawing the conclusion that “extra” interfaces couldbe formed duringthe process ofinterruption. The prolongation of film growth on the “extra” interface in- volves anewnucleation ofthe metaloxidefilm, theformation and growth of a film island structure and, finally, the forma- tion of a continuous layer. All the steps in this process of V. Khatko et al. / Sensors and Actuators B 109 (2005) 128–134 133 film growth take a specific time to complete. Therefore, the total film thickness decreases by increasing the number of interruptions during the deposition process. The analysis of AUG depth profiles shows that the formation of “extra” in- terfaces is not related to the free surface bond saturation by the atoms from residual atmosphere. Obviously, the “extra” interfaces are formed due to the surface relaxation during the interruption time. This differs from the results obtained for metal films deposited in an atmosphere containing oxy- gen [33], where the sputtered films adsorb oxygen from the residual atmosphere. The enhancement of the sensing properties observed for WO 3 films depositedwiththree interruptions isdue to thede- creasein grainsizeintothemetaloxidefilms. Thisconclusion is confirmed byAFM data. Certainly, further confirmationby alternative methods would be necessary. Work is in progress to perform more analyses, in particular transmission electron microscopy, on the samples. 4. Conclusions WO 3 thin films were deposited by reactive rf sputter- ing from a pure tungsten target. The deposition process was conducted without interruption and with one, two and three interruptions. On the base of these films, sensing lay- ers were prepared and their response to NO 2 , CO, ammonia and ethanol was investigated. It was shown that the sensing layers prepared with the maximum (i.e. three) number of in- terruptions show the best sensing properties. The increase in sensitivity is related to the decrease of grain size in the WO 3 thin films observed as the number of interruptions was increased. This conclusion is based on AFM data. References [1] S.R. Morrison, Selectivity in semiconductor gas sensors, Sens. Ac- tuators 12 (1987) 425–440. [2] G. Sberveglieri, S. Groppelli, P. Nell, V. Lantto, H. Torvela, P. Romp- painen, S. Lepp ¨ avuori, Response to nitric oxide of thin and thick SnO 2 films containing trivalent additives, Sens. Actuators, B Chem. 1 (1990) 79–82. [3] J. Mizsei, V. Lantto, Simultaneous response of work function and resistivity of some SnO 2 -based samples to H 2 and H 2 S, Sens. Ac- tuators, B Chem 4 (1991) 163–168. [4] K.D. Shierbaum, U. Weimar, W. Gopel, Comparison of ceramic, thick-film and thin-film chemical sensors based upon SnO 2 , Sens. Actuators, B Chem. 7 (1992) 709–716. [5] E. Llobet, G. Molas, P. Molinas, J. Calderer, X. Vilanova, J. Brezmes, J.E. Suegras, X. Correig, Fabrication of highly selective tungsten oxide ammonia sensors, J. Electrochem. Soc. 147 (2000) 776–779. [6] C. Bittencourt, R. Landers, E. Llobet, G. Molas, X. Correig, M.A.P. Silvaa, J.E. Sueiras, J. Calderer, Effect of oxygen partial pressure and annealing temperature on the formation of sputtered tungsten oxide films, J. Electrochem. Soc. 149 (2002) H81–H86. [7] R. Ionescu, E. Llobet, Wavelet transform-base fast feature extrac- tion from temperature modulated semiconductor gas sensors, Sens. Actuators, B Chem. 81 (2002) 289–295. [8] R. Ionescu, E. Llobet, S. Al Khalifa, J.W. Gardner, J. Brezmes, X. Vilanova, X. Correig, Response model for thermally-modulated tin oxide based microhotplate gas sensors, in: Proceedings of the 16th European Conference on Solid State Transducers EUROSENSORS XVI, Prague, 2002, pp. 247–248. [9] J. Puigcorbe, A. Vila, J.R. Morante, Thermal fatigue modelling in mi- cromachined gas sensors, in: Proceedings of the 16th European Con- ference on Solid State Transducers EUROSENSORS XVI, Prague, 2002, pp. 257–258. [10] S.R. Morrison, Semiconductor gas sensors, Sens. Actuators 2 (1982) 329–343. [11] G.K. Flingelli, M.M. Fleischer, H. Meixner, Selective detection of methane in domestic environments using a catalyst sensor system based on Ga 2 O 3 , Sens. Actuators, B Chem 48 (1998) 258–262. [12] M. Frietsch, F. Zudock, J. Goschnick, M. Bruns, CuO catalytic mem- brane as selectivity trimmer for metal oxide gas sensors, Sens. Ac- tuators, B Chem 65 (2000) 258–262. [13] M. Fleischer, S. Kornely, T. Weh, H. Meixner, Selective gas detection with high-temperature operated metal oxides using catalytic filters, Sens. Actuators, B Chem 69 (2000) 205–210. [14] M. Frietsch, V. Trouillet, I. Kiselev, J. Goschnick, The influence of different gradient coating on the detecting properties of a metal oxide gas sensor microarray, in: Proceedings of the 16th European Con- ference on Solid State Transducers EUROSENSORS XVI, Prague, 2002, pp. 595–596. [15] G J. Li, X H. Zhang, S. Kawi, Relationships between sensitivity, catalytic activity, and surface areas of SnO 2 gas sensors, Sens. Ac- tuators B Chem. 60 (1999) 64–70. [16] G. Zhang, M. Liu, Effect of particle size and dopant on properties of SnO 2 -based gas sensors, Sens. Actuators B Chem. 69 (2000) 144–152. [17] A. Chiorino, G. Ghotti, F. Prinetto, M.C. Carotta, C. Malagu, G. Martinelli, Preparation and characterization of SnO 2 and WO X –SnO 2 nanosized powders and thick films for gas sensing, Sens. Actuators B Chem. 78 (2001) 89–97. [18] P. Ivanov, E. Llobet, X. Vilanova, J. Brezmes, X. Correig, J. Hubalek, K. Malysz, I. Gracia, C. Cane, Screen-printed nano-grain tin ox- ide films for micro-hotplate gas sensors, in: Proceedings of the fourth Conferencia de Dispositivos Electronicos, Calella de la Costa, Barcelona, 2003, p. 145. [19] B. Lillis, E. Hurley, S. Galvin, A. Mathewson, H. Berney, A novel, high surface area, capacitance based silicon sensor for DNA hybridi- sation detectiondkjdot, in: Proceedings of the 16th European Con- ference on Solid State Transducers EUROSENSORS XVI, Prague, 2002, pp. 693–694. [20] O.K. Varghese, D. Gong, M. Paulose, K.G. Ong, C.A. Grimes, E.C. Dickey, Highly ordered nanoporous alumina films: Effect of pore size and uniformity on sensing performance, J. Mater. Res. 17 (2002) 1162–1171. [21] E.C. Dickey, O.K. Varghese, K.G. Ong, D. Gong, M. Paulose, C.A. Grimes, Room temperature ammonia and humidity sensing using highly ordered nanoporous alumina films, Sensor 2 (2002) 91– 110. [22] V. Khatko, E. Logothetis, R. Soltis, J. Hangas, J. McBride, Develop- ment of highly active catalyst for Si-microcalorymetric gas sensor, in: Sven Kruger, Wolfgang Gessner (Eds.), Advanced Microsystems for Automotive Application, Springer, Berlin, 2000, pp. 27–37. [23] V. Khatko, R. Soltis, J. McBride, K. Nietering, Catalytic properties of Pd/SiO 2 and Pt/SiO 2 multilayer stacks, Sens. Actuators, B Chem. 77 (2001) 548–554. [24] N. Rumak, V. Khatko, S. Vasiliev, Formation of metal-semiconductor contacts by ion-beam methods, in: Proceedings of the Belarusian Academy of Sciences, Ser. Physical-Technical, 4, 1990, pp. 82–84 (in Russian). [25] S. Vasiliev, N. Rumak, V. Khatko, Formation of MOS-structures with quasilayer metal gate, Electron. Eng. Ser. Microelectron. 3 (1990) 10–14, in Russian. [26] V. Khatko, Catalytic property variation of sputtered Pt–SiO 2 thin film, in: Proceedings of International Conference “New Technologies 134 V. Khatko et al. / Sensors and Actuators B 109 (2005) 128–134 of Multicrystal Module Production”, Minsk-Naroch, Belarus, 2000, pp. 71–74 (in Russian). [27] G. Sberveglieri, L. Depero, S. Groppelli, P. Nelly, WO 3 sputtered thin films for NO x monitoring, Sens. Actuators, B Chem. 26–27 (1995) 89–92. [28] M. Penza, M.A. Tagliente, L. Mirenghi, C. Gerardo, C. Martucci, G. Cassano, Tungsten trioxide (WO 3 ) sputtered thin films for a NO x gas sensor, Sens. Actuators, B Chem. 50 (1998) 9–18. [29] C. Lemire, D.B.B. Lollman, A. Al Mohammad, E. Gillet, K. Aguir, Reactive rf magnetron sputtering deposition of WO 3 thin films, Sens. Actuators, B Chem. 84 (2002) 43–48. [30] C. Bittencourt, R. Landers, E. Llobet, X. Correig, J. Calderer, Role of oxygen partial pressure and annealing temperature on the formation of W O bonds in WO 3 films, Semicond. Sci. Technol. 17 (2002) 522–525. [31] R. Ionescu, E. Llobet, X. Vilanova, J. Brezmes, J.E. Sueiras, J. Calderer, X. Correig, Quantitative analysis of NO 2 in the presence of CO using a single tungsten oxide semiconductor sensor and dynamic signal processing, Analyst 127 (2002) 1237–1246. [32] J. Hub ´ alek, K. Malysz, J. Pr ´ a ˇ sek, X. Vilanova, P. Ivanov, E. Llobet, J. Brezmes, X. Correig, Z. Sv ˘ er ´ ak, Pt-loaded Al 2 O 3 catalytic filters for screen-printed WO 3 sensors highly selective to benzene, Sens. Actuators, B, Chem. 101 (2004) 277–283. [33] T.T. Bardin, J.C. Pronko, R.C. Budhan, J.S. Lin, R.F. Bunshah, The effect of oxygen concentration in sputter-deposited molybde- num films, Thin solid films 165 (1988) 243–247. . Sensors and Actuators B 109 (2005) 128–134 New technology of metal oxide thin film preparation for chemical sensor application Viacheslav. properties of metal oxide sensors, the aim of this paper is to study the influence of in- terrupting the deposition process on the sensing properties of WO 3 thin