e B A 9 fil diss (A tem spo ith w en ms osition many e semic to oper adsorbe ease th e resis ns and operate between 200 and 400 °C [2,3]. TiO2 sensors use the same potential to influence sensor stability [13]. In the present work, we use Thin Solid Films 519 (2010) 434–438 Contents lists available at ScienceDirect Thin Soli w.e sensingmechanism as SnO2; however they are operational up to 600 °C andhave the potential to be thematerial of choicenear this temperature [4,5]. Current commercially available sensors are in the form of bulk or thickfilm ceramicmaterials. Recently, thin film gas sensors have shown large increases in response over that of their thick film counterparts [6]. This has enabled sensitivities to increase and detection limits to be as a sol–gel spin coating method to create thin films of TiO2 on alumina substrates which show stable gas sensing up to 600 °C. This technique allows control of the processing temperature which determines grain size and crystallinity in the films. In this work we show the effects of these two material properties on the sensing characteristics of TiO2 thin films which show good stability and sensitivity during 600 °C operation. low as parts per billion (ppb) [7]. Thin films al sensor to be fabricated, which can operate on and in sensor arrays. Specifically, TiO2 thin fi shown to be sensitive to hydrogen, carbon ⁎ Corresponding author. E-mail address:
[email protected] (S. Bose). 0040-6090/$ – see front matter © 2010 Elsevier B.V. A doi:10.1016/j.tsf.2010.07.040 ical modifications on tin vailable sensors able to We have previously shown that crystallinity is a critical material property affecting the sensing response of TiO2 thick films and has the layer conduction barrier. Structural and chem oxide (SnO2) have enabled commercially a Scanning electron microscopy 1. Introduction The need for real time gas comp monitor and control the efficiency of sparked research interest inmetal-oxid to their inherent stability and ability harsh environments [1]. Oxygen ions, create a surfacedepletion layer and incr Presence of a reducing gas lowers th reacting with the adsorbed oxygen io measurements to help industrial processes has onductor gas sensors due ate in high temperature d on the oxide particles, e resistance of the sensor. tance of the sensors by reducing the depletion while operating at high temperatures [8]. Processing conditions such as substrate material, film thickness, and operating temperature are all known to affect the gas sensing performance [9]. Nanoporous films created by anodization of Ti were also sensitive to hydrogen [10]. Addition of metal catalysts such as platinum is also possible and can improve the gas sensitivity [11]. Thin films, however, are inherently less stable and show reduced sensing performance at temperatures above 300 °C [9–12]. Improved material properties in thin film sensors are required for high operating temperature gas sensor arrays which are stable and show improved gas response [1]. so enable a smaller sized less power consumption lms have recently been monoxide, and methane 1.1. Experimenta Previously, w lead zirconate t modified version of TiO2 on silicon ll rights reserved. X-Ray diffraction Carbon monoxide Methane Titanium dioxide thin films for high temp Zachary Mark Seeley, Amit Bandyopadhyay, Susmita School of Mechanical and Materials Engineering, Washington State University, Pullman, W a b s t r a c ta r t i c l e i n f o Article history: Received 4 December 2009 Received in revised form 21 June 2010 Accepted 8 July 2010 Available online 14 July 2010 Keywords: Titanium dioxide Gas sensors Thin film Sol–gel Spin coating Titanium dioxide (TiO2) thin of titanium isopropoxide electroded aluminum oxide temperature and operating energy, and gas sensing re anatase phase was found w 900 °C. Grain size increased crystallite size and phase. S operating temperatures. Fil toward CO. j ourna l homepage: ww rature gas sensors ose ⁎ 9164-2920, United States m gas sensors were fabricated via the sol–gel method from a starting solution olved in methoxyethanol. Spin coating was used to deposit the sol on l2O3) substrates forming a film 1 μm thick. The influence of crystallization perature on crystalline phase, grain size, electronic conduction activation nse toward carbon monoxide (CO) and methane (CH4) was studied. Pure crystallization temperatures up to 800 °C, however, rutile began to form by ith increasing calcination temperature. Activation energy was dependent on sing response toward CO and CH4 was dependent on both calcination and crystallized at 650 °C and operated at 450 °C showed the best selectivity © 2010 Elsevier B.V. All rights reserved. d Films l sev ie r.com/ locate / ts f l details e have used sol–gel processing to create thin films of itanate on silicon wafers [14,15]. We have used a of the same processing technique to create thin films wafers and alumina substrates. controlled using ball flow-meters and kept at a constant gas flow rate of 150 sccm throughout the experiment. Starting gas reagents were dry high purity air, nitrogen, 1000 ppm CO in nitrogen, and 1000 ppm CH4 in nitrogen. Details and a diagram of the sensing set-up can be found elsewhere [13]. Duplicates of each film were fabricated and tested in the sensing chamber in order to verify results and gain knowledge of reproducibility. The standard deviations of these results are reported in the form of error bars to show differences between samples. Calculation of selectivity (S) toward CO with respect to CH4 is defined here as the ratio of responses to 500 ppm of each target gas: S = CO500 CH 5004 : ð2Þ 2. Results and discussion 435Z.M. Seeley et al. / Thin Solid Films 519 (2010) 434–438 1.2. Precursor solution Titanium-based metal-organic sols were created by dissolving titanium (IV) isopropoxide (Ti[OCH(CH3)2]4 97% min., Alfa Aesar) in 2-methoxyethanol (CH3OCH2CH2OH 99.3+%, Alfa Aesar) in an argon atmosphere. Ionic concentration of titanium was kept at 0.5 M. The solution was refluxed in an oil bath at 120 °C for 6 h under constant argon flow to ensure homogeneity. Thermogravimetric analysis and differential scanning calorimetry (TGA and DSC, NETZSCH STA 409 PC, Germany) were performed on the precursor solution to determine calcination and crystallization temperatures at a heating rate of 5°/ min to 1500 °C. 1.3. Film fabrication Thin films were prepared on two different substrates. A silicon wafer (100) was used as a smooth substrate to measure film thickness and an alumina substrate with interdigitated gold electrodes (Case Western Reserve University: Center for Micro and Nano Processing) was used to measure film resistance. Film deposition was accom- plished via spin coating the precursor solution onto the substrates. A photo resist spin coater was used at 3000 rpm for 30 s. After each deposition, the filmwas dried on a hot plate at 400 °C for 10 min. After every 4th deposition, the film was calcined between 600 and 900 °C for 10 min. A total of 16 depositions were used to create the TiO2 thin films. 1.4. Film characterization The films on alumina substrates were characterized for phase analysis using X-ray diffraction (XRD, Siemens D500 Kristalloflex, Madison WI) with a Cu-Kα radiation source at settings of 30 mA and 35 kV and a step size of 0.01° 2θ. Crystallite size was measured using the full width at half maxima of the anatase (101) and rutile (110) peaks and calculated using the Scherrer equation [16]. From this data a qualitative comparison of crystallite size wasmade for films calcined at different temperatures. Approximate film thickness was measured by cross sectioning the substrate and film for analysis under a field emission scanning electron microscope (FE-SEM; FEI, Sirion, OR) with an accelerating voltage of 20 KeV. Film microstructure and grain size were determined from FE-SEM images of the top surface. In order to measure temperature dependent resistance and gas sensing responses, films were placed in a 30 mm inner diameter quartz tube running through a tube furnace. A baseline gas composition of 10% oxygen balanced by nitrogenwas introduced through the quartz tube at a flow rate of 150 sccm. Silver leadwires connected to the interdigitated electrodes on the substrate were used to measure the two probe dc electrical resistance of the films using a Keithley SourceMeter 2400. Operating temperature in the tube furnace was held constant at 425, 450, 500, 600, 650, and 700 °C and the resistance was allowed to stabilize for 24 h at each temperature. Baseline resistance was recorded before any gas response measurements were made. The gas response of the films was determined by changing the gas flow composition while measuring the dynamic film resistance. Gas response (R) is defined here as the ratio of the film resistance in the presence of the target gas (Rg) to the film resistance in the background gas (Ro): R = Rg Ro : ð1Þ Oxygen responsewasmeasured by reducing the oxygen content to 2% balanced by nitrogen. Carbon monoxide and methane responses were measured by separately introducing 100 and 500 ppm of each target gas into the background flow of 10% oxygen balanced by nitrogen. Gas composition flowing through the quartz tube was 2.1. Crystallization and microstructure DSC and TGA profiles for the precursor solution are shown in Fig. 1. The endothermic peak at 110 °C corresponding to a large weight loss is due to the evaporation of solvent. The three exothermic peaks seen at 270, 390, and 500 °C corresponding to small weight losses are due to burning of the organic constituents in the precursor. Above 550 °C, no additional weight loss is observed indicating that calcination is complete; however two exothermic shoulders at 650 and 850 °C remain in the DSC profile. These shoulders are attributed to the crystallization of anatase TiO2 and anatase to rutile transformation, respectively. Lastly, the large exothermic peak above 1000 °C corresponds to grain growth, sintering, and densification in the TiO2. XRD patterns for calcined TiO2 thin films on alumina substrates are shown in Fig. 2. Anatase (JCPDS# 73-1764) is the only crystalline phase present for calcination temperatures between 600 and 800 °C. The anatase crystallite sizes, given in Table 1, increasewith calcination temperature. This indicates that anatase crystallization and grain growth are taking place up to 800 °C. Calcination at 900 °C results in a significant amount of rutile (JCPDS# 87-0710) formation along with a dramatic increase in crystallite size. Analysis from both DSC and XRD results conclusively agree that anatase is crystallizing between the temperatures of 600 to 800 °C, and the anatase to rutile transformation is taking place between 800 and 1000 °C. The anatase phase in our thin films is thermally stable to 800 °C compared with other thin films created by sputtering, particulate sol–gel, or butoxide-based sol–gel synthesis which show rutile formation at temperatures between 550 and 800 °C [11,17–19]. Fig. 1. Thermogravimetric analysis and differential scanning calorimetry profiles for titanium isopropoxide in methoxyethanol precursor solution after refluxing at 120 °C. Fig. 3. SEM micrographs of TiO2 thin film cross sections. (A) Film on alumina substrate with gold electrodes, and (B) film on smooth Si wafer. 436 Z.M. Seeley et al. / Thin Solid Films 519 (2010) 434–438 These results indicate the potential for our films to be more stable and show good sensing properties at higher operating temperatures. Cross section FE-SEM images of the TiO2 film on alumina and silicon substrates are shown in Fig. 3. Measurements from both substrates suggest a film thickness between 800 and 1000 nm for a film created by 16 depositions. These results agree with our earlier findings on similar films that each spin coated deposition produces approximately 60–70 nm [14,15]. FE-SEM images of thefilm top surface are shown in Fig. 4 to show the TiO2 grain size and microstructure after calcination between 600 and 900 °C. Filmsdonot contain any significant amount of porosity although some cracking is visible due to nonuniform shrinkage during the calcination and crystallization processes. Grain size is seen to increase with increasing temperature as summarized in Table 1. During calcination between 600 and 700 °C, grains are becomingmore defined corresponding to the crystallization of anatase and only a small increase in grain size is observed. Comparison between calcination at 700 and 800 °C reveals that a significant amount of anatase grain growth is occurringduring this temperature range. Calcination at900 °Cproduced large clearly visible grains corresponding to rutile formation. This agrees with previous findings that anatase to rutile transformation is accompanied by significant grain growth [20,21]. From these results, it is concluded that anatase crystallization is occurring between 600 and 700 °C, anatase grain growth between 700 and 800 °C, and anatase-to- rutile phase transformation between 800 and 900 °C. 2.2. Electronic and sensing properties Fig. 5 shows the resistance of 700 °C calcined film at operating temperatures ranging between 425 and 700 °C in the baseline gas environment of 10% O2. Arrhenius plots (Fig. 5 inset) with a linear fits Fig. 2. X-ray diffraction patterns for TiO2 thin films on alumina substrates crystallized between 600 and 900 °C. were used to determine the activation energies for conduction in the TiO2 thin films calcined at different temperatures, summarized in Table 1. A steady increase in activation energy from 0.93 to 1.23 eV is noticed as film calcination temperature increases from 600 to 800 °C. These results agree with similar findings that activation energy Table 1 Crystallite size, particle size, crystalline phase, and activation energy for thin films calcined between 600 and 900 °C. Calcination temperature (°C) XRD crystallite size (nm) FE-SEM grain size (nm) Crystalline phase Activation energy — EA (eV) 600 28 60 Anatase 0.93 650 30 75 Anatase 1.16 700 30.9 80 Anatase 1.21 800 39.6 130 Anatase 1.23 900 51 150 Anatase+ Rutile 1.16 increases with increased grain size due to the lower enthalpy of vacancy formation near grain boundaries [22,23]. However, the 900 °C calcined film shows a further increase in grain size but a decrease in activation energy. This result suggests that the formation of rutile is responsible for the reduction in activation energy. The band gap of rutile (3.0 eV) is smaller than that of anatase (3.2 eV) [24]. Therefore, the reduction in activation energy may be a consequence of the vacancy donor sites being located closer to the conduction band in the smaller band gap of rutile. The dynamic sensing plot showing the change in sensor resistance for various O2, CO, and CH4 concentrations operated at 600 °C, is shown in Fig. 6 for the film calcined at 700 °C. The film shows a stable baseline resistance signifying a potential for long term operation at elevated temperatures. Response kinetics are fast showing full recovery between target gas insertions. 437Z.M. Seeley et al. / Thin Solid Films 519 (2010) 434–438 Gas responses toward O2, CO, and CH4 are shown in Fig. 7 for films calcined between600and900 °Candoperatedbetween425 and700 °C. Three fundamental observations can be made from these results. First, all sensors showed poor response to all target gasses when operated at 650 and 700 °C. At this temperature it is likely that the adsorbed oxygen ions are not stable on the TiO2 surface and are desorbed without the reaction with a reducing species [25]. Lack of these adsorption/ desorption reactions is responsible for the reduced gas response. Second, films calcined at lower temperatures (600–700 °C) showed a much better response to all target gasses compared with films calcined at high temperatures (800–900 °C). This can be attributed to the smaller crystallite size providing more surface area for gas interaction in the films calcined at lower temperatures. Also, lower activation energies allow a higher concentration of electrons in the conduction band which Fig. 4. FE-SEM micrographs of the top surface of thin films calc Fig. 5. Resistance as a function of operating temperature in the background gas atmosphere of 10% O2 balanced by N2 for the TiO2 thin film calcined at 700 °C. Arrhenius plot (inset), the slope is used to calculate activation energy. take part in the sensing mechanism. Third, maximum CO response occurred between 450 and 500 °C while the maximum CH4 response is seen at 600 °C. A similar trend has been observed for other metal-oxide sensors and is attributed to the higher energy required to decompose CH4 [26,27]. From these observations,we conclude that both processing and operating temperatures play important roles in the gas sensitivity and selectivity of the films. In general, films showed better response to CO and poor response to CH4 similar to previous findings for TiO2 sensors [5]. The selectivity toward CO, defined by Eq. (2), is plotted in Fig. 8 for all film calcination and operation temperatures. The film calcined at 900 °C showed minimal CO selectivity due to the universally poor gas response. The films calcined between 600 and 800 °C are most selective toward CO during operation at 425–500 °C, with the film calcined at 650 and operated at 450 °C showing the best selectivity. Selectivity of this film ined at (A) 600 °C, (B) 700 °C, (C) 800 °C, and (D) 900 °C. Fig. 6. Oxygen, carbon monoxide, and methane kinetic sensing patterns for TiO2 thin film calcined at 700 °C and operated at 600 °C. is attributed to the different thermal energies required to oxidize CO and CH4. This film consistently showed the best material properties for gas sensors: high anatase crystallinity, small particle size, uniform film coverage and low activation energy. 3. Summary Thin films composed of pure TiO2 were fabricated via the spin Fig. 7. Thin film response towards 2% oxygen, 500 ppm carbon monoxide, and 500 ppm methane as a function of calcination and operating temperatures. Background gas was 10% oxygen balanced by nitrogen. Fig. 8. Thin film selectivity toward CO with respect to CH4 for films calcined between 600 and 900 °C and operated between 425 and 700 °C. CO selectivity defined by Eq. (2). 438 Z.M. Seeley et al. / Thin Solid Films 519 (2010) 434–438 coating sol–gel method from isopropoxide starting reagents. Sixteen depositions led to a film thickness of 1 μm. Films formed pure anatase phase up to 800 °C and rutile began to form by 900 °C. Increased crystallization temperature induced grain and particle growth which reduced the response to CO. Grain size, crystallinity, and crystal structure had competing effects on the electron conduction activation energy of the films. Thin films displayed stable behavior at high temperatures and good sensitivity toward carbon monoxide. Selec- tivity toward CO with respect to CH4 was found to be dependent on operating temperature indicating that different thermal energies are required for reaction with the two gasses. These films show promise for stable operation of thin film sensing elements at high operating temperatures and the potential for selectivity toward CO. Acknowledgements Authors acknowledge financial support from theWSU Foundation, Office of Research (OR) and Office of Naval Research (Grant # N00014-1-05-0583). References [1] A. Akbar, P. Dutta, C. Lee, Int. J. Appl. Ceram. Technol. 3 (2006) 302. [2] G. Korotcenkov, Sens. Actuators, B 107 (2005) 209. [3] W. Gopel, K. Schierbaum, Sens. Actuators, B 26 (1995) 1. [4] P. Duttam, A. Ginwalla, B. Hogg, B. Patton, B. Chwieroth, Z. Liang, P. Gouma, M. Mills, S. Akbar, J. Phys. Chem. B 103 (1999) 4412. [5] N. Savage, S. Akbar, P. Dutta, Sens. Actuators, B 72 (2001) 239. [6] K. Zakrzewska, Vacuum 74 (2004) 335. [7] E. Comini, G. Sberveglieri, V. Guidi, Sens. Actuators, B 70 (2000) 108. [8] C. Lu, Z. Chen, Sens. Actuators, B 140 (2009) 109. [9] I. Al-Homoudi, L. Rimai, R. Niak, R. Baird, G. Auner, G. Newaz, Mater. Res. Soc. Symp. Proc. 888 (2006) 283. [10] A.Z. Sadek, J.G. Partridge, D.G. McCulloch, Y.X. Li, X.F. Yu, W. Wlodarski, K. Kalantar-zadeh, Thin Solid Films 518 (2009) 1294. [11] M. Epifani, A. Helwig, J. Arbiol, R. Diaz, L. Francisco, P. Siciliano, G. Mueller, J. Morante, Sens. Actuators, B 130 (2008) 599. [12] D. Mardare, N. Iftimie, D. Luca, J. Non-Cryst. Solids 354 (2008) 4396. [13] Z. Seeley, A. Bandyopadhyay, S. Bose, Mater. Sci. Eng. B 164 (2009) 38. [14] T. Myers, P. Banerjee, S. Bose, A. Bandyopadhyay, J. Mater. Res. 17 (2002) 2379. [15] F. Akasheh, T. Myers, J. Fraser, S. Bose, A. Bandyopadhyay, Sens. Actuators, A 111 (2004) 275. [16] B.D. Cullity, J.W.Weymouth, Elements of X-ray Diffraction, 2nd ed. Addison-Wesley, Reading, 1978 284. [17] X. Wenbin, D. Shurong, W. Demiao, R. Gaochao, Physica B 403 (2008) 2698. [18] M. Mohammadi, D. Fray, M. Ghorbani, Solid State Sci. 10 (2008) 884. [19] S. Gao, A. Xian, L. Cao, R. Xie, J. Shang, Sens. Actuators, B 134 (2008) 718. [20] X. Ding, X. Liu, J. Alloys Compd. 248 (1997) 143. [21] X. Ding, X. Liu, J. Mater. Res. 13 (1998) 2556. [22] C. Demetry, X. Shi, Solid State Ionics 118 (1999) 271. [23] B. Huber, A. Brodyanski, M. Scheib, A. Orendorz, C. Ziegler, H. Gnaser, Thin Solid Films 472 (2005) 114. [24] S. Mo, W. Ching, Phys. Rev. B 51 (1995) 13023. [25] C. Park, S. Akbar, J. Mater. Sci. 38 (2003) 4611. [26] D. Kohl, J. Phys. D Appl. Phys. 34 (2001) R125. [27] M. Carotta, S. Gherardi, C. Malagu, M. Nagliati, B. Vendemiati, G. Martinelli, M. Sacerdoti, I. Lesci, Thin Solid Films 515 (2007) 8339. Titanium dioxide thin films for high temperature gas sensors Introduction Experimental details Precursor solution Film fabrication Film characterization Results and discussion Crystallization and microstructure Electronic and sensing properties Summary Acknowledgements References