Thick film titania sensors for detecting traces of oxygen
E. Sotter, X. Vilanova ∗, E. Llobet, A. Vasiliev, X. CorreigMINOS, Department of Electronic Engineering, University Rovira i Virgili, Avda. Paısos Catalans 26, 43007 Tarragona, Spain
Received 27 December 2006; received in revised form 9 May 2007; accepted 10 May 2007Available online 17 May 2007
bstract
The development of pure titania and Nb-doped titania based sensors is reported. Active materials were synthesized via sol–gel and calcined at◦ ◦
emperatures between 600 C and 900 C. The different materials were characterized by XRD, Raman, SEM and Area BET. The samples were
hen deposited on alumina substrates through a thick-film technique. A comparison of the response to traces of O2 in N2 balance of the differentamples was carried out at working temperatures between 300 ◦C and 600 ◦C. The detection of oxygen traces (10 ppm) at a relatively low operatingemperature (500 ◦C) is demonstrated.
2007 Elsevier B.V. All rights reserved.
topa
ptwTai
itgtttnr
eywords: Oxygen sensor; Titania; Thick film
. Introduction
Due to their many advantages such as low cost, small size andobustness, semiconductor sensors appear as a good solutionor oxygen trace detection. In some industrial processes suchs in the production of carbonated beverages, the presence ofxygen must be detected and controlled in the ppm range. Someuthors have reported the detection of oxygen at ppm levelssing gas sensors [1,2]. In most cases the sensors were developedhrough thin film technology. For industrial applications, the
ost widespread is thick film technology because its costs areower when small/medium series of sensors are fabricated [1].dditionally, thick film technology is more flexible when small
mounts of catalysts or sensitizers need to be included in the gasensitive film [3–5]. However the detection of traces of oxygens still a very difficult goal to reach using thick film sensors,nd usually high operating temperatures (>700 ◦C) are needed.lthough a thick film oxygen sensor working at 400 ◦C was
eported by Sharma et al. [6], the detected oxygen concentrationas near 1200 ppm, which is rather high for most applications.Titanium dioxide is the semiconductor material most widely
sed for oxygen detection [7–10]. Titania (usually in a rutilehase) based sensors are bulk conductivity sensors. The oxygenetection mechanism implies the diffusion of oxygen ions in
he bulk of the material and this occurs provided the material isperated at high temperatures (700–1000 ◦C). This leads to highower consumption, which is not desirable for most electronicpplications.
On the other hand, titania showing an anatase crystallinehase has more free electrons than rutile titania [11]. For anatseitania, oxygen detection can be associated to a surface reaction,hich takes place at lower temperatures (400–500 ◦C) [12,13].hen, it can be derived that keeping an anatase structure wouldllow for the detection of oxygen at lower temperatures, whichs desirable for sensor design [6,14].
When titania is doped with pentavalent ions, e.g., Nb5+, suchons get into the anatase titania crystalline structure, giving riseo a hindering in the phase transition to rutile and an inhibition inrain growth. While in undoped titania, the change from anataseo rutile starts at about 600 ◦C, in doped titania, the transitionemperature is higher, around 750 ◦C. This effect is attributedo the extra valence of niobium ions in comparison with tita-ium ones, which reduces oxygen vacancies in the anatase phase,etarding the transformation to rutile [15–17].
Furthermore, grain growth is inhibited due to the stressnduced in the anatase structure by the substitutional Nb5+ ions,ith a slightly higher ionic radius value with respect to Ti4+.maller grains imply more active area, which increases the sur-
ace to volume ratio and thus sensitivity [12,17].
It has been reported that Nb-doped titania shows higher sensi-ivity towards oxygen and shorter response time than pure TiO218]. The doped material also shows lower impedance at low
dtlfutas small as 1.5 mm × 0.3 mm × 0.15 mm. The thickness of thealumina substrate was decreased down to 0.1 mm. The sensorchip was bonded to TO-8 package. The diameter of the packagewas ∼10 mm [21].
68 E. Sotter et al. / Sensors and
perating temperatures and hence, the design of the associatedlectronic circuitry is simpler [6]. Arbiol [15,16] reported thathere is an optimum for the concentration of Nb in TiO2. The besttomic ratio between Nb and Ti was found to be 3%. However,he gas sensing properties of this material were not investigatedn those papers.
In this paper, pure titania and niobium doped titania nanopow-ers were synthesized by a simplified sol–gel route. In order toet the crystalline structure of the active materials, they werealcined at four different temperatures: 600 ◦C, 700 ◦C, 800 ◦Cnd 900 ◦C. The obtained materials were characterized by differ-nt techniques. The objective of these characterizations was tobtain information about the material structure (chemical com-osition, crystalline phase, surface area, porosity and grain size)hat could be related to its detection properties.
To determine the sensing capabilities of each material, theyere deposited over alumina substrates that included contacts
nd a heater. Then they were tested under 20 ppm, 15 ppm and0 ppm of O2 in N2 at different temperatures between 300 ◦C and00 ◦C. These oxygen concentrations are far lower than thoseypically reported in the previous literature (i.e., near 1000 ppm).
The organization of the paper is as follows. In the next sec-ion, the synthesis of the gas sensitive materials is discussed,etails on the fabrication of the sensors are given and the methodsmployed to characterize the phase, morphology, compositionnd oxygen sensitivity of the active films are reviewed. In thehird section the results of the different characterizations are pre-ented and discussed. Finally a summary of the main results cane found in the conclusions.
. Experimental
.1. Material synthesis
Non-doped TiO2 samples were synthesized through a sol–geloute, starting from an alkoxide precursor. Following the exper-ment done by Ruiz [17,19], Titanium(IV) isopropoxide, alsoalled tetraisopropyl orthotitanate Ti[OCH(CH3)2]4 99% purity,as mixed with isopropanol to get a 0.5 M solution to avoid earlyrecipitations of oxides.
In the case of Nb-doped TiO2, niobium ethoxide,b(OC2H5)5 99.99%, was also added to a 0.5 M solution of
etraisopropyl orthotitanate mixed with isopropanol in the appro-riate concentration to obtain an Nb/Ti atomic ratio of 3%, whichas the optimal doping concentration found by Arbiol [15,16].Due to the fact that the organic precursors were prone to oxi-
ation, this process was conducted under inert atmosphere. Thenert atmosphere was created by using an inflatable polyethy-ene chamber. The air inside the bag was taken out from itvacuum was created) and nitrogen was introduced later. Thisrocedure was repeated three times. After the precursors wereiluted in isopropanol, the process could continue under normaltmosphere.
Another solution of water and nitric acid, HNO3 70%, wasrepared in parallel. Then the mixture of organic precursorsiluted in isopropanol was added dropwise to the acid solutionnder stirring. The final composition of the constituent was set to
Fch
tors B 127 (2007) 567–579
atisfy [Ti]:[HNO3]:[H2O] = 1:1:100 in molar ratio. In the casef hydrolysis-condensation of titanium isopropoxide, increasinghe initial water concentration produced higher nucleation rates,hich resulted in a decrease in average particle size. Nagpal et
l. [20] showed that the main factor on particle size is the effectf water concentration. Based on this report and in Ruiz experi-ent [19], we selected a [H2O]/[Ti] ratio of 100. Because of the
arge amount of nitric acid, the hydrolysis proceeded withoutorming a precipitate, giving rise to a transparent sol at pH 1.
Afterwards, the pH of the sol was increased carefully bydding dropwise approximately 30 ml of an aqueous solution1 M) of ammonium hydrogen carbonate (pH 9), until a consis-ent gel was achieved.
The gel was dried in an oven UNE 300 from Memmert Co.irst, the temperature was set to 120 ◦C for 20 h to evaporateater. Then, temperature was increased to 250 ◦C for 10 h to
liminate some compounds generated during synthesis. At thend, the gel was turned into big agglomeration of particles, inorm of rocks. These big particles were milled to reach powderppearance.
The firing was carried out in a programmable muffle CarbiliteWF 1200. The synthesized material was, once milled, split in
our parts to be fired in air at 600 ◦C, 700 ◦C, 800 ◦C and 900 ◦C,espectively. A temperature rise of 10 ◦C/min, taken from Arbiolt al. [16], was applied to reach each firing temperature. Then,he samples remained at firing temperature for 2 h. To finish, aree cooling rate was applied to the materials.
.2. Sensor fabrication
The sensor substrate was fabricated by multilayer thick filmeposition of a Pt based heater, an insulation layer and con-acts to the sensing layer. A ruthenium oxide diffusion barrierayer was employed to prevent inter-diffusion between the Ptrom the contact pads and the gas-sensitive metal oxide. These of a two-side construction (Fig. 1) allows for miniaturizinghe sensor chip. The final dimensions of the sensor chip were
ig. 1. Structure of the substrate. 1: alumina substrate, 2: sensing layer, 3: Ptontact pads, 4: connecting wires, 5: ruthenium oxide diffusion barrier, 6: Pteater and 7: Pt contact pads.
Actua
aTttplhsam
2
dptci
bw((Pasr7
rosXtθ
m0awsXwma
w
wVc[A
ti
caytiwyeTWst
sepadSfet
dSthsepmfiF
sotnafmwwgactflaasci
E. Sotter et al. / Sensors and
The TiO2 based nanopowders were dispersed in glycerolnd then the resulting paste was dropped over the electrodes.he as-deposited films were dried at 300 ◦C using a tempera-
ure ramp. A temperature rise of 20 ◦C/min was used to avoidhe occurrence of cracks in the metal oxide layer. A computerrogrammable power supply was employed. Finally, the activeayers were annealed for 2 h. Sensors whose active materialad been calcined at 600 ◦C were annealed at 600 ◦C. The sen-ors that employed materials calcined at 700 ◦C or higher werennealed at 700 ◦C. The drying and annealing processes wereade in situ using the heating element of the sensors.
.3. Sensor characterisation
The concentration of doping atoms (niobium atoms) in theoped titania was determined by using inductively coupledlasma (ICP) spectroscopy. With this technique it is possibleo verify the chemical composition of the samples. Moreover, itan give information about the amount of any noble metal placedn the surface or inside the bulk of a metal oxide sample [22].
The material employed in this analysis was the doped oneefore the firing treatment. For the measurements, the samplesere first treated to get an acid solution. Thus, the powders
50 mg) were dissolved with HCL (1 ml), HF (1 ml) and H2O1 ml) in a pressurized microwave decomposition system (fromAAR Inc). Then, the acid solution of titania was analyzed inn ICP-OES analyzer (induced coupled plasma-optical emissionpectroscopy) from Perkin-Elmer model OPTIMA 3200RL. Theadiofrequency source was 40 MHz working at a power between50 W and 1500 W.
The analysis of phase transition was studied by using X-ay diffraction (XRD). With this technique it is possible notnly to distinguish the presence of different phase states in theample analyzed, but also to quantify each crystalline phase.RD measurements were made using a Siemens D5000 diffrac-
ometer (Bragg–Bretano parafocusing geometry and vertical–θ goniometer) fitted with a curve graphite diffracted-beamonochromator, incident and diffracted-beam Soller slits, a
.06◦ receiving slit and scintillation counter as a detector. Thengular 2θ diffraction range was between 20◦ and 45◦. The dataere collected with an angular step of 0.05◦ at 3 s per step and
ample rotation. Cu K� radiation was obtained from a copper-ray tube operated at 40 kV and 30 mA. Quantitative analysisas performed by employing the Rietveld XRD quantificationethod, in which the weight fraction wp for phase p is calculated
s follows:
p = (SpMpVp/Bp)∑Np
p=1(SpMpVp/Bp)(1)
here Np is the number of crystalline phases present, and Sp, Mp,p and Bp are the Rietveld scale factor, the unit cell mass, the unitell volume and the Brindley correction for phase p, respectively23]. For quantitative analysis, the TOPAS software from Bruker
dvanced X-ray Solutions was used.Raman technique was also employed to analyze the phase
ransition in doped materials. Raman spectroscopy is basedn the diffusion processes given by the electronic polarization
i
S
tors B 127 (2007) 567–579 569
aused by ultraviolet or visible light. Raman gives informationt molecular level allowing the vibrational and rotational anal-sis of chemical species. Briefly, the radiation interacts withhe molecular vibrational energy levels due to the light scatter-ng phenomena, in which the electromagnetic radiation interactsith a pulsating polarizable electron cloud. For the Raman anal-ses we used a Raman spectrometer Jobin Yvon T64000. Thexcitation source was a visible laser Ar+ coherent INNOVA 300.he microscopy was Olympus BH2 with high spatial resolution.e used a triple monochromator (1800 g/mm) and a bidimen-
ional CCD detector cooled with liquid nitrogen. The power ofhe Raman laser was set at 200 W.
The morphology of the active layers was investigated usingcanning electron microspcopy (SEM). A JSM 6400 scanninglectron microscope was used. Before SEM analysis, the sam-les were coated with a thin gold layer, which was sputtered tovoid charging effects. Micro-images were taken using a focalistance of 8 mm, and a voltage of 25 kV. Grain size data fromEM pictures were obtained by using the software ImageJ 1.33urom Wayne Rasband. One hundred grains were considered inach analysis for the results to be statistically representative ofhe active films.
Area BET measurements were performed on the nanopow-ers to know the surface area and the porosity of each material.ince the interaction between the active material and the gas
akes place on the surface, the knowledge of this parameter mayelp in the selection of the best material to be used for the oxygenensor. A Micromeritics ASAP2020 gas sorption analyzer wasmployed for determining surface area, porosity, pore size andore distributions of the titania nanomaterials. One gram of eachaterial was used for the analyses. These samples were degasi-ed during 2 h at 90 ◦C and then for additional 6 h at 150 ◦C.inally, the adsorption of N2 was measured at 77 K.
To study the oxygen sensing properties of the different sen-ors, these were placed in an airtight test chamber (with a volumef 16 cm3). Three sensors of each material were employed forhis experiment. Before starting a set of measurements, pureitrogen (certified to contain less than 7 ppm of oxygen) wasllowed to continuously flow through the measurement systemor 12 h to ensure that oxygen was flushed out. Every measure-ent consisted of two stages. In the first stage, pure nitrogenas let to flow through the test chamber and the sensor baselineas established. Then, a calibrated mixture of nitrogen and oxy-en was mixed with pure nitrogen using mass-flow controllersnd the resulting mixture was allowed to flow through the testhamber. The total flow was set constant to 140 ml/min duringhe whole measurement process. The accuracy of each mass-ow meter was +1% of its full scale. The sensors were testedt four different operating temperatures: 300 ◦C, 400 ◦C, 500 ◦Cnd 600 ◦C. Three replicates of each measurement for each sen-or were done. The change in resistivity of the active layersaused by the presence of oxygen was measured by employ-ng a Keithley 6517 A electrometer. The sensor response (SR)
s defined as follows:
R = �R
RN2
(2)
5 Actua
wonot
3
3
darvc
3
mtta
arwaiatbp
itos
ooact
70 E. Sotter et al. / Sensors and
here �R is the change in sensor resistance in the presencef oxygen and RN2 is the resistance of the sensor under pureitrogen. In the first step, the response was evaluated to 20 ppmf oxygen. Then, the material with the higher response was alsoested at 15 ppm and 10 ppm of oxygen.
. Results and discussion
.1. ICP analysis
As explained in Section 2.3, the content of Nb in niobium-oped titanium oxide was estimated by ICP analysis. Thisnalysis revealed that the content of niobium in the doped mate-ial was 2.75 ± 0.2 at.%. This result is very close to the nominalalue of 3 at.% (i.e., the value sought). Therefore, we can con-lude that the doping process was successful.
.2. XRD analysis
A comparison between the XRD patterns for the different
aterials studied is shown in Fig. 2. In this figure it is possible
o see the fast evolution of phase in pure titania with calcinationemperature, since in this material the change of phase fromnatase to rutile is almost complete at 700 ◦C. The quantitative
t
oa
Fig. 2. XRD patterns and quantitative analysis of pure (above) and Nb-do
tors B 127 (2007) 567–579
nalysis shows that at 600 ◦C, pure titania comprises anatase andutile phases in a similar proportion. No trace of brookite phaseas found at this temperature. At 700 ◦C, the anatase phase has
lmost disappeared and we can consider that the whole materials in rutile phase. At 800 ◦C the sample is 100% in rutile phasend there is no further change in the crystalline structure ofhe material when temperature is increased to 900 ◦C. This isecause the rutile phase is completely stable and the change ofhase is irreversible.
Making an extrapolation of the quantitative analysis results,t can be stated that phase transition in pure titania started atemperatures of about 500 ◦C or even lower. This early changef phase means an almost complete stabilization of its crystallinetructure near 700 ◦C.
Considering the XRD pattern of Nb-doped titania, the effectsf the addition of niobium ions to the titania lattice can bebserved. Unlike in pure titania, in Nb-doped titania calcinedt temperatures below 700 ◦C, the anatase phase is the mainrystalline structure. At 800 ◦C there is still a significant quan-ity of anatase phase in the material (12%). The complete change
o rutile does not occur up to 900 ◦C.
Arbiol [15,16] gives the physical explanation of the effectsf adding Nb ions to the TiO2 lattice on phase transition fromnatase to rutile. According to Arbiol [15,16], during the doping
ped titania (below), calcined at 600 ◦C, 700 ◦C, 800 ◦C and 900 ◦C.
Actuators B 127 (2007) 567–579 571
pT(itNvcctOfaitbr
dpttpaipi
ict
tdl
ttf
d
wd((iXgofgtisrr
Fct
iaaosIsh98(nF69
smi
3
aecXfI9dc
E. Sotter et al. / Sensors and
rocess, Nb5+ ions enter substitutionally in the lattice, occupyingi4+ vacancies [24] without affecting the basic TiO2 structureanatase or rutile) [25] since the ionic radius of Nb5+ (0.70 A)s comparable to that of Ti4+ (0.68 A) [1,6,26,27]. But, in ordero assure the charge neutrality of the resulting material, theseb5+ ions should be compensated by a decrease in oxygenacancies. In anatase–rutile transformation, the anatase pseudo-lose-packed planes of oxygen {1 1 2} are retained as the rutilelose-packed planes {1 1 0} and a cooperative rearrangement ofitanium and oxygen ions occur within this configuration [28].xygen vacancies placed in anatase planes act as nucleation sites
or the anatase to rutile phase transformation [29], and thereforeminimum ratio of oxygen vacancies in anatase {1 1 2} planes
s needed in order to assure the phase transition process. Then,he phase inhibition is attributed to the extra valence in nio-ium ions, which reduces oxygen vacancies in anatase phase,etarding the transformation to rutile [26,30].
Considerable quantities of brookite phase were found in Nb-oped titania calcined at 600 ◦C (25%) and 700 ◦C (9%). Thishase is usually found before the formation of anatase (first) andhen rutile crystalline states. However it has been reported thathe brookite phase can change directly to rutile phase withoutassing through anatase phase [31]. The changes in the percent-ges of crystalline phases observed at 700 ◦C suggest that theres a partial transformation of brookite phase directly into rutilehase. At 800 ◦C there is no longer evidence of brookite phasen the material.
The brookite phase has less free carriers than anatase andts electrical behavior is very alike to rutile [32]. Its presenceould hinder the response of titania towards oxygen at workingemperatures near 500 ◦C.
The presence of anatase phase is higher in Nb-doped TiO2han in pure TiO2. Therefore it could be anticipated that Nb-oped TiO2 appears to be more suitable for oxygen detection atower temperatures (400–600 ◦C).
From the XRD patterns shown in Fig. 2, it is also possibleo see that the informative peaks of Nb-doped titania are widerhan those of pure TiO2. Crystallite size, d, can be estimatedrom the XRD patterns by the Scherrer equation:
= kλ
D cos θ(3)
here k is a constant the value of which varies from 0.7 to 1.7epending on crystallite shape, λ is the wavelength of Cu K�i.e., 1.54056 A), D is defined as the full width at half maximumFWHM) and θ is the Bragg angle. Considering that the FWHMs inversely proportional to crystallite size, a wider peak in theRD pattern implies a smaller crystallite, and hence a smallerrain. Fig. 3 compares the FWHM (main peak of each phase)f the crystalline structure of each material synthesized as aunction of calcination temperature. It can be concluded that therowth of TiO2 crystallites with a rise in calcination tempera-ure is hindered by the addition of Nb. This is also due to the
ncorporation of Nb in the titania lattice. Niobium ions occupyubstitutionally titanium sites due to the similarity of their ionicadii. Nevertheless, Nb5+ radius (0.70 A) is a bit larger than Ti4+
adius (0.68 A). The result of this is that Nb atoms induce stress
cahr
ig. 3. Comparison among the full width half maximum (FWHM) of eachrystalline phase of pure and Nb-doped TiO2 as a function of the calcinationemperature.
n the Titania lattice, which may hinder the growth of anatasend rutile phase crystallites in the material, as found by Sharmand Bhatnagar [1]. These conclusions are supported by the studyf Williamson-Hall plots (not shown, but derived from the datahown in Fig. 2), which reveal crystallite size and strain effects.n pure titania samples, the microstrain remains almost con-tant (near 1300) regardless the firing temperature. On the otherand, the crystallite size increases steadily (from 20 nm to near5 nm) when the firing temperature is increased from 600 ◦C to00 ◦C. For Nb-doped titania samples, the microstrain is highernear 3000 for samples fired at 600 ◦C) and is reduced down toear 1500 when the firing temperature is raised up to 900 ◦C.or these samples, crystallite size increases from 20 nm to near5 nm when the firing temperature increases from 600 ◦C up to00 ◦C.
Small crystallites lead to small grains and therefore to a higherurface area. Then it is expected that Nb-doped titania will haveore surface area than pure titania, which will result in a better
nteraction between the gas and the active material.
.3. Raman analysis
The Raman spectra between 10 cm−1 and 680 cm−1 for TiO2nd Nb-doped TiO2 are shown in Fig. 4. This figure shows thevolution of the crystalline structures in both materials as thealcination temperature is increased. As it was also observed forRD results, Raman results confirm that the change of phase
rom anatase to rutile for pure TiO2 is almost complete at 700 ◦C.n the case of the doped material, phase transition is completed at00 ◦C. After a comparison of the spectra of both TiO2 and Nb-oped TiO2 it is possible to derive that the samples with a higherontent of anatase structure are the Nb-doped TiO2 materials
◦ ◦
alcined at 600 C and 700 C. Considering the samples calcinedt 800 ◦C and 900 ◦C, even those that are Nb-doped present aigh concentration of rutile crystalline phase, which involves aeduction of the quantity of free electrons on the material surface.
572 E. Sotter et al. / Sensors and Actuators B 127 (2007) 567–579
Fig. 4. Raman spectra of TiO2 and Nb-doped TiO2 calcined at different temperatures.
F 0 ◦Cc
1arstdnBsrdc
3
Itt
n
tdat
sIcpure titania calcined at 600 ◦C and Nb-doped TiO2 calcinedbelow 800 ◦C, are typical of mesoporous materials (isothermtype IV). The main feature of these isotherms is the occurrence
Table 1Some Raman features in the range between 700 cm−1 and 1750 cm−1 of Nb-doped titania calcined at 600 ◦C and 700 ◦C
ig. 5. Raman spectrum at high frequencies of Nb-doped titania calcined at 60arbon compounds present on the material surface.
Fig. 5 shows the Raman spectra between 1200 cm−1 and750 cm−1 for Nb-doped TiO2 calcined at 600 ◦C (Fig. 5 left)nd 700 ◦C (Fig. 5 right). The Raman spectra over 1200 cm−1
evealed some peaks corresponding to the presence of carbonpecies on the surface of materials that were calcined at loweremperatures. The concentration of these carbon compoundsecreases as the calcination temperature increases, and at 800 ◦Co carbon compounds were observed on the material surface.ecause there is no other source of carbon after the material
ynthesis, it is supposed that those deposits of carbon are theesidual of the synthesis process, which could not be eliminateduring the calcination. Table 1 summarizes the compounds thatorrespond to the peaks found in those spectra.
.4. BET area
The results from BET area analysis are summarized in Fig. 6.t can be seen that the surface area decreases when calcination
emperature increases. This is attributed to both the growth ofhe material crystallites and the reduction of porosity.
The BET area in Nb-doped TiO2 is higher than in pure tita-ia. This supports the idea that the addition of Nb ions to the
11
1
(left) and 700 ◦C (right). Dashed curves represent the peaks corresponding to
itania lattice hinders the growth of the material crystallites. Inoped titania calcined at temperatures above 800 ◦C, the surfacerea is even lower than for non-doped titania calcined at loweremperatures.
In Fig. 7, the adsorption isotherms for N2 at 77 K arehown for the different materials investigated. According to thenternational Union of Pure and Applied Chemistry (IUPAC)lassification of adsorption isotherms [33–35], the isotherms of
350 D peak of microcrystalline graphite [45,46]490 Contributions from the photon density of states
in finite-size crystals of graphite [44,47]600 G peak of graphite [45,46]
E. Sotter et al. / Sensors and Actua
ota9wfti
wdm
3
via
cao8
tfrcrartog
sloTtTtsfcr
Ttotprra
Fig. 6. BET surface area of the materials.
f hysteresis loops due to capillary condensation during adsorp-ion. On the other hand, isotherms of pure materials calcinedt temperatures above 700 ◦C and doped materials calcined at00 ◦C are attributed to non-porous or macroporous materialsith strong adsorbate–adsorbent interactions (type II). The main
eature of these isotherms is the absence of hysteresis loops. Thushe porosity of titania diminishes when calcination temperaturencreases. The addition of niobium retards this effect.
Taking into account the IUPAC classification of isothermsith hysteresis [34], pure titania calcined at 600 ◦C and Nb-oped titania calcined below 800 ◦C can be classified asaterials with narrow pore size distribution.
.5. SEM analysis
Grain size was estimated for the different materials studiedia an SEM analysis. Fig. 8 summarizes these results. Accord-ng to the histogram of TiO2 calcined at 600 ◦C, the grains havemean diameter of approximately 60 nm. This size is high in
oSt
Fig. 7. Adsorption and desorption isotherms of N2 at
tors B 127 (2007) 567–579 573
omparison with Nb-doped titania calcined at the same temper-ture (∼45 nm). When the temperature increases, the grain sizef pure TiO2 increases steadily until reaching a diameter about3 nm at 900 ◦C.
The increment in the mean grain size is closely related withhe quantity of rutile phase present in the material. The grainsormed by anatase crystals are smaller than those formed byutile crystals. This is because rutile crystals, unlike anataserystals, tend to coalesce when crystallinity is improved untileaching a complete stabilization [15]. The small quantity ofnatase phase in pure TiO2 calcined at 600 ◦C, observed in XRDesults, contributes to retard the grain growth. Therefore, whenhis phase disappears at 700 ◦C, rutile crystals, which are thenly present in the material, start to coalesce forming biggerrains.
At 900 ◦C, the size of the grains in pure titania is almost theame as that at 800 ◦C. It appears that the fast grain growth atower calcination temperatures stops at 800 ◦C. The grain sizef pure TiO2 calcined at 900 ◦C is similar to that of Nb-dopediO2 calcined at the same temperature. This may be attributed
o the early stabilization of the material crystallinity in pureiO2. The pure titania does not have any additive that retards
he phase transition. Thus, the material reaches the crystallinetabilization quickly. When this state is reached, the crystals stopusing with others. This fact can explain why when a determinedalcination temperature is reached, the grain size in pure titaniaemains almost constant.
As it can be seen in Fig. 8, the grain size of Nb-dopediO2 calcined at 600 ◦C, 700 ◦C and 800 ◦C is smaller than
hat of pure TiO2 calcined at the same temperatures. The sizef grains in Nb-doped material calcined at 800 ◦C is almosthe same as that in pure TiO2 calcined at 600 ◦C. This sup-orts the fact that the addition of Nb ions to the titania latticeetards the phase change and inhibits the grain growth. Theseesults are in good agreement with those reported by otheruthors [1,36].
The inhibition of phase transition in titania by the additionf Nb ions was already discussed in Section 3.2. By comparingEM results and those obtained by XRD analysis, it is possible
o conclude that the increase in grain diameter and the content of
77 K of TiO2 (left) and Nb-doped TiO2 (right).
574 E. Sotter et al. / Sensors and Actuators B 127 (2007) 567–579
togra
rpaccaogt
iclt
sa
i(lmbie
Fig. 8. SEM images and his
utile phase are related. The higher the quantity of rutile phaseresent in the sample is, the bigger its grains are. As it waslready stated, the reason for this behavior is the tendency tooalescence shown by rutile crystals but not shown by anataserystals [15]. Therefore, the grains formed by anatase crystalsre usually smaller that those formed by rutile crystals. For mostf the samples studied, anatase and rutile crystals coexist in arain. In these cases, the presence of the anatase phase retardshe growth of the grains.
Nb ions in the crystalline structure of titania play also an
mportant role in the inhibition of grain growth. As already dis-ussed for XRD results, when Nb is incorporated to the titaniaattice, niobium ions occupy substitutionally titanium sites, dueo the similarity of their ionic radii. However, Nb ions induce
swfiT
ms of all studied materials.
tress in the titania lattice, which may hinder the growth ofnatase and rutile phase crystallites in the material.
Grain size in Nb-doped titania grows linearly with an increasen calcination temperature between 600 ◦C (48.9 nm) and 800 ◦C64 nm). The rate is about 7.5 nm each 100 ◦C. At 900 ◦C theinear growth is no longer valid and the grain size suffers a dra-
atic increase. The explanation of this phenomenon was giveny Arbiol [15] and Marien et al. [37], as follows. The Nb5+
ons remain inside the anatase bulk until the temperature is highnough for the niobium ions to gain the necessary mobility to
inter. At such high temperature, the resulting Nb aggregatesould be expulsed out from the anatase structure and would benally located on the TiO2 surface in an oxidized Nb phase.he content of oxidised Nb at the surface of the film would
Actua
betwittassioeT
3
sTdoobfw
Bnatpt
atibr
Tn
tmplaArwmrw
aatiaiufInpt6
bfrfita
Fc
E. Sotter et al. / Sensors and
e below 3%, which is too low to be detected by XRD. How-ver, the results derived from Williamson-Hall plots show thathe microstrain in Nb-doped titania samples clearly diminishesith calcination temperature, reaching at 900 ◦C a value that
s only slightly higher than that for pure titania. This supportshe idea that Nb migrates towards the surface of the film. Athe same time, once the main part of the niobium has left thenatase structure, a fast transformation to rutile would happen,ince at those high temperatures (800–900 ◦C), the phase tran-ition is highly favored. The crystals, now free from the stressntroduced by niobium ions, can grow more easily. Additionally,nce in the rutile phase, crystals will tend to coalesce. This canxplain the sudden increase in grain size observed for Nb-dopediO2 calcined at 900 ◦C.
.6. Oxygen detection
The first material tested was pure titania. The results are repre-ented in Fig. 9 (left). All the responses are rather low (0.1–0.3).iO2 calcined at 700 ◦C shows the worse responses (0.16–0.20),ecreasing when the operation temperature is increased. On thether hand, the response seems to improve when increasing theperation temperature for other cases. This tendency is onlyroken for pure titania calcined at 800 ◦C, in which the responsealls from 0.26 at 500 ◦C to 0.2 at 600 ◦C. The best responsesere obtained from TiO2 calcined at 600 ◦C (0.22–0.27).Based on the previous characterizations (XRD, Raman, Area
ET and SEM), it can be concluded that the low responsive-ess to oxygen found for pure TiO2 based materials could bettributed to the predominance of the rutile phase in their crys-alline structure, more precisely to the low surface area, loworosity and high grain size that are characteristic of this crys-alline phase.
The rutile phase, as it was commented before, is poorly activet temperatures below 700 ◦C. The detection mechanism in rutile
itania based sensors is principally related to the ion vacanciesn the material bulk, which need high working temperatures toe formed. The pure TiO2 samples are mostly constituted by autile phase, in particular those calcined at 700 ◦C and above.
tsbc
ig. 9. Comparison of the responses of sensors made from pure titania (left) and Nbarrier gas employed was pure nitrogen.
tors B 127 (2007) 567–579 575
herefore, the working temperatures employed in this study areot enough to activate the working mechanism of these sensors.
Additionally, the large size of grains, which is associated tohe rutile crystalline phase, also hinders the sensitivity of these
aterials towards oxygen. Active surface area is inversely pro-ortional to grain size, and then when the grains grow there isess surface area to interact with the gas. The detection processlso depends on the quantity of grain-to-grain interfaces [38].s a general rule, the size of grains in pure TiO2 based mate-
ials is larger than those in any other materials tested in thisork. Therefore, the low sensitivity of the sensors based on thisaterial is not a surprising result. Fig. 10 (left) shows a typical
esponse of a sensor based on pure TiO2 calcined at 700 ◦C andorking at 500 ◦C, towards 20 ppm of oxygen in N2.According to XRD results, pure TiO2 calcined at 600 ◦C has
large quantity of anatase crystalline phase (48.83%). Thisnatase phase present in the material should help to improvehe reactivity with oxygen. Nevertheless, results show that thiss not the case. The different crystalline phases (i.e., brookite,natase or rutile) present in titania play a very important rolen its capability to detect oxygen at the operating temperaturessed. However, there are others features such as grain size, sur-ace area and porosity, which may also influence the detection.n the case of pure titania calcined at 600 ◦C, the grain size isear 60 nm (Fig. 8), the surface area is 6 m2/g (Fig. 6) and theorosity is very low (Fig. 7 left). Compared with the same fea-ures in doped materials, the features in pure titania calcined at00 ◦C are not good enough for gas detection.
The responses of sensors fabricated with Nb-doped titaniaased materials towards 20 ppm of oxygen and working at dif-erent temperatures are represented in Fig. 9 (right). The sensoresponses towards oxygen of Nb-doped samples (at differentring temperatures) are significantly better than those of pure
itania based sensors. As it was revealed by XRD, Raman, BETnd SEM results, the addition of niobium ions to the titania lat-
ice modify the physical characteristics of the resulting material,uch as crystalline phase, surface area and grain size, allowing aetter interaction with oxygen. Moreover, these Nb foreign ionsontribute directly to the catalytic process of detection by creat-
-doped titania (right) toward 20 ppm of oxygen, at different temperatures. The
576 E. Sotter et al. / Sensors and Actuators B 127 (2007) 567–579
F itaniaw
i[
wkamc5tahi
iorlcst
aarIipta
stimfcto
(rihp
rdccbtnlmsf
sgttotlc6tsc
7tt2
ig. 10. Typical responses toward 20 ppm of oxygen of sensors made from pure tas 500 ◦C and the carrier gas employed was pure nitrogen.
ng defects in the material that help oxygen ions to be adsorbed1,15,16].
Fig. 9 (right) shows that the best results for Nb-doped TiO2ere obtained for a calcination temperature of 700 ◦C. To ournowledge, this is the first time that good sensitivity is reportedt such low oxygen concentrations. The response of this specificaterial stands out from the rest. Fig. 10 (right) shows the typi-
al response of a sensor obtained from this material working at00 ◦C, towards 20 ppm of oxygen in N2. The average responseime of these sensors (operated at 500 ◦C) is below 60 s and itsverage recovery time is near 90 s. Response and recovery timeseavily depend on the operating temperature (e.g., they increasef the operating temperature decreases).
The response of this sensor increases with a rise in work-ng temperature up to 500 ◦C, where the maximum response isbtained. Then, there is an inflection point in the temperature-esponsiveness curve, obtaining at 600 ◦C a response value evenower than the value obtained at 300 ◦C. This inflection pointould be attributed to a maximum in the adsorption and dis-ociation of oxygen molecules taking place at the optimumemperature (i.e., near 500 ◦C).
Despite the fact that grain size of Nb-doped titania calcinedt 800 ◦C is lower than the one for the same material calcinedt 900 ◦C and that the surface area is higher in the former, theesponses towards oxygen of both materials are rather similar.t has to be pointed out that these two materials are mainlyn rutile phase. So it can be concluded that the crystallinehase is the dominant factor for oxygen responsiveness ratherhan other physical characteristics such as grain size or surfacerea.
On the other hand, Nb-doped titania calcined at 600 ◦Chowed the worse responsiveness. According to XRD results,he quantity of anatase phase in this material is even higher thann the Nb-doped titania calcined at 700 ◦C. The former has also
ore surface area than the latter and its grain size is lower. There-
ore, a higher response of the sensor based on Nb-doped titaniaalcined at 600 ◦C could be expected. A possible explanation forhis low responsiveness can be found in the significant contentf brookite phase found in Nb-doped titania calcined at 600 ◦C
itaV
(left) and Nb-doped titania (right) calcined at 700 ◦C. The working temperature
24.74%). As already stated, the brookite phase has less free car-iers than the anatase phase. Although its electrical behaviours very alike to rutile, the presence of a brookite phase mayave a worse effect than the rutile phase in the oxygen sensingroperties of titania.
Additionally, the high frequency content of Raman spectraeveals another reason for the lower oxygen response of Nb-oped titania calcined at 600 ◦C. In Fig. 5, some peaks thatorrespond to carbon species can be found. The peaks and theirorrespondences are summarized in Table 1. As already statedefore, these carbon deposits are contamination residues fromhe synthesis process that could not be eliminated during calci-ation at such low temperature. An example of this is the peakocated at 1039 cm−1, which clearly matches with one of theain peaks in the spectrum of ammonium carbonate, used for the
ol–gellification. The other carbon compounds could be formedrom the organic part of precursors.
The problem with these carbon deposits is that they cover theurface of the active material and hinder its interaction with oxy-en. Because these carbon structures are also poorly catalytic,he result is a deactivation of the catalysis [39–41]. In conclusion,he considerable quantity of brookite phase and carbon depositsn the surface of Nb-doped titania calcined at 600 ◦C impairshe catalytic properties of the active film. This is reflected in theow response of this material towards oxygen. The problem ofarbon deposits surely affects pure titania sensors calcined at00 ◦C. However, these sensors show better response to oxygenhan pure titania sensors calcined at 700 ◦C because phase tran-ition from anatase to rutile is almost complete in pure titaniaalcined at 700 ◦C (and this is not the case in Nb-doped samples).
The sensors fabricated with Nb-doped TiO2 calcined at00 ◦C, which showed the best responsiveness, were also testedo lower concentrations of oxygen. Measurements show thathey respond with enough sensitivity to 10 ppm, 15 ppm and0 ppm of oxygen. The response is defined as it was shown
n Eq. (2). The error among the whole measurements with thehree sensors of a same material is also shown and is defineds the rate between the standard deviation (S.D. = √
Var, wherear = (1/n − 1)
∑ni=1 (Xi − X)2) and the mean value X of the
E. Sotter et al. / Sensors and Actua
Table 2Responses of sensors made from Nb-doped titania calcined at 700 ◦C and work-ing at 500 ◦C toward different oxygen concentrations
Oxygen concentrationin N2 (ppm)
Sensor response(�R/RN2 )
Error (%)
10 1.65 4.315 1.78 2.72
T
mc
4
b9
iwo
btdtwstt
tatr
gsctdnb
dtbsnstc
dr
vcwnotrrrtisispcbccwtrfi
sipscft
A
bg
R
[6] R.K. Sharma, M.C. Bhatnagar, G.L. Sharma, Mechanism in Nb doped TiO2
0 1.99 5.2
he carrier gas was pure nitrogen.
easurements. The responses of the sensors towards these con-entrations are summarized in Table 2.
. Conclusions
Pure and Nb-doped titania nanopowders were synthesizedy a sol–gel route and calcined at 600 ◦C, 700 ◦C, 800 ◦C and00 ◦C.
ICP technique was used to corroborate the content of niobiumn the doped material. The result is near to the nominal value,hich is 3%. This confirms that the doping process was carriedut correctly.
The XRD and Raman results showed that the addition of nio-ium ions to the titania lattice retards the phase transition inhe crystalline structure of the material. It was observed that foroped materials calcined at 600 ◦C or 700 ◦C the dominant crys-alline phase was anatase. For the other samples studied, rutileas the most abundant crystalline phase. XRD technique also
howed that the size of crystallites grew with a rise in calcinationemperature. This growth is hindered by the addition of niobiumo the titania matrix.
BET analyses showed that the surface area decreases whenhe calcination temperature increases. It was observed that theddition of niobium retards the reduction in surface area withemperature and also improves the porosity of the resulting mate-ial.
SEM analysis allowed for estimating the evolution in grainrowth for the different layers studied. It was observed that theize of grains in doped materials is smaller than in pure titania foralcination temperatures between 600 ◦C and 800 ◦C. At 900 ◦C,he size of grains was very similar to each other in pure and Nb-oped samples. The introduction of niobium in the titania latticeot only hinders the crystal growth but also the coalescenceetween them, which retards the grain growth.
The results of the physical characterization performed on theifferent materials indicated that doped samples calcined at lowemperatures (600 ◦C or 700 ◦C) could be expected to show theest responsiveness towards oxygen. The high surface area andmall grain size of such materials benefit the detection mecha-ism. Additionally, their crystalline phase, i.e., mostly anatase,hould enable the detection of oxygen at moderate operatingemperatures (300–600 ◦C). Finally, the niobium species present
ould also contribute to the catalytic process of detection.
Actually, the measurements performed showed that Nb-oped samples, with response values up to 2.0, were moreesponsive to oxygen than pure titania samples with response
tors B 127 (2007) 567–579 577
alues below 0.26. Among the Nb-doped samples, only thosealcined at 700 ◦C had good responsiveness towards oxygenith responses between 1.4 and 2.0. The highest responsive-ess was obtained at a working temperature of 500 ◦C. In spitef the fact that the doped materials calcined at 600 ◦C had bet-er physical characteristics than those calcined at 700 ◦C, theiresponsiveness was poorer. A possible explanation for this lowesponse may be the brookite phase contained in the doped mate-ials calcined at 600 ◦C; at this temperature the crystallinity ofhe brookite particles is improved with the consequent dimin-shing of surface defects, which affect the adsorption of oxygenpecies at the surface. That is, the occurrence of a brookite phases even worse than the presence of rutile phase for the oxygenensing properties of titania. Raman spectroscopy shows anotherossible reason for the low response of these low-temperaturealcined materials. Some peaks corresponding to different car-on species were found in their structure. These deposits ofarbon are contamination from the synthesis process, whichould not be eliminated during the calcination. The problemith these deposits of carbon is that they cover the surface of
he active material and hinder its interaction with oxygen. This iseflected in the low response towards oxygen found for materialsred at low temperatures.
In summary, we have shown that thick-film Nb-doped TiO2ensors can be used to detect traces of oxygen (as low as 10 ppm)n N2. This responsiveness compares favorably to previouslyublished results where the concentrations of oxygen tested wereignificantly higher. Nb allows for anatase to be the dominantrystalline phase. This enables the sensor to work as a sur-ace conductivity sensor and thus to lower its optimal operatingemperature down to 500 ◦C.
cknowledgements
This work was funded in part by Carburos Metalicos S.A. andy the Spanish Commission for Science and Technology underrant no. TEC2003-06301.
eferences
[1] R.K. Sharma, M.C. Bhatnagar, Improvement of the oxygen gas sensitivityin doped TiO2 thick films, Sens. Actuators B: Chem. 56 (1999) 215–219.
[2] Y. Hu, O.K. Tang, J.S. Pan, H. Huang, W. Cao, The eEffect of anneal-ing temperature on the sensing properties of low temperature nano-sizedSrTiO3 oxygen gas sensor, Sens. Actuators B: Chem. 108 (2005) 244–249.
[3] V. Guidi, M.A. Butturi, M.C. Carotta, B. Cavicchi, M. Ferroni, C. Malagu,G. Martinelli, D. Vincenzi, M. Sacerdoti, M. Zen, Gas sensing throughthick film technology, Sens. Actuators B: Chem. 84 (2002) 72–77.
[4] P. Ivanov, E. Llobet, X. Vilanova, J. Brezmes, J. Hubalek, X. Correig,Development of high sensitivity ethanol gas sensors based on Pt-dopedSnO2 surfaces, Sens. Actuators B: Chem. 99 (2004) 201–206.
[5] P. Ivanov, E. Llobet, F. Blanco, A. Vergara, X. Vilanova, I. Gracia, C. Cane,X. Correig, On the effects of the materials and the noble metal additives toNO2 detection, Sens. Actuators B: Chem. 118 (2006) 311–317.
oxygen gas sensor, Sens. Actuators B: Chem. 46 (1998) 194–201.[7] U. Kirner, K.D. Schierbaum, W. Gopel, B. Leibold, N. Nicoloso, W. Wepp-
ner, D. Fischer, W.F. Chu, Low and high-temperature TiO2 oxygen sensors,Sens. Actuators B: Chem. 1 (1990) 103–107.
5 Actua
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
B
EsUesc
XshfVgd
Esh
78 E. Sotter et al. / Sensors and
[8] J. Sheng, N. Yoshida, J. Karasawa, T. Fukami, Platinum doped titania filmsoxygen sensor integrated with temperature compensating thermistor, Sens.Actuators B: Chem. 41 (1997) 131–136.
[9] A. Takami, Development of titania heated exhaust gas oxygen sensor,Ceram. Bull. 67 (1988) 1956–1960.
10] A.M. Azad, S.A. Akbar, S.G. Mhaisalkar, L.D. Birkefeld, K.S. Goto,Solid-state gas sensors: a review, J. Electrochem. Soc. 139 (1992)3690–3695.
11] H. Tang, K. Prasad, R. Sanjines, F. Levy, TiO2 Anatase thin films as gassensors, Sens. Actuators B: Chem. 26 (1-3) (1995) 71–75.
12] Y. Xu, X. Zhou, O.T. Sorensen, Oxygen sensors based on semiconductingmetal oxides: an overview, Sens. Actuators B: Chem. 65 (2000) 2–4.
13] K. Zakrzewska, Gas sensing mechanism of TiO2-based thin films, Vacuum74 (2004) 335–338.
14] F. Hossein-Babaei, M. Keshmiri, M. Kakavand, T. Troczynski, A resistivegas sensor based on undoped p-type anatase, Sens. Actuators B: Chem. 110(2005) 28–35.
15] J. Arbiol, Metal Additive Distribution in TiO2 and SnO2 Semiconduc-tor Gas Sensor Nanostructured Materials, Doctoral thesis, Universitat deBarcelona, Departament d’Electronica, 2001.
16] J. Arbiol, J. Cerda, G. Dezanneau, A. Cirera, F. Peiro, A. Cornet, J.R.Morante, Effects of Nb doping on the TiO2 anatase-to-rutile phase transi-tion, J. Appl. Phys. 92 (2002) 853–861.
17] A. Ruiz, G. Dezanneau, J. Arbiol, A. Cornet, J.R. Morante, Study of theinfluence of Nb content and sintering temperature on TiO2 sensing films,Thin Solid Films 436 (2003) 90–94.
18] R.K. Sharma, Development and Characterization Solid State Materialsfor Oxygen Gas Sensors, Doctoral thesis, Indian Institute of Technology,Department of Physics, 1996.
19] A. Ruiz, TiO2 Nanoparticles for Gas Sensor Applications, Doctoral thesis,Universitat de Barcelona, Departament d’Electronica, 2003.
20] V.J. Nagpal, R.M. Davis, J.S. Riffle, In situ steric stabilization of tita-nium dioxide particles synthesized by a sol–gel process, Colloid Surf. A:Physicochem. Eng. Aspects 87 (1994) 25–31.
21] J.H. Kim, J.S. Sung, Y.M. Son, A.A. Vasiliev, V.V. Malyshev, E.A.Koltypin, A.V. Eryshkin, D.Y. Godovski, A.V. Pisyakov, S.S. Yakimov,Propane/butane semiconductor gas sensor with low power consumption,Sens. Actuators B: Chem. 44 (1997) 452–457.
22] R. Diaz, J. Arbiol, F. Sanz, A. Cornet, J.R. Morante, Electroless addition ofplatinum to SnO2 nanopowders, Chem. Mater. 14 (8) (2002) 3277–3283.
23] Bruker Advanced X-ray Solutions, Difract Plus Topas 2.1, Tehnical Refer-ence, 2003.
24] P.H. Zimmermann, Temperature dependence of the EPR spectra ofniobium-doped TiO2, Phys. Rev. B 8 (1973) 3917–3927.
25] A. Bernasik, M. Radeeka, M. Rekes, M. Sloma, Electrical properties of Crand Nb-doped TiO2 thin films, Appl. Surf. Sci. 65-66 (1993) 240–245.
26] M.C. Carotta, M. Ferroni, D. Gnani, V. Guidi, M. Merli, G. Martinelli,M.C. Casale, N. Norato, Nanostructured pure and Nb-doped TiO2 as thickfilm gas sensors for environmental monitoring, Sens. Actuators B: Chem.58 (1999) 310–317.
27] M. Ferroni, M.C. Carotta, V. Guidi, G. Martinelli, F. Ronconi, O. Richard,D.V. Dyck, J.V. Landuyt, Structural characterization of Nb–TiO2 nanosizedthick-films for gas sensing application, Sens. Actuators B: Chem. 68 (2000)140–145.
28] R.D. Shannon, J.A. Pask, Kinetics of the anatase–rutile transformation, J.Am. Ceram. Soc. 48 (1965) 391–398.
29] S. Hishita, I. Mutoh, K. Koumoto, H. Yanagida, Inhibition mechanism ofthe anatase–rutile phase transformation by rare earth oxides, Ceram. Int. 9(1983) 61–67.
30] M.K. Akhtar, S.E. Pratsinis, S.V.R. Mastrangelo, Dopants in vapor-phasesynthesis of titania powders, J. Am. Ceram. Soc. 75 (1992) 3408–3416.
31] Y. Djaoued, R. Bruning, D. Bersani, P.P. Lottici, S. Badilescu, Sol–gelnanocrystalline brookite-rich titania films, Mater. Lett. 58 (2004)
2618–2622.
32] M. Koelsch, S. Cassaignon, C. Ta, T. Minh, J.F. Guillemoles, J.P. Jolivet,Electrochemical comparative study of titania (anatase, brookite and rutile)nanoparticles synthesized in aqueous medium, Thin Solid Films 451/452(2004) 86–92.
fVmlt
tors B 127 (2007) 567–579
33] J. Rouquerol, D. Avnir, C.W. Fairbridge, D.H. Everett, J.M. Haynes, N.Pernicone, J.D.F. Ramsay, K.S.W. Sing, K.K. Unger, Recommendations forthe characterization of porous solids (technical report), Pure Appl. Chem.66 (1994) 1739–1758.
34] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J.Rouquerol, T. Siemieniwska, Reporting physisorption data for gas/solidsystems, Pure Appl. Chem. 57 (1985), 603-619.
35] P. Somasundaran, Adsorption at the gas–solid and liquid–solid interface,Colloid Surf. 9 (1) (1984) 96–97.
36] L.E. Depero, L. Sangaletti, B. Allieri, E. Bontempi, A. Marino, M. Zocchi,Correlation between crystallite sizes and microstrains in TiO2 nanopow-ders, J. Cryst. Growth 198 (1999) 516–520.
37] J. Marien, T. Wagner, G. Duscher, A. Koch, M. Ruhle, Nb on (1 1 0) TiO2
(rutile): growth, structure and chemical composition of the interface, Surf.Sci. 446 (2000) 219–228.
38] C. Xu, J. Tamaki, N. Miura, N. Yamazoe, Grain-size effects on gas sensi-tivity of porous SnO2-based elements, Sens. Actuators B: Chem. 3 (1991)147–155.
39] Z.L. Zhang, V.A. Tsipouriari, A.M. Efstathiou, X.E. Verykios, Reformingof methane with carbon dioxide to synthesis gas over supported rhodiumcatalysts. I. Effects of support and metal crystallite size on reaction activityand deactivation characteristics, J. Catal. 158 (1996) 51–63.
40] Z. Paal, A. Wootsch, R. Schlogl, U. Wild, Carbon accumulation, deactiva-tion and reactivation of Pt catalysts upon exposure to hydrocarbons, Appl.Catal. A: Gen. 282 (2005) 135–145.
41] N.M. Rodriguez, P.E. Anderson, A. Wootsch, U. Wild, R. Schlogl, Z. Paal,XPS, EM, and catalytic studies of the accumulation of carbon on Pt black,J. Catal. 197 (2001) 365–377.
42] J. Kohanoff, Phonon spectra from short non-thermally equilibratedmolecular dynamics simulations, Comput. Mater. Sci. 2 (1994) 221–232.
43] D. Knight, W. White, Characterization of diamond films by Raman spec-troscopy, J. Mater. Res. 4 (1989) 385–393.
44] J. Schwan, S. Ulrich, V. Batori, H. Ehrhardt, Raman spectroscopy on amor-phous carbon films, J. Appl. Phys. 80 (1996) 440–447.
45] M.A. Tamor, W.C. Vassel, Raman “Fingerprinting” of amorphous carbonfilms, J. Appl. Phys. 76 (1994) 3823–3829.
46] R.O. Dillon, J.A. Woollam, Use of Raman scattering to investigate disorderand crystalline formation in as-deposited an annealed carbon films, Phys.Rev. B 29 (1984) 3482–3489.
47] R.J. Nemanich, S.A. Solin, First- and second-order Raman scattering fromfinite-size crystals of graphite, Phys. Rev. B 20 (1979) 392–401.
iographies
dgar Sotter, obtained his electronic engineering degree in 2001 at the Univer-idad del Norte in Barranquilla, Colombia. He obtained his PhD degree at theniversitat Rovira i Virgili (Tarragona, Spain) under the doctoral program on
lectronic engineering in 2006. His main research topic is related to the synthe-is of semiconductor materials and their implementation in the development ofhemical gas sensors.
avier Vilanova, graduated in telecommunication engineering from the Univer-itat Politecnica de Catalunya (UPC), (Barcelona, Spain) in 1991, and receivedis PhD in 1998 from the same university. He is currently an associate pro-essor in the Electronic Engineering Department at the Universitat Rovira iirgili (Tarragona, Spain). His research activities are related to semiconductoras sensors development and characterisation, as well as, gas sensors systemsesign.
duard Llobet, graduated in telecommunication engineering from the Univer-itat Politecnica de Catalunya (UPC), (Barcelona, Spain) in 1991, and receivedis PhD in 1997 from the same university. He is currently an associate pro-
essor in the Electronic Engineering Department at the Universitat Rovira iirgili (Tarragona, Spain). His main areas of interest are in the fabrication, andodeling of semiconductor chemical sensors and in the application of intel-
igent systems to complex odour analysis. Dr. Llobet is a senior member ofhe IEEE.
Actua
AitSibl(i
ws
X
E. Sotter et al. / Sensors and
lexey Vasiliev, graduated from Moscow Institute of Physics and Technologyn 1980, obtained his PhD in 1986 for the “Study of the kinetics of low-emperature reactions of atomic fluorine by ESR method”. Gained his Dr. ofcience degree (habilitation) in solid state microelectronics in 2004 for the
nvestigation of “Physical and chemical principles of design of gas sensorsased on metal oxide semiconductors and MIS structures with solid electrolyteayer”. Recently is working in Sensor group of the University Rovira i VirgiliTarragona, Spain). Research interests are related with the study of the kinet-cs and mechanisms of heterogeneous processes related with chemical sensing,
shtic
tors B 127 (2007) 567–579 579
ith design and prototyping of gas sensors and instruments based on theseensors.
avier Correig, graduated in telecommunication engineering from the Univer-
itat Politecnica de Catalunya (UPC), (Barcelona, Spain) in 1984, and receivedis PhD in 1988 from the same university. He is a full professor of electronicechnology in the Electronic Engineering Department at the Universitat RoviraVirgili (Tarragona, Spain). His research interests include heterojunction semi-onductor devices and solid-state gas sensors.
