ELECTRONIC NOSES APPLICATIONS AND TECHNOLOGIES

Nanostrutured thick film sensors for CO (g) based on Al doped SnO₂

¹Cabezas M.D.; ²Lamas D.G.; ³Baby R.E.; ⁴Cabanillas E.; ²Walsöe de Reca N.E.

¹CITEFA, ²CITEFA-CONICET, ³CONICET-UTN (FRBA),⁴CONICET-CNEA
CINSO (Centro de INvestigaciones en SOlidos) CITEFA-CONICET, Juan Bautista de La Salle 4397, Villa Martelli (B1603ALO), Buenos Aires, Argentina, Fax: (54) (11)-4709-8241, E-Mail:walsoe@citefa.gov.ar

Received November 05, 2004. In final form May 30, 2005

Abstract
In this work, the aluminium doping of SnO₂ lattice was studied, synthesising a semiconductive nanocrystalline doped material by a gel-combustion method with citric acid as fuel, to be used for thick film sensors to detect CO (g). Aluminium concentration was varied from 1 to 5 at %, nanocrystalline powders with crystallite mean size of 8-12 nm and low agglomeration were obtained. The semiconductive material was characterised with X-rays diffraction, energy dispersive electron microprobe and transmission electron microscopy. Nanocrystalline powders were used to prepare conductive pastes with suitable organic solvents employing our own formulations different from those of commercial materials. Pastes were deposited as serigraphic thick films on electronic purity alumina wafers provided with Pt electrodes. Sintering of thick sensing films was performed in a range of medium temperatures to avoid important crystallite growth since it is convenient to maintain the nanocrystalline structure in order to increase sensor sensitivity. Sensors were characterised in mixtures of [CO(g)+air] at temperatures from 300 to 400 °C. Results indicated that aluminium was incorporated in the SnO₂ lattice in low proportions leaving low concentrations of amorphous Al₂O₃ segregated. Sensors showed a high sensitivity to the target gas. In addition, it was also found that the incorporation of aluminium to SnO₂ lowers the sensor operating temperature.

Resumen
En este trabajo de investigación, se estudió el dopado con aluminio de SnO₂, sintetizando el material semiconductor dopado nanocristalino, mediante el método de gelificación-combustión empleando ácido cítrico como combustible, con el objeto de usarlo en sensores de película gruesa para detectar CO (g). La concentración de Al se varió entre 1 y 5 at % y se obtuvieron polvos nanocristalinos, con tamaño promedio de cristalitas entre 8 y 12 nm y bajo grado de aglomeración. El material semiconductor fue caracterizado por difracción de rayos X, microsonda electrónica dispersiva en energías y microscopía electrónica de transmisión. Los polvos nanocristalinos fueron usados para preparar pastas conductoras con solventes orgánicos convenientes, empleando formulaciones propias diferentes de aquéllas de los materiales comerciales. Las pastas se depositaron como películas gruesas por serigrafía sobre sustratos de alúmina de pureza electrónica, provistos con electrodos de Pt. El sinterizado de las películas gruesas sensoras se realizó en un rango medio de temperaturas para evitar un crecimiento importante de las cristalitas ya que es conveniente mantener la estructura nanocristalina para aumentar la sensibilidad del sensor. Los sensores fueron caracterizados en mezclas de [CO (g)+aire] en un rango de temperatura entre 300 y 400 ^oC. Los resultados indicaron que el aluminio se había incorporado a la red del SnO₂ en bajas proporciones quedando, además, muy bajas concentraciones de Al₂O₃ segregada amorfa. Los sensores mostraron una alta sensibilidad al gas a detectar. Además, se encontró que la incorporación del Al al SnO₂ permitió reducir la temperatura de operación del sensor.

Introduction
Tin dioxide (SnO₂) is one of the most studied metal oxides because of its interesting properties and applications. Perhaps the most important of them are: coatings as transparent conductors and the fabrication of sensors for reducing gases like CO, H₂ or CH₄. The sensors response is based on measurements of surface electrical resistance, in comparison with the resistance value in air, in the presence of the gas to be detected. Sensor response is defined as the ratio S = R_air/R_air+gas. This variation is due to the reducing gas reaction with the O^- (oxygen adsorbates on the SnO₂ surface, formed in air at the typical working temperatures, 300-500 ºC, which reduce the potential barrier produced by the adsorbates limiting the electrons movement).
Yamazoe et al. [1] proved that the sensors response increases significantly if the crystallite size (D) of the nanocrystalline tin dioxide is about twice the adsorption depth, 2L (L is the depth of the space-charge layer), of oxygen adsorbates. This fact means that the sensor performance improves not only when D decreases but also if L increases, since a major proportion of material takes place in the reaction of oxygen adsorbates with the target gas even if the grain size is not excessively small.
Besides, Yamazoe et al. have found that doping with trivalent cations enables to increase the response since the carriers concentration is reduced and consequently, L increases [1]. As an example of this phenomenon, they showed Al₂O₃ doped SnO₂ sensors which were built for H₂ detection but they did not report data on CO(g) sensing. It is important to point out that these authors have prepared sensors material by coprecipitation, followed by calcination at high temperature. They could not evaluate the Al solubility in the SnO₂ lattice, in spite of reporting that it had to be lower than 1% as determined by resistivity measurements.
The aim of this work was to study the possibility of increasing the Al solubility in the SnO₂ lattice because this fact could improve the gas sensor sensitivity since carriers concentration could decrease even more. Nanocrystalline SnO₂ was synthesised by the nitrate-citrate gel-combustion method [2, 3] since the employment of similar routes in other systems made possible to increase the solubility limit of the dopant, in a metastable condition, while retaining the homogeneity in composition of materials [4, 5]. This is due to the fact that the powder is obtained by the rapid disintegration of the homogeneous gel, so the system cannot evolve towards its equilibrium state.
Even though Al can be introduced in SnO₂ lattice in low proportions, results of the present work show that the solubility limit of Al in SnO₂ can be extend by the use of the gel-combustion method in comparison with coprecipitation. Crystallite sizes were considerably small, between 8 and 12 nm. Sensors showed a high sensitivity, between 8 and 10 for 200 ppm of CO in air. It has also been found that the incorporation of Al to SnO₂ enables to decrease the working temperature.

Experimental Procedure
1.Powders and thick films preparation
SnO₂ with 0, 2 and 5 at % Al powders have been obtained by the nitrate-citrate gel-combustion method [2,3]. The employed raw materials were pure metallic Sn and Al, dissolved in diluted HNO₃. The use of chlorides of both metals has been avoided because Cl^- ions, if retained, can damage the sensors performance. Citric acid was added in a proportion of six acid moles per metal mole and neutrality was attained with NH₄OH. The solution was then thermally concentrated at a temperature of 80 ºC on a heating plate until a dark gel was obtained. Finally, at temperature above 250 ºC, the dried gel burned due to a vigorous exothermic reaction between nitric and citric acids. The ashes produced by combustion were calcined for 1 hour at 700 ºC.
Pastes were prepared with the synthesised powders and organic solvents to get thick films for sensors, painted on electronic purity Al₂O₃ substrates, provided with Pt contacts, dried at 100 ºC with a final sintering for an hour at 700 ºC.

2. Powders and thick films characterisation
Powders and films were studied by X-Ray diffraction (XRD) with a Philips PW 3710 diffractometer, employing Cu-Ka radiation to identify phases and to determine the crystallites size. For this last evaluation the Scherrer equation: D = 0.9 l / b cos q was used, where D is the crystallite size, l is the wave length of the incident radiation (1.5418 Å for Cu-Ka radiation), b is the half maximum peak width and q is the peak position. The instrument width, necessary to correct b, was evaluated employing a Al₂O₃ with a mean crystallite size of 25 mm.
Powders morphology was studied by Transmission Electron Microscopy (TEM) with a Philips CM200 electron microscope. Chemical composition and Al distribution were evaluated by energy dispersive electron microprobe (EDS) using a EDAX CM20 microprobe.

3. Sensors characterisation
To evaluate the sensors performance, the electrical resistance in synthetic air and in a diluted solution of 200 ppm of CO in synthetic air was measured with a 617 Keithley electrometer between 300 and 400 ºC. In both measurements the gas flow was 200 ml/min.

Results and Discussion
XRD patterns of the studied materials did not present any shift lattice parameters with increasing doping concentration, suggesting that aluminium solubility in SnO₂ was low. In spite of that, Al₂O₃ peaks were not observed.This fact was assigned to the low doping concentration but, experiments with high Al concentrations suggested that it was more probable that this phase had remained amorphous. The powders crystallite size resulted between 8 and 12 nm, considerably smaller than that obtained for coprecipitated material, Yamazoe et al. [1]. As an example, Figure 1 shows the diffractogram corresponding to undoped SnO₂ powder exhibiting a crystallite mean size of 12 nm. It was also found that the crystallite size was slightly reduced by an increasing of aluminium content. The thick films exhibited similar crystallite size to that of powders because the calcination temperature was relatively low.

Figure 1: Diffractogram corresponding to non-doped SnO₂ powder.

Figure 2 : TEM micrograph of nanocrystalline 5 at% Al doped SnO₂ powders (bright field).

TEM studies enabled to confirm the crystallite size as determined by XRD. For example, the Figure 2 shows a TEM bright field image of SnO₂ – 5 at% Al powder with a crystallite size of 10 nm. Besides, it is important to point out that powders are formed by spherical particles, characteristic of powders obtained by gel-combustion method with a very low state of agglomeration. The EDS study of powder demonstrated that a part of the Al is not incorporated into the SnO₂ lattice since small segregated Al-riched regions were observed as well.

Figure 3. Response (for 200ppm CO + air) of thick film pure SnO₂ and of doped - 2 and 5 at% Al doped SnO₂ sensors.

To evaluate the Al solubility in SnO₂ lattice, the variation of the electrical resistance with the Al concentration was measured. The resistance of undoped SnO₂ films was about 10⁷ W. The resistance of 0.2 at% Al doped SnO₂ films resulted within the same magnitude order but ~3 times more resistive while the 5 at% Al doped SnO₂ films exhibited a difference of more than one order of magnitude higher. This continuos increase of resistance is consistent with the decrease of the carriers concentration and with the increase of L (which is possible if Al is incorporated into the SnO₂ lattice). Yamazoe et al. [1] reported a higher resistance variation for coprecipitated powders. These authors found a resistance change three orders of magnitude higher for 1 at% Al doped SnO₂ but they did not observe an important resistance increase with higher doping concentration. This fact was associated to Al₂O₃ segregation even for very low Al concentrations, concluding that the Al solubility limit in SnO₂ was lower than 1at%. The considerable difference in the behaviour of materials according to their synthesis (coprecipitation or gel-combustion) suggests that the gel-combustion method enables to increase significantly the Al solubility in SnO₂.
In Figure 3, the response of pure, 2 at% and 5 at% Al doped SnO₂ is plotted versus temperature (°C) for 200 ppm CO in air. If pure SnO₂ sensors were compared with 2 at% Al doped SnO₂ sensors, it was observed that the latter exhibited a higher sensitivity for the whole considered temperature range. The maximal sensitivity resulted: ~8 for non-doped SnO₂ and 10 for 2 at% Al doped SnO₂. The optimal working temperature (T_w) resulted ~375 °C for both types of sensors. However, 5 at% Al doped SnO₂ sensors exhibited a lower optimal working temperature: 325 ºC and their sensitivity resulted of 8. This decrease of T_w may be assigned to a shift of the formation temperature characteristic of adsorbates O^- by the incorporation of Al in the SnO₂ lattice. This phenomenon has been reported in the technical literature for doping with trivalent elements like Bi [6].

Conclusions
Results indicate that the gel-combustion method enables to increase the Al solubility in SnO₂ in comparison with the coprecipitation method and, at the same time, it reduces the crystallite size. Both facts improve the sensitivity of the sensors built with the gel-combustion synthesised materials as it was proved with the obtained thick films sensors. The response of the sensors reported in this work was considerably high, between 8 and 10 for 200 ppm CO in synthetic air. Furthermore, it is possible to reduce the temperature of operation of sensors by increasing the Al concentration.

Acknowledgements
Authors are indebted to the Agencia Nacional de Promoción Científica y Tecnológica by Grant PICT No. 8688 which enabled to carry out this research.

References

[1] Xu, C.; Tamaki, J.; Miura, N.; Yamazoe, N., Sensors & Actuators B-Chemical, 1991, 3, 147-155.         [ Links ]

[2] Fraigi, L. B.; Lamas, D. G.; Walsöe de Reca, N. E., NanoStructured Materials, 1999, 11, 311-318.         [ Links ]

[3] Fraigi, L. B.; Lamas, D. G.; Walsöe de Reca, N. E., Materials Letters, 2001, 47, 262-266.         [ Links ]

[4] Lamas, D. G.; Walsöe de Reca, N. E., Journal of Materials Science, 2000, 35, 5563-5567.         [ Links ]

[5] Lamas, D. G.; Lascalea, G. E.; Juárez, R. E.; Djurado, E.; Pérez L.; Walsöe de Reca, N. E., Journal of Materials Chemistry, 2003, 13, 904-910.         [ Links ]

[6] Sarala Devi, G.; Manorama, S. V.; Rao V. J., Sensors & Actuators B-Chemical, 1999, 56, 98-105.         [ Links ]