Sensing of hydrocarbons and CO in low oxygen conditions with tin dioxide sensors: possible conversion paths

Sensing of hydrocarbons and CO in low oxygen conditions with tin dioxide sensors: possible conversion paths

Sensors and Actuators B 103 (2004) 362–368 Sensing of hydrocarbons and CO in low oxygen conditions with tin dioxide sensors: possible conversion path...

211KB Sizes 0 Downloads 26 Views

Sensors and Actuators B 103 (2004) 362–368

Sensing of hydrocarbons and CO in low oxygen conditions with tin dioxide sensors: possible conversion paths Wolf Schmid∗ , Nicolae Bˆarsan, Udo Weimar Institute of Physical and Theoretical Chemistry, University of Tübingen, Auf der Morgenstelle 8, 72076 Tübingen, Germany Available online 7 June 2004

Abstract The conversion of CO and different hydrocarbons by palladium-doped tin oxide sensors was studied in conditions with low oxygen and low water concentration for temperatures between 200 and 400 ◦ C. Although, under normal conditions, oxygen plays an important role in the detection of these reducing analytes, relatively large sensor signals could be observed. For most conditions, carbon dioxide and water could be observed as reaction products even at low oxygen content. © 2004 Elsevier B.V. All rights reserved. Keywords: Tin dioxide; Low oxygen conditions; Consumption

1. Introduction Tin oxide sensors usually work in environmental conditions: with varying backgrounds of water vapour, but constant oxygen concentration (20.9 vol.%). Water vapour plays an important role for the detection, as was shown in previous publications [1–3]. Usually it is assumed that for reducing gases like carbon monoxide (CO) and hydrocarbons (HCs), tin oxide sensors work through oxidation of the analyte. It could be shown that interaction of the doping material plays also an important role [4]. Astonishing at first glance is that also under conditions with a low concentration of oxygen (below 100 ppm), tin oxide sensors still show a sensitivity against CO and HCs. This work investigates the sort and the amount of products related to sensing during the detection of CO and HCs with tin oxide sensors.

2. Experimental The measurements were performed at different sensor temperatures (between 200 and 400 ◦ C) and different analyte concentrations (cf. Table 1) in dry and humid nitrogen (10% relative humidity at room temperature) with 0 and 2% oxygen added. An IR gas analyser (Innova Airtech, Type ∗ Corresponding author. Tel.: +49-7071-2978768; fax: +49-7071-2955960. E-mail address: [email protected] (W. Schmid). URL: http://www.ipc.uni-tuebingen.de/weimar/.

0925-4005/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2004.04.079

1301/1312) was used to monitor the gas composition upstream and downstream the sensors and simultaneously, the sensor resistance was measured with a multimeter (Keithley 2000). Home made SnO2 -based thick film sensors (Fig. 1) doped with Pd were used, consisting of an alumina substrate and screen-printed platinum interdigitated electrodes (front side) and resistive heater (back side). The sensitive material (Pd-doped SnO2 ) was applied with screen printing onto the substrate. Subsequentially to the screen printing, the sensors underwent a thermal annealing at 700 ◦ C for 10 min. An SEM picture of the sensitive layer is shown in Fig. 2. The grains are approximately 10 nm in diameter, with a narrow grain size distribution. The seven identically prepared sensors were mounted jointly in one chamber, which was alternatively supplied by two separate gas flows, one with nitrogen only, the other with varying concentrations of the analytes in nitrogen to keep the sensors in constant conditions all the time, even when making reference measurements. The pressure inside the flow system was monitored with a barometer (Vaisala) and used as a correction for the gas analyser. The oxygen partial pressure was measured with a process oxygen analyser (ABB) and recorded continuously together with the resistance of the sensors. To distinguish between the consumption of the sensitive material and the consumption of the substrate, additional measurements were performed with substrates equipped with electrodes and heater only. Test gases from gas cylinders were used throughout the measurements.

W. Schmid et al. / Sensors and Actuators B 103 (2004) 362–368

363

Table 1 Measured gases and concentrations Analyte Measured concentrations (ppm) Carbon monoxide

Propane

10 20 50 100

15 30 50 100

50µ 5µ 900µ

Fig. 2. SEM picture of the SnO2 sensing layer.



25.4mm x 4.2

Fig. 1. Schematic of the home made thick film tin oxide sensors.

Due to limitations of the set-up and of the gases used in this work, there was a background of oxygen (≈50 ppm) and humidity (≈500 ppm), however, the conditions without added oxygen or humidity were denoted “0% O2 ” and “dry”, respectively.

3. Results and discussion For conditions with low oxygen concentration and low humidity (denoted “dry, 0% O2 ” in the figures; these conditions will be referred to as a starting point throughout the description), propane and CO behave quite differently: for

propane, sensor signal as well as consumption (difference of the analyte concentration upstream and downstream the sensors) and CO2 production show a saturation for higher concentrations (Fig. 3), with a maximum around 350 ◦ C for the sensor signal (see Fig. 4). CO on the other hand shows no saturation (except for 200 ◦ C) of the sensor signal (Fig. 5), and the amount of CO2 which is produced in the sensing process and the consumption (not shown) are almost linear with the CO concentration. The sensor signal is decreased with rising temperature (Fig. 6). When 2% (vol.) oxygen are added, the situation changes dramatically: for CO, the sensor signal is strongly reduced, and the temperature dependence of the sensor signal is inverted (Fig. 6). For propane, the sensor signal is shifted down by a factor of around 2 (Fig. 4). Adding oxygen results in a higher consumption and more produced carbon dioxide for CO (Fig. 7), while the opposite is observed for propane (Fig. 8).

120

200˚C 240˚C 270˚C 310˚C 350˚C 400˚C

sensor signal (R0/Rgas)

100

80

60

40

20

0% O2

0 0

20

40

60

80

100

propane concentration [ppm] Fig. 3. Sensor signal for different concentrations of propane in dry N2 without oxygen.

364

W. Schmid et al. / Sensors and Actuators B 103 (2004) 362–368

sensor signal (R0/Rgas)

100

dry, 0% O2 humid, 0% O2 dry, 2% O2 humid, 2% O2

10

100ppm propane 200

250

300

350

400

temperature [˚C] Fig. 4. Sensor signal for 100 ppm propane in N2 at different temperatures.

For propane, adding humidity shifts the sensor signal down by a factor of around 2 (Fig. 8) and enhances the production of CO2 when no oxygen is available, while the concentration of produced CO2 remains unchanged when 2% oxygen are available. For CO, the presence of humidity enhances the production of CO2 (with or without oxygen, Fig. 7). With 2% oxygen, the sensor signal is almost unaffected when 10% relative humidity are added, while it is shifted down with 0% oxygen (Fig. 6). For bare substrates (without sensitive layer), the consumption and the produced CO2 concentration were around the detection limit. Only high concentrations of CO together with the presence of 2% oxygen showed a small consump-

200˚C 240˚C 270˚C 310˚C 350˚C 400˚C

200

sensor signal (R0/Rgas)

tion and production of CO2 . This finding is in opposition to the behaviour of CO in air, where the substrates were responsible for a large fraction of the consumption (cf. 1). For propane, the measured CO2 concentrations were around the detection limit, independent from propane concentration, temperature as well as water and oxygen level. For CO, the sensor signal (Fig. 6) is not directly correlated with the CO2 production (Fig. 7): a larger sensor signal does not necessarily go with more consumption and more produced CO2 . However, the experimental findings can be explained by assuming the following competitive detection mechanism: in the presence of very small amounts of oxygen (intended 0%) and water; CO can adsorb at the surface of the sensitive layer and act as an electron donor (Fig. 9). The

150

100

50

0% O2

0 0

20

40

60

80

100

CO concentration [ppm] Fig. 5. Sensor signal for different concentrations of CO in dry N2 without oxygen.

W. Schmid et al. / Sensors and Actuators B 103 (2004) 362–368

sensor signal (R0/Rgas)

1000

365

dry, 0% O2 humid, 0% O2 dry, 2% O2 humid, 2% O2

100

10

100ppm CO 200

250

300

350

400

temperature [˚C] Fig. 6. Sensor signal 100 ppm CO in N2 at different temperatures.

release of an electron does not result in a localised charge, but the electron is rather inserted into the conduction band of the material, thus increasing the conductivity of the material. The molecular configuration of the adsorption, i.e. bonding of the carbon or the oxygen atom to a tin or oxygen atom of the surface could not be clarified up to now, but is subject to investigation. When small amounts of oxygen are added (Fig. 10), the CO can either be directly chemisorbed or it can react with ionosorbed oxygen, also resulting in an electrical effect (i.e. sensor signal) through the insertion of electrons (originating from the ionosorbed oxygen) into the conduction band. With larger amounts of oxygen available, this direct chemisorption seems to be hindered. Thus, at ambient oxygen con-

centration, the reaction with ionosorbed oxygen is the only possible reaction. As shown by the experimental results, the effect in resistance is quite dramatic (>1 order of magnitude change). The maximum coverage is already reached for relative small concentrations of CO, as observed by [5], but above 100 ppm, so the saturation was not observed in the experiments in this work. The generation of CO2 is probably due to the residual O2 still present in the gas and at the surface. At higher oxygen concentrations, the generation of CO2 is the main reaction, but results in less electrical effect than the direct adsorption, which is confirmed by the decrease in sensor signals and increase in CO2 generation.

40

produced CO2 [ppm]

100ppm CO

30

20

dry, 0% O2 humid, 0% O2 dry, 2% O2 humid, 2% O2

10

0 250

300

350

400

temperature [˚C] Fig. 7. Produced CO2 for 100 ppm propane (left) and for 100 ppm CO (right).

366

W. Schmid et al. / Sensors and Actuators B 103 (2004) 362–368

25

produced CO2 [ppm]

20

dry, 0% O2 humid, 0% O2 dry, 2% O2 humid, 2% O2

15

10

5

0

100ppm propane -5 250

300

350

400

temperature [˚C] Fig. 8. Produced CO2 for 100 ppm propane (left) and for 100 ppm CO (right).

electron into the conduction band. Details on this reaction and on the nature of the protonated water and similar species (Zundel-structures, hydrated protons) can be found in [3]. Of these three possibilities, the direct adsorption results in the largest electrical effect, followed by the reaction with hydroxyl groups and the reaction with ionosorbed oxygen. For propane, the explanation is quite similar: in the presence of very small amounts of oxygen and water (denoted “dry, 0% O2 ”), propane can adsorb on the tin oxide surface (without consumption and reaction to CO2 ) and cause an electrical effect (Fig. 12). As SnO2 is known to be a weak basic oxide [6], it is assumed that C3 H8 dissociates to a propyl group and hydrogen on the SnO2 surface: C3 H8,gas  C3 H7,ads + Hads (1)

CO + chemisorbed CO

e Fig. 9. Direct chemisorption of CO on SnO2 . Through the chemisorption, an electron is inserted into the conduction band and thus, conductivity of the material is increased.

When small amounts of water are present, CO can alternatively react with certain hydroxyl groups on the surface (Fig. 11) resulting in CO2 production and protonated water as well as an electrical effect through the insertion of an CO CO

no O2, no CO

CO

CO

CO CO CO CO CO CO

CO SnO2

O2, O

O

CO2

CO

CO O

CO

CO O

CO

CO2

O CO CO CO

O CO

CO

SnO2

O2, CO2

CO O O O O O O O

CO O

CO

CO2

CO2

CO O

O O O O O O

CO

CO2 CO

O O O O O O

CO

SnO2

Fig. 10. Competitive mechanism for the interaction of CO with the sensing layer in dry conditions.

CO2

W. Schmid et al. / Sensors and Actuators B 103 (2004) 362–368

propane which is anyhow adsorbed, can now react with the plenty of OH groups (from the adsorption of water) available at the surface resulting in an increased consumption and production of CO2 . Presence of oxygen hinders the dissociative adsorption of propane (more than water does), but a reaction with the ionosorbed oxygen is only possible at higher temperatures, so the sensor signal is decreased and consumption and production of CO2 are decreased. As the details of the interaction are not fully understood, they are subject to further investigation.

CO2

CO

Sn+-OHad

H3O+

367

e Fig. 11. Reaction of CO with isolated hydroxyl groups to CO2 . The hydrogen freed from the OH group is transferred to an adsorbed water molecule resulting in protonated water and the insertion of an electron into the conduction band.

This dissociation was observed for CH4 and in the case of C3 H8 it should be even favoured because of the lower bond energies of the C−H bonds (431 kJ/mol for CH4 compared to 410 kJ/mol for C2 H6 /C3 H8 [7]). Hads acts as a donor, in combination with lattice oxygen Olat one obtains rooted OH groups: Hads + Olat  (Olat H)+ + e−

(2)

The reaction described in Eq. (2) inserts electrons into the conduction band thus leading to a conductance increase. The propyl group can adsorb on a lattice oxygen forming a rooted propoxy-like species and a hydrogen atom (Fig. 12): The experimental results (see above) and thermodynamic calculations suggest that the reaction leading to carbon dioxide is favoured, as no intermediates (like alcohols, aldehydes or carboxylic acids) were found in the infrared spectra. However, with more sensitive detection methods and/or higher concentrations of propane, the partly oxidised products may be traceable as by-products. In the experiments with propane, adding oxygen and/or water results in a smaller sensor signal (Fig. 4), while the effect on the sensor signal points in opposite directions: presence of humidity enhances consumption and the reaction to CO2 ; while the presence of oxygen seems to inhibit the consumption/production of CO2 ; except for the highest temperature (Fig. 8). A possible explanation could be the following: humidity decreases the sensor signal through partial blocking of the adsorption sites for propane. The fraction of

4. Summary and conclusion In the course of this work, an experimental set-up was established which allows for the measurement of the consumption of gas sensors during the variation of the oxygen concentration. The concentrations of the test gases were measured with a gas analyser based on IR spectrometry, the oxygen concentration was monitored with a process oxygen analyser. Both CO and propane show clear sensor signals even in the absence of oxygen. This is due to a direct chemisorption (for CO) and a dissociative adsorption (for propane, respectively), which lead to insertion of electrons into the conduction band and thus, change in resistance. This interaction seems to be hindered by the presence of humidity or oxygen, probably through blocking of adsorption sites.

5. Outlook The role of water-related adsorbed species during the sensing process is not fully clear as well as the (positive or negative) catalytic interaction between sensor material, dopant and electrode material. To extend the view to the surface, direct spectroscopic examination of the adsorbed species during sensing is needed (e.g. in situ IR studies). As the set-up used in this work had some shortcomings, the improvement of the set-up may lead to more precise results. The extension of the measurements to different test gases, different sensing layers or larger temperature ranges would also be promising investigation targets.

References C2 H5 H

C

H

C2H5

H Sn O

Sn O

H Sn

O

Sn O

O

C

Sn O

H Sn

O

H O

propoxy like species

Fig. 12. Dissociative adsorption of propane leading to a propoxy-like species and a hydrogen atom. Through insertion of an electron into the conduction band the conductivity is increased.

[1] J. Kappler, A. Tomescu, N. Barsan, U. Weimar, CO consumption of Pd doped SnO2 based sensors, Thin Solid Films 391 (2001) 186–191. [2] W. Schmid, N. Bˆarsan, U. Weimar, Sensing of hydrocarbons with tin oxide sensors: possible reaction path as revealed by consumption measurements, Sens. Actuators B 89 (2003) 232–236. [3] S. Emiroglu, N. Barsan, U. Weimar, V. Hoffmann, In-situ diffuse reflectance infrared spectroscopy study of CO adsorption on SnO2 , Thin Solid Films 391 (2001) 176–185. [4] A. Cabot, A. Vilà, J. Morante, Analysis of the catalytic activity and electrical characteristics of different modified SnO2 layers for gas sensors, Sens. Actuators B 84 (2002) 12–20.

368

W. Schmid et al. / Sensors and Actuators B 103 (2004) 362–368

[5] S. Hahn, N. Barsan, U. Weimar, S. Ejakov, J. Visser, R. Soltis, Oxygen and water interactions at the surface of SnO2 based sensors, in: proceeding of Eurosensors XVI conference, Prague, Czech Republic, 2002.

[6] S. Munnix, M. Schmeits, Electronic structure of tin dioxide surfaces, Phys. Rev. B 27 (1983) 7624–7635. [7] Robert T. Morrison, Robert N. Boyd, Lehrbuch der organischen Chemie, third ed., VCH Weinheim, 1986, p. 129.