SnO2 gas sensors to carbon monoxide and hydrogen

SnO2 gas sensors to carbon monoxide and hydrogen

Sensors and Actuators B 155 (2011) 659–666 Contents lists available at ScienceDirect Sensors and Actuators B: Chemical journal homepage: www.elsevie...

825KB Sizes 1 Downloads 2 Views

Sensors and Actuators B 155 (2011) 659–666

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb

The influence of catalytic activity on the response of Pt/SnO2 gas sensors to carbon monoxide and hydrogen Ireneusz Kocemba ∗ , Jacek Rynkowski ˙ Institute of General and Ecological Chemistry, Technical University of Łód´z, 90-924 Łód´z, Zeromskiego 116, Poland

a r t i c l e

i n f o

Article history: Received 20 April 2010 Received in revised form 10 January 2011 Accepted 20 January 2011 Available online 3 March 2011 Keywords: Semiconductor oxides Gas sensors Pt/SnO2 catalyst CO oxidation TPD-O2

a b s t r a c t The article presents the results of research studies on ceramics SnO2 sensors with Pt catalysts. The role of catalysis in gas sensing mechanisms was investigated. In order to obtain samples with different catalytic activity but with identical Pt loading, the Pt/SnO2 catalysts were calcined at different temperatures (400–800 ◦ C). Structural analysis of these samples was performed. Among the sensors manufactured with Pt/SnO2 , the highest sensitivity was shown for the sensor obtained with Pt/SnO2 sample sintered at 800 ◦ C. The correlation between catalytic activity and sensor sensitivity is given. © 2011 Elsevier B.V. All rights reserved.

1. Introduction It is quite obvious that nowadays the detection and monitoring of different pollutant gases such as CO are necessary. Carbon monoxide, an extremely toxic gas, is often referred to as the “silent killer” since it is colourless, tasteless and odourless and even short exposure (<2 h) to the concentration of CO in air as low as 800 ppm is lethal. Among the possible ways of CO detection, semiconductor gas sensors (SGS) typically based on metal oxides (e.g. SnO2 , TiO2 , ZnO, In2 O3 , WO3 , etc.) are best known [1–3]. Tin oxide (SnO2 ) is the most often used oxide in gas sensor technology. It is a basic material for preparing resistive sensors, which are widely used for detection of toxic, combustible and pollutant gases. SnO2 is also useful in heterogeneous catalysis, most often used as a support of noble metals. Among others, Pt/SnO2 is a very good catalyst for the reaction of CO oxidation. Sensors made of Pt/SnO2 usually show very high sensitivity to CO [4]. The mechanism of CO detection realised by resistive sensors is connected with catalytic oxidation of this gas. A simple reaction scheme describing the process of carbon monoxide detection can be written as: O2 + 2e → 2O−

(1)

CO + O− → CO2 + e

(2)

∗ Corresponding author. Tel.: +48 42 6313134; fax: +48 42 6313128. E-mail address: [email protected] (I. Kocemba). 0925-4005/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2011.01.026

The first reaction depicts the process of oxygen adsorption. In pure air, oxygen molecules are chemisorbed on the surface of SnO2 in different ionic forms such as O2 − , O− or O2− [5], leading to the reduction of semiconductor conductivity. Upon exposure to the reductive gases such as CO (Reaction (2)), the chemisorbed oxygen species react with CO and electrons are subsequently reintroduced to the conduction band, leading to the increase in conductivity. In a successful gas sensor, the changes in conductivity must be very great and proportional to the concentration of detected gases. A more detailed mechanism is described in papers [6–8]. Since the sensing mechanism is connected with oxygen adsorption (Reaction (1)) and surface reaction (Reaction (2)), the process of detection can be considered as a kind of contact catalytic reaction. It means that typical phenomena of adsorption and catalysis are crucial for the detection mechanism. Thus some selected theories of adsorption and catalysis can be very useful in the analysis of the detection process. In the case of carbon monoxide oxidation over Pt/SnO2 catalysts, several mechanisms are discussed in literature [9–13]. For instance, in 1997 Grass and Lintz [13] suggested a probable pathway of CO oxidation via Langmuir–Hinschelwood mechanism, assuming a migration of oxygen to the reaction sites situated at the border between an oxide and noble metal particles. According to this mechanism, there are separate sites for CO and O2 adsorption. Carbon monoxide adsorbs on metal, while oxygen adsorbs on SnO2 . In this case, CO and O2 do not have to compete for the same adsorption sites on the platinum surface. Thus strong adsorption of oxygen promotes catalytic activity. On the other hand, due to Reaction (1), high adsorption of oxygen should also promote sensitivity of gas sensor.

660

I. Kocemba, J. Rynkowski / Sensors and Actuators B 155 (2011) 659–666

The well-known mechanism called “chemical sensitisation”, connected with the oxygen spillover effect [14], has been proposed for such promotion [15]. In such a case the sensitivity of sensors should be dependent on platinum dispersion, since the efficiency of spillover depends on dispersion. Dispersion is one of the most important parameters determining the activity of supported noble metal catalysts. The dispersion of the metal is defined as the ratio of the number of atoms available for chemisorption to the total number of metal atoms. In literature, there are several papers [16–20] describing the particle size effect on various elementary steps involved in CO oxidation. The investigations of supported gold catalysts have indicated that catalysts with small or ultra small gold particles show the highest activity in CO oxidation, even at room temperature [21]. In the case of platinum, the role of particle size is rather ambiguous. The main aim of the present work was to consider the relationship between the catalytic activity of Pt/SnO2 catalysts in CO oxidation and sensitivity of gas sensors prepared on the basis of these catalysts. In our previous work [22] and the papers of other authors [23,24] some relationship and correlation between the catalytic and detection properties were discussed. Platinum is a superior catalyst, but its role in the promotion of sensitivity of metal oxide gas sensors is far from being well understood and is thus still a matter of investigation. 2. Experimental

Specific surface area of SnO2 and Pt/SnO2 samples was determined by BET method using Carlo Erba Sorptomatic 1900 apparatus. X-ray diffraction (XRD) patterns were recorded at room temperature using a polycrystalline D 5000 Siemens X-ray difractometer (CuK␣ radiation). Data were collected in the range of 2 = 20–80◦ with a step size of 0.03◦ and step time of 10 s. JCPDS-ICDD files were used for phase identification. The particle morphology of samples was investigated by scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS) using S-4700 Hitachi apparatus. The species on the surface were examined by TOF-SIMS (Time-Of-Flight Secondary Ion Mass Spectrometry). The TOFSIMS measurements were performed in the static mode using an ION–TOF instrument (TOF-SIMS IV) equipped with a 25-kV pulsed 69 Ga+ primary ion gun. 2.3. TPR measurements Temperature programmed reduction (TPR) measurements were carried out in PEAK-4 apparatus, designed and constructed by us [26], using H2 /Ar (5 vol.% H2 , 95 vol.% Ar) gas mixture, a gas flow rate of 40 cm3 min−1 , in the temperature range 25–760 ◦ C with a linear ramp rate of 15 ◦ C min−1 . The presence of H2 in the effluent gas due to the reduction of the sample, was monitored by the thermal conductivity detector (TCD). The concentration and flow of the gases were controlled by calibrated mass flow controllers (Brooks).

2.1. Sample preparation 2.1.1. SnO2 support Tin chloride (SnCl4 ·5H2 O, analytical grade) and ammonia (NH3 ·H2 O, analytical grade) were used as raw materials. SnCl4 ·5H2 O was dissolved in distilled water to form a transparent solution. An aqueous solution of SnCl4 was hydrolysed with an ammonia solution. The obtained white precipitates were washed with distilled water several times, dried at 200 ◦ C for 12 h) and finally calcined at 800 ◦ C in air for 4 h. 2.1.2. Pt/SnO2 catalysts Catalysts Pt/SnO2 were prepared by wet impregnation of SnO2 with a solution of H2 PtCl6 in water of appropriate concentration to obtain 1 wt. % Pt. In order to obtain samples with different catalytic activity but with identical Pt loading, after drying at 150 ◦ C for 16 h, the samples were calcined in air for 2 h at different temperatures: 400, 500, 600, 700 and 800 ◦ C and denoted as S400, S500, S600, S700 and S800, respectively. 2.1.3. Pt/SnO2 sensors The sensor elements were of ceramic type. Each sample of Pt/SnO2 was milled and mixed with Al2 O3 as binder. The role of Al2 O3 in this construction was described in our previous work [25]. Then the samples were pressed (35 T cm−2 ) between two Pt electrodes to tablets being 5 mm in diameter and 0.2 mm thick. Next, the sensor was mounted into the measurement glass chamber with a volume of ca. 1 dm3 , and was heated in a flow of dry air (50 cm3 min−1 ) at 400 ◦ C for 72 h in order to obtain a stable value of resistance. The sensors were denoted identical to SnO2 and the catalysts which were used to their preparation (SnO2 , S400, S500, S600, S700 and S800). 2.2. Structural and morphological characterization of samples Differential thermal analysis (DTA) and thermogravimetric analysis (TGA) were carried out in the atmosphere of air (Setsys 16/18, Setaram).

2.4. Chemisorption studies Oxygen adsorption was measured by temperatureprogrammed desorption (O2 -TPD) with an application of a zirconia detector. The sample of 0.2 g was placed in a reactor in the stream of high purity oxygen with a flow rate of 40 cm3 min−1 at 500 ◦ C for 2 h. After cooling the reactor to room temperature, oxygen flow was replaced by argon (99.9999% purity) and the temperature was raised to 800 ◦ C at a heating rate of 20◦ min−1 . The titration technique was used to determine the dispersion of platinum. The sorbed hydrogen was titrated by oxygen. We assumed that stoichiometry of chemisorption (H/Pt) is equal to 1, and the equation describing this process is as follows: PtHads + 3/4O2 = PtO + 1/2H2 O

(3)

The catalyst of 0.2 g was placed in a glass tube reactor with an internal diameter of 5 mm. The catalyst was activated at 180 ◦ C for 5 h in a stream of gas mixture H2 /Ar (5 vol.% H2 , 95 vol.% Ar) with a flow rate of 40 cm3 min−1 . Then the reactor was cooled to room temperature and a flow of H2 /Ar was replaced by helium. Next, the pulses of 0.05 cm3 oxygen were introduced on the reactor using the six-way valve. The changes of oxygen concentration were detected by thermal conductivity detector (TCD). The injected oxygen reacts with chemisorbed hydrogen. If the first few injections are totally consumed, no change in signal from the detector is recorded. The amount of sorbed oxygen was calculated from the sum of oxygen pulses consumed by the catalyst. 2.5. Catalytic test Temperature programmed surface reaction (TPSR) method was used to measure catalytic activity in the reactions of CO and H2 oxidation in the temperature range 25–350 ◦ C. This process was carried out in a flow apparatus PEAK-4. The mass of the sample of 0.2 g, the stream of the reacting gas CO/air (0.06 vol.% CO, 99.94 vol.% air) and a linear increase in temperature (5◦ min−1 ) were used.

I. Kocemba, J. Rynkowski / Sensors and Actuators B 155 (2011) 659–666

661

Behind the reactor, the concentration of CO2 was constantly measured by the CO2 infrared gas analyser (Fuji Electric System Co., type ZRJ-4). The catalytic activity was expressed as reciprocal of the temperature of 50% CO conversion (1/T50 ). In the case of H2 oxidation, all parameters were the same apart from the reacting gas, which was the mixture of H2 and air (0.1 vol.% H2 , 99.9 vol.% air). To analyze the products of H2 oxidation, a gas chromatograph, equipped with thermal conductivity detector (TCD) and a column containing 5A molecular sieve (diameter – 1/8 in. and long – 1 m) was used. 2.6. Sensing characterization The sensor resistance was measured using a conventional voltage divider (circuit voltage = 10 V). The resistance of the sensors was measured in dry air and in the mixture of 0.06 vol. % CO in air or 0.01 vol. % H2 in air. Gas sensitivity was defined as the ratio of electrical resistance of the sensor in dry air (Rair ) to that in the analysed gas (RS ). 60000

101

40000

SnO2 SnO2 precursor

211

The measurements were carried out in the temperature range 25–400 ◦ C. In order to test the reliability and reproducibility, the measurements of sensitivity were repeated several times at a given temperature. The adjustment and monitoring of the total gas flow, temperature and resistance of sensors were controlled by a computer.

50000

30000 20000

200

(4)

110

Rair RS

Intensity [a.u.]

S=

Fig. 1. Thermal analysis of the SnO2 precursor.

10000 0 0

20

40

60

80

100

2θ [o]

3. Results

Fig. 2. The XRD patterns of SnO2 precursor and SnO2 sample.

3.1. Characterization of the samples Fig. 1 shows termogravimetric and differential thermal analysis patterns of SnO2 precursor dried at 200 ◦ C before calcination. Two steps of weight loss can be observed on TG curve: at 100 ◦ C (ca. 8.5%) and smaller at 220–550 ◦ C (ca. 5.8%), which can be attributed to the removal of physical and chemisorbed water, respectively. An exothermic peak at 360 ◦ C is probably connected with the transformation of hydroxy or oxy intermediates to tin oxide. The mass of SnO2 stabilizes at above 550 ◦ C. The BET surface area of SnO2 sample decreases from 123 m2 g−1 for SnO2 precursor to 2.5 m2 g−1 after calcination at 800 ◦ C. The X-ray diffraction patterns of SnO2 precursor and SnO2 , shown in Fig. 2, correspond to the cassiterite phase. The SnO2 precursor demonstrates a very low crystallinity. The average crystallite size of both samples was determined using Scherrer formula [27], according to the broadening of the (1 1 0) diffraction line [25]. The

calculated values are 20 nm for SnO2 precursor and 167 nm for SnO2 after calcination at 800 ◦ C. In order to get a better insight into the morphology of SnO2 , SEM photographs were taken (Fig. 3). The comparison of two magnifications 500× and 15,000× reveals large agglomerates of the order of 20 ␮m (Fig. 3a), which consist of small (≈0.3 ␮m) spherical particles of SnO2 (Fig. 3b). The discrepancy between SnO2 particle sizes observed by XRD and SEM methods may be caused by the tendency of separate crystallites to agglomeration [28]. The X-ray diffraction patterns of Pt/SnO2 calcined at different temperatures are shown in Fig. 4. No lines characteristic of platinum are observed on the diffractogrammes of the samples calcined at 400 and 500 ◦ C. It means that either platinum species are amorphous or the crystallites are too small to be detected by XRD method

Fig. 3. SEM picture of the surface of SnO2 : (a) magnification 500× and (b) 15,000×.

I. Kocemba, J. Rynkowski / Sensors and Actuators B 155 (2011) 659–666

Hydrogen consumption [a.u.]

662

12 10 8 6

S600

SnO

4 2 0 0

100

200

300

400

500

600

Temperature [oC] Fig. 5. TPR profile of SnO2 and S600 sample.

Tin dioxide is an easily reducible oxide, so the pre-treatment of Pt/SnO2 in hydrogen, which is essential for the purification of the surface before chemisorption measurements, should be carried out at a temperature low enough to prevent the reduction of the support. Fig. 5 shows the TPR profiles of SnO2 . The reduction of SnO2 starts at about 200 ◦ C. Therefore we applied the temperature of 180 ◦ C for the pre-treatment of the catalyst. A strong interaction between Pt and SnO2 due to the surface reduction of tin oxide was observed by authors of work [33] after reduction of Pt/SnO2 catalysts at 400 ◦ C. They postulated a core/shell structure, where the core of Pt was covered with SnO2 shell. An X-ray diffraction pattern of Pt/SnO2 sample showed some lines, which have been attributed to intermetallic compounds such as PtSn2 and PtSn4 . In order to check the chemical composition of the surface of Pt/SnO2 catalyst after its calcination at the highest temperature (800 ◦ C) followed by the activation at 180 ◦ C in hydrogen, the TOFSIMS method was used. The spectra showed different ionic species including tin such as SnO, SnO2 , SnO2 H, SnO3 H or platinum, PtO, PtO2 , PtCl, PtOCl. However, there were no intermetallic compounds like Pt–Sn, Pt–O–Sn or Pt–Sn–O. TOF-SIMS results ensured us that the temperature of 180 ◦ C was chosen properly. Dispersion of the studied catalysts, estimated on the basis of hydrogen–oxygen titration, is presented in Table 1. A significant decrease in Pt dispersion with an increase in the calcination temperature can be observed. The S800 sample shows the dispersion nearly nine times lower than the S500 one. Such values of dispersion seem to be clearly too high. Taking into consideration the surface area of SnO2 as low as 2.5 m2 g−1 , much lower values should be expected. We suppose that the overestimated values of dispersion are connected with the phenomenon of oxygen spillover. After hydrogen titration, oxygen adsorbed on platinum can spill from the metal surface on the SnO2 support. As a result, the volume of oxygen uptaken during chemisorption measurements is abnormally high. Thus calculated values of dispersion may not be strictly correct. Nevertheless, they follow a general rule that the temperature of calcination influences the platinum dispersion. For platinum supported on SnO2 the oxygen spillover is often postulated as a stage of the detection process. Thus the

Fig. 4. XRD patterns of Pt/SnO2 catalysts.

(<5 nm). Diffraction patterns of the Pt/SnO2 catalysts calcined at temperatures 600, 700 and 800 ◦ C show a line of low intensity at about 39.8◦ , which can be ascribed to the Pt (1 1 1) diffraction (ICDS 00-004-0802). The crystallite sizes of Pt particles in the catalysts under study calculated using the Scherrer formula are shown in Table 1. As expected, the size of Pt crystallites increases with the calcination temperature. Table 1 also shows BET surface area and SnO2 particle size of the Pt/SnO2 catalysts in comparison with SnO2 oxide support. The BET surface area of the catalysts is in each case slightly higher than that of SnO2 , whereas the size of SnO2 particles ca. is three times lower. The most striking feature of the results presented in Table 1 is the fact that the deposition of platinum on SnO2 markedly stabilizes the crystallite size of SnO2 . Irrespective of the calcination temperature in a wide range 400–800 ◦ C, the crystallite size of SnO2 changes insignificantly. SnO2 is a material which shows rapid grain growth during sintering [29]. According to literature data [30], the microstructure of SnO2 may be stabilised by an addition of a small amount of various metal oxides. The mechanism of this phenomenon is likely to be connected with separation of SnO2 grains. The fine oxide particle incorporated into SnO2 during the sintering process prevents the growth of “neck” between the adjacent grains stabilizing the crystallites size. It may be supposed that this mechanism also takes place in the case of investigated Pt/SnO2 samples. 3.2. TPR and chemisorptions studies Metal dispersion can be determined based on chemisorption methods. These methods have become widely applied, also for catalysts containing a very small amount of supported metal [31]. Unfortunately, there are some factors which affect chemisorption. In the case of noble metals, Strong Metal Support Interactions (SMSI) can influence chemisorption properties of catalysts to a great extent [32]. The real mechanism of metal support interaction is not always clear, but it is usually observed for reducible supports and after the reduction of catalyst at high temperatures. Table 1 Structural (XRD), textural (BET) and chemisorption properties of the catalysts. Sample

Surface area [m2 /g]

SnO2 particle sizea [nm]

Pt particle size [nm]a

Dispersion [%]

Oxygen adsorption [␮mol]/gcat

SnO2 S400 S500 S600 S700 S800

2.5 3.8 4.1 3.5 3.2 3.2

170 54 51 58 60 61

– <5 <5 6.2 7.1 11.2

– 36.5 74.1 15.2 12.2 8.7

– 233 199 166 63 33

a

Based on Scherrer’s formula.

I. Kocemba, J. Rynkowski / Sensors and Actuators B 155 (2011) 659–666

663

60

x3 x2

Sensitivity (Ra/RCO)

50

Operating temperature [oC] 100

40

150 200

30

250 300

20

350

10 0 S400

S500

Fig. 6. TPD-O2 profiles of Pt/SnO2 catalysts.

S600

S700

S800

SnO2 SnO2

Sensor

Fig. 8. Sensitivity of sensors to CO at different operation temperatures.

chemisorption of oxygen over Pt/SnO2 has a crucial meaning for an understanding of the mechanism of detection. Temperature programmed desorption of oxygen (TPD-O2 ) is a very convenient tool of investigating of oxygen adsorption over SnO2 [34,35]. The TPD-O2 profiles for all the investigated catalysts are presented in Fig. 6. The profiles show two overlapped stages of desorption, relatively well distinct for the sample S400 (maxims at 320 and 750 ◦ C). The area of the recorded peaks strongly decreases with an increase in the calcination temperature. Table 1 shows the total amount of oxygen desorbed from the catalysts. The sample S800 shows the lowest adsorption, whereas the sample S400 shows the highest one. An amount of oxygen desorbed from S400 catalyst is ca. seven times higher than that for the S800 sample. 3.3. Catalytic test The results of CO oxidation during the TPSR are presented in Fig. 7. The activity of the catalysts decreases with the increase in their calcination temperature. For the most active catalyst (S400) the reaction starts at room temperature while CO is completely oxidized at ca. 120 ◦ C. The conversion of CO for the least active Pt/SnO2 sample (S800) starts at 130 ◦ C and is completed at 210 ◦ C. All Pt/SnO2 catalysts under study were very active in the reaction of hydrogen oxidation. 100% conversion was observed already at room temperature.

Fig. 8 presents the sensitivity of sensors to CO, prepared from SnO2 and Pt/SnO2 catalysts at different operating temperatures in the temperature range 100–350 ◦ C. The sensitivity of the sen-

100

Catalyst S400 S500 S600 S700 S800 SnO2

Conversion [%]

80

60

40

20

0 50

100

3.5. Discussion In order to explain the obtained results, we assumed that all factors which are responsible for properties of sensors may be categorized into three main groups

3.4. Sensing characterization

0

sor made of SnO2 is generally low and slightly increases with the increase in the operation temperature, reaching its maximum about 4 at 350 ◦ C. As it was mentioned earlier, it is a well-known phenomenon that an addition of platinum to SnO2 leads to the increase in its sensitivity. Thus the sensors made from catalysts are much more sensitive. The maximum sensitivity is reached for the operating temperature 150 ◦ C. Moreover, one can observe an increase in the sensors sensitivity with the increase in Pt/SnO2 calcination temperature, which is most significant for S700 and S800 sensors. The S800 catalyst, of which the most active sensor was made, showed the largest size of Pt particles and the lowest dispersion. Therefore, high dispersion does not guarantee high sensitivity although parallelism between these magnitudes could be expected. In order to check how other reductive gases change the sensitivity of sensors, the tests of sensitivity to hydrogen were carried out. Fig. 9 summarises the results. The sensitivity to H2 for all sensors is markedly lower than that to CO. Moreover, sensitivities of sensors made of SnO2 and Pt/SnO2 are similar. One can conclude that in that case of H2 detection, both the presence and the state of platinum on SnO2 surface are of minor importance.

150

200

250

300

350

Temperature [o C]

Fig. 7. CO conversion versus temperature for SnO2 and Pt/SnO2 catalysts.

- electronic, - physicochemical, - catalytic. Electronic factors show the correlations between properties of sensors and electronic structure of gas sensitive layers. The physicochemical factors reveal the relationship between chemical constitution of sensitive layers, their physical properties (surface area, metal dispersion, etc.) and the detection properties. The third group demonstrates relationships between the catalytic behaviours and the gas-sensing properties of semiconductor oxides catalytic. On the other hand, both electronic and physicochemical factors determine catalytic properties of the gas sensitive layers. The catalytic properties of these layers arise from their chemical compositions, physical properties and electronic structure. Thus all factors which are responsible for detection properties of sensors are focussed in catalytic properties of gas sensitive layer. The general scheme describing this relationship is shown in Fig. 10. Therefore taking this scheme into account, different behaviour of investigated sensors can be explained by an analysis of the mech-

664

I. Kocemba, J. Rynkowski / Sensors and Actuators B 155 (2011) 659–666

6 Operating temperature [oC]

Sensitivity [Ra/RH 2]

5

100

4

150 200

3

250 2

300 350

1 0 S400

S500

S600

S700

S800

SnO2

Sensor Fig. 9. Sensitivity of sensors to H2 at different operation temperatures.

anism of CO and H2 oxidation over Pt/SnO2 catalysts. According to Eq. (1 and 2), the high adsorption of oxygen should promote both catalytic activity and sensitivity of gas sensors. Fig. 11 shows the changes in catalytic activity and sensitivity of catalysts/sensors under study versus O2 adsorption measured by the TPD-O2 process. The catalytic activity and ability to oxygen adsorption change in a parallel way. On the other hand, the highest values of sensitivity were obtained for the sensors made of the catalysts showing low oxygen adsorption and catalytic activity (S800 and S700). Pt/SnO2 catalysts have been extensively characterized because of their efficiency at a low temperature of CO oxidation. The most characteristic feature of such systems is that they show a significantly higher catalytic activity in CO oxidation than either Pt or SnO2 alone [36]. The effect is synergistic and apparently involves separate but complementary roles for platinum and tin dioxide phases. According to the mechanism given by authors of work [36], the synergetic effect can be explained assuming that CO is chemisorbed by platinum, but it is not chemisorbed by tin dioxide. However, oxygen is chemisorbed on SnO2 surface. The adsorbed species can migrate to the sites situated at the border between

oxide and platinum particles, where the reaction occurs. Since CO and O2 do not have to compete for the same surface sites, Pt/SnO2 catalyst shows much higher activity in CO oxidation than “conventional” catalysts supported on classical supports such as Pt/Al2 O3 . In our work, this relatively high catalytic activity is observed for S400 and S500 samples, which can be easily understood, taking into consideration the fact that these catalysts are characterised by the highest Pt dispersion and the highest ability to oxygen adsorption. To summarize, the following equations can be proposed to describe the process of CO detection in dry air at the temperature of maximum sensitivity: 1. Pure air a. direct adsorption O2 on SnO2 surface SnO2 + 0.5O2 + e → SnO2 · · ·O− b. adsorption of O2 on Pt crystallites Pt/SnO2 + 1/2O2 → (Pt· · ·O∗)/SnO2

Physicochemical factors

(Pt· · ·O∗)/SnO2 + CO → CO2 + Pt/SnO2

Pt/SnO2 + CO → (Pt· · ·CO)/SnO2

Fig. 10. Relationship between catalytic and detection properties of sensors.

O−

d. reaction with inum particles

100

0.008

80 60

0.004

40 20

0

0 199

233

Oxygen adsorption [μmol g-1] Fig. 11. Sensitivity and catalytic activity as a function of O2 adsorption.

Sensitivity [Ra/RCO]

140 120

[1/o C]

Catalytic activity as 1/T 50

(Pt· · ·O∗)/SnO2 + (Pt· · ·CO)/SnO2 → CO2 + Pt/SnO2

160 0.012

(9)

c. reaction with adsorbed O* on Pt crystallites

180

0.016

166

(8)

b. adsorption of CO on Pt crystallites

Detection properties

63

(7)

2. In the presence of CO a. reaction with O adsorbed on Pt via Rideal–Eley mechanism

Catalytic properties

33

(6)

c. adsorption of O2 on SnO2 via spillover of O* (Pt· · ·O∗)/SnO2 → Pt/(O− · · ·SnO2 ) − e

Electronic factors

(5)

(10)

ions on the border between oxide and plat-

(Pt· · ·CO)/SnO2 + (O− · · ·SnO2 ) → CO2 + Pt/SnO2 + e e. direct reaction with mechanism

O−

(11)

adsorbed on SnO2 via Rideal–Eley

SnO2 · · ·O− + CO → CO2 + SnO2 + e

(12)

f. reaction with O2 via Rideal–Eley mechanism (Pt· · ·CO)/SnO2 + O2 → CO2 + Pt/SnO2

(13)

Reaction (12) can be neglected because at the temperature of maximum sensitivity, (150 ◦ C) the process of CO oxidation over pure SnO2 was not recorded. Thus, it must be marked here that

I. Kocemba, J. Rynkowski / Sensors and Actuators B 155 (2011) 659–666

only Reaction (11) can change conductivity of sensors and its rate is of great significance for the detection sensitivity. Of course, this reaction has Langmuir–Hinshelwood character, so its rate at a constant temperature depends on the concentration of reactive species adsorbed on the surface of the catalyst [37]. For the two gases A and B, which are simultaneously chemisorbed on the catalyst surface and can react with each other, the reaction rate may be expressed as:

vs = ks A B

(14)

where, ks is the rate constant,  A and  B are the surface coverage of gas A and B, respectively. Therefore taking Reactions (8–13) into account, the rate of CO oxidation on Pt/SnO2 catalyst can be expressed by three equations. The first one is determined by Reaction (11)

v1 = k2 [O− ] [CO]

(15)

the second one by Reactions (10)

v2 = k1 [O∗ ] [CO]

(16)

and the third one by Reaction (13) via Rideal–Eley mechanism

v3 = k3 [CO] pO2

(17)

where pO2 is oxygen partial pressure The total rate of oxidation vs will be given by the sum of v1 , v2 and v3 :

vs = v1 + v2 + v3

(18)

Because only Reaction (11) can change conductivity of sensors, the relation between v1 and (v2 + v3 ) can be the factor which determines the sensitivity of sensors. Three cases can be distinguished: I. v1 > (v2 + v3 )

(19)

II. v1 = (v2 + v3 )

(20)

III. v1 < (v2 + v3 )

(21)

In the first case – I, the sensitivity should be the highest because O− coverage will strongly decrease during detection (reaction). In the second (II) and third (III) cases, the sensitivity should be low or very low although the catalytic activity can be high. It is so because in such cases coverage O− is either constant or decreases insignificantly. Thus, both the preparation of Pt/SnO2 and conditions of sensors operation should be assorted in such a manner to favour Reaction (11). Taking the above considerations into account, one can assume that sensors made of S400 and S500 samples of Pt/SnO2 catalysts show low sensitivity due to their high catalytic activity. High dispersion promotes CO oxidation according to Reactions 10 and 13. Thus, for these sensors the relationship expressed by inequality I is true. The sensor made of S600 sample, which shows moderate catalytic activity, satisfies the equation II. However, the sensors which were made of S700 and S800 samples, the catalytic activity and dispersion of which were poor, satisfy inequality III. Furthermore, when the condition described by inequality III is fulfilled and when the sample Pt/SnO2 shows low ability to oxygen adsorption, sensitivity can be additionally intensified by carbon monoxide spillover. Carbon monoxide adsorbed on platinum in the system Pt/SnO2 can spill over SnO2 surface as an electron donor gas. This process is described by equation: (Pt· · ·CO)/SnO2 → Pt/(SnO2 · · ·CO+ ) + e

(22)

Under such conditions, the resistance of the sensor strongly decreases. Besides, CO on the surface of SnO2 can react with lat-

665

tice oxygen creating nonstoichiometric tin dioxide with very low resistance. This process can be described by the reaction: Pt/(SnO2 · · ·CO+ ) + oL 2− → Pt/SnO2−x + xCO2 + e

(23)

where, O−2 L means lattice oxygen and SnO2−x nonstoichiometric tin dioxide. This process will be more effective when the sensor shows low ability to oxygen adsorption. It is well-known that CO oxidation may occur with the participation of lattice oxygen from the oxide surface due to the so-called Mars-van Krevelen mechanism [37]. In this mechanism the molecules are oxidized by consuming lattice oxygen of the oxide catalyst, which in turn is re-oxidized by gas-phase oxygen. We suppose that this mechanism describes the properties of S800 sensors during CO detection, giving an ultra high sensitivity. Such a mechanism of detection can be indirectly confirmed by the measurements of time after which the sensor recovered its initial resistance. We carried out such comparative experiments for S400 and S800 sensors. When the flow of CO in air was interrupted and pure air was passed through the test chamber, sensor S400 regained its initial resistance after 5 min., while S800 sensor only after 35 min. The time for the sensor S800 was so long because, due to the detection process, its surface was partly reduced Eq. (23) and an optimum temperature of sensor operation (150 ◦ C) was not sufficiently high for quick reoxidation of Pt/SnO2 surface, which was not the case for a very active S400 catalyst. The authors of the work [38] claim that at 100 ◦ C, when stoichiometric mixture of carbon monoxide and oxygen labelled by 18 O2 isotope was passed over a common isotope of 1% Pt/SnO2 catalyst, 85% 12 C18 O2 and 15% 12 C16 O18 O were formed. This means that lattice oxygen from SnO2 had to be involved in the CO oxidation. Thus these studies also confirm that Reactions (22) and (23) can occur in the temperatures range at which high sensitivity of S800 sensor was observed. Considering the catalytic effects in a similar way, we can explain the low sensitivity of the sensors to H2 . Let us recall that all Pt/SnO2 samples investigated in this work showed very high activity in the reaction of H2 (100% conversion at room temperature). Therefore we can suppose that during H2 detection an oxidation of hydrogen over sensors surface can be described by two the following reactions: (Pt· · ·O∗)/SnO2 + 3/2H2 → H2 O + (H· · ·Pt)/SnO2

(24)

2(H· · ·Pt)/SnO2 + O2 → H2 O + (Pt· · ·O∗)/SnO2

(25)

Neither of these reactions change the concentration of oxygen ions adsorbed on SnO2 surface. That is why sensitivity is low in spite of very high catalytic activity of hydrogen oxidation. 4. Conclusion On the basis of our work, we can state that catalytic activity in CO oxidation and sensitivity to CO of Pt/SnO2 sensors do not change parallel. Sensitivity is determined by the rate of CO oxidation. When it is too low or too high, the sensitivity is low in spite of high Pt dispersion. There are optimum values of the reaction rate at which the highest sensitivity can be reached. Moreover, if simultaneous adsorption of oxygen is low, favourable conditions for CO spillover from platinum into SnO2 surface arise. The presence of CO on the surface can strongly influence conductivity of the sensor as an electron donor gas. Carbon monoxide on the surface of SnO2 can also react with lattice oxygen, creating nonstoichiometric tin dioxide with very high conductance. Under such conditions sensors can show exceptionally high sensitivity.

666

I. Kocemba, J. Rynkowski / Sensors and Actuators B 155 (2011) 659–666

References [1] G.S. Henshaw, R. Ridley, D.E. Williams, Room-temperature response of platinised tin dioxide gas-sensitive resistors, J. Chem. Soc., Faraday Trans. 92 (1996) 3411–3417. [2] W. Fliegel, G. Behr, J. Werner, G. Krabbes, Preparation, development of microstructure, electrical and gas-sensitive properties of pure and doped SnO2 powders, Sens. Actuators B: Chem. 18–19 (1994) 474–477. [3] D.E. Williams, Semiconducting oxides as gas-sensitive resistors, Sens. Actuators B: Chem. 57 (1999) 1–16. [4] N. Barsan, J.R. Stetter, M. Findlay, W. Gopel, High performance gas sensing of CO: comparative tests for SnO2 -based semiconducting and for electrochemical sensors, Sens. Actuators B: Chem. 66 (2000) 31–33. [5] M. Batzil, U. Diebold, The surface and materials science of tin oxide, Prog. Surf. Sci. 79 (2005) 47–154. [6] D.S. Vlachos, C.A. Papadopoulos, J.N. Avaritsiotis, Characterisation of the catalyst–semiconductor interaction mechanism in metal-oxide gas sensors, Sens. Actuators B: Chem. 44 (1997) 458–461. [7] V. Brynzari, G. Korotchenkov, S. Dmitriev, Theoretical study of semiconductor thin film gas sensitivity: attempt to consistent approach, J. Electron. Technol. 33 (2000) 225–235. [8] N. Barsan, M. Schweizer-Berberich, W. Gopel, Fundamentals and practical applications to design nanoscaled SnO2 gas sensors a status report, Fresenius J. Anal. Chem. 365 (1999) 287–304. [9] G.C. Bond, L.R. Malloy, M.L. Fuller, Oxidation of carbon monoxide over palladium–tin(IV) oxide catalysts: an example of spillover catalysis, J. Chem. Soc. Chem. Commun. (1975) 796. [10] G. Dong, J. Wang, Y. Gao, S. Chen, A novel catalyst for CO oxidation at low temperature, Catal. Lett. 58 (1999) 37–41. [11] N.D. Gangel, N.M. Gupta, R.M. Iyer, Microcalorimetric study of the interaction of CO, O2 and CO + O2 with Pt/SnO2 catalysts, J. Catal. 126 (1990) 13. [12] D.R. Schryer, B.T. Upchurch, B.D. Sidney, K.G. Brown, G.B. Hoflund, R.K. Herz, A proposed mechanism for Pt/SnOx catalysed CO oxidation, J. Catal. 130 (1991) 314–317. [13] K. Grass, H. Lintz, The kinetics of carbon monoxide oxidation on tin(IV) oxide supported platinum catalysts, J. Catal. 172 (1997) 446. [14] W. Curtis, J.L. Falconer, Spillover in heterogeneous catalysis, Chem. Rev. 95 (1995) 759–788. [15] N. Yamazoe, New approaches for improving semiconductor gas sensors, Sens. Actuators B: Chem. 5 (1991) 7–19. [16] T. Tanabe, Y. Nagai, K. Dohmae, H. Sobukawa, H. Shinjoh, Sintering and redispersion behavior of Pt on Pt/MgO, J. Catal. 257 (2008) 117–124. [17] N.P. Siswana, D.L. Trimm, Metal support interactions in the catalytic oxidation of carbon monoxide, Catal. Lett. 46 (1997) 27–29. [18] K. Arnby, A. Torncrona, B. Andersson, M. Skoglundh, Investigation of Pt/-Al2 O3 catalysts with locally high Pt concentrations for oxidation of CO at low temperatures, J. Catal. 221 (2004) 252–261. [19] A. Bourane, D. Bianchi, Heats of adsorption of the linear CO species on Pt/Al2 O3 using infrared spectroscopy: impact of the Pt dispersion, J. Catal. 218 (2003) 447–452. [20] A. Bourane, S. Derrouiche, D. Bianchi, Impact of Pt dispersion on the elementary steps of CO oxidation by O2 over Pt/Al2 O3 catalysts, J. Catal. 228 (2004) 288–297. [21] M. Haruta, N. Yamada, T. Kobayashi, S. Iijima, Gold catalysts prepared of coprecipitation for low-temeperature oxidation of hydrogen and carbon monoxide, J. Catal. 115 (1989) 301. [22] I. Kocemba, S. Szafran, J. Rynkowski, T. Paryjczak, Relationship between the catalytic and detection properties of SnO2 and Pt/SnO2 systems, Adsorpt. Sci. Technol. 20 (2002) 897–907. [23] C.B. Lim, S. Oh, Microstructure evolution and gas sensitivities of Pd-doped SnO2 based sensor prepared by three different catalyst-ad-dition processes, Sens. Actuators B: Chem. 30 (1996) 223.

[24] P.K. Dutta, M. Frank De Lucia, Correlation of catalytic activity and sensor response in TiO2 high temperature gas sensors, Sens. Actuators B: Chem. 115 (2006) 1–3. [25] I. Kocemba, S. Szafran, J. Rynkowski, T. Paryjczak, The properties of strongly pressed tin oxide-based gas sensors, Sens. Actuators B: Chem. 79 (2001) 28–32. [26] I. Kocemba, The peak-4 instrument for studying catalysts by temperatureprogrammed methods, Przem. Chem. 82 (2003) 142. [27] H.P. Klug, L.E. Alexander, X-ray Diffraction Procedures, Wiley, New York, 1967. [28] G. Korotcenkov, V. Brinzari, J. Schwank, M. DiBattista, A. Vasiliev, Peculiarities of SnO2 thin film deposition by spray pyrolysis for gas sensor application, Sens. Actuators B: Chem. 77 (2001) 244. [29] V.V. Kovalenko, A.A. Zhukova, M.N. Rumyantseva, A.M. Gaskov, V.V. Yushchenko, I.I. Ivanova, T. Pagnier, Surface chemistry of nanocrystalline SnO2 : effect of thermal treatment and additives, Sens. Actuators B: Chem. 126 (2007) 52–55. [30] G. Yang, Z. Haibo, Z. Biying, Monolayer dispersion of oxide additives on SnO2 and their promoting effects on thermal stability of SnO2 ultrafine particles, J. Mater. Sci. 35 (2000) 917–923. [31] T. Paryjczak, Gas Chromatography In Adsorption And Catalysis, PWN Warszawa, 1986. [32] D.E. Resasco, R.J. Fenoglio, M.P. Suarez, J.O. Cechini, Different strong metalsupport interaction effects on Rh/TiO2 and Pt/TiO2 , catalysts, J. Phys. Chem. 90 (1986) 4330–4333. [33] T. Okanishi, T. Matsui, T. Takeguchi, R. Kikuchi, K. Eguchi, Chemical interaction between Pt and SnO2 and influence on adsorptive properties of carbon monoxide, Appl. Catal. A: Gen. 298 (2006) 181–187. [34] S. Saukko, U. Lassi, V. Lantto, M. Kroneld, S. Novikov, P. Kuivalainen, T.T.J. Rantala, J. Mizsei, Exprimental studies of O2 –SnO2 surface interaction using powder thick films and monocrystalline thin films, Thin Solid Films 490 (2005) 48–53. [35] J.M. Kanervo, T.J. Keskitalo, R.I. Slioor, A.O.I. Krause, Temperature-programmed desorption as a tool to extract quantitative kinetic or energetic information for porous, catalysts, J. Catal. 238 (2006) 382–393. [36] R.K. Herz, CO-Oxidation Catalysts: Low-Temeparture CO Oxidation Over Noble Metal Reducible Oxide (NMRO) Catalysts, 3076, NASA Conference Publication, 1990, p. 21. [37] G. Centi, B. Cornils, W.A. Hermann, R. Schlogel, C.-H. Wong (Eds.), Catalysis from A to Z, Wiley-VCH, Weinheim, 2000. [38] B.T. Upchurch, G.M. Wood, D.R. Schryer, R.V. Hess, I.m. Miller, E.J. Kielin, Rare-isotope and kinetic studies of Pt/SnO2 catalyst, 3076, NASA Conference Publication, 1990, p. 193.

Biographies Ireneusz Kocemba received his diploma in Chemistry at the Technical University of Łódz´ (Poland) in 1982. Since then he has been working in the Institute General and Ecological Chemistry where he received a PhD degree in the field of adsorption and catalysis in 1988. His current research concentrates on gas sensors based on semiconductors oxides and in the investigations of correlations between catalysis and detection. Jacek Rynkowski was graduated from the Technical University in Łódz´ I 1970. (PhD-1978, DSc-1988). Professor in the Institute General and Ecological Chemistry, ´ Present research interests: characterisation of supTechnical University of Łódz. ported metallic and bimetallic catalysts, environmental catalysts, semiconductor gas sensors.