4.3. Catalyst characterization techniques
Experimental surface characterization methods were used to study ageing-induced changes in the active metals and washcoat oxides. The characterization techniques included both microscopic and spectroscopic methods, such as Scanning Electron Microscopy (SEM) and X-ray Diffraction (XRD). Several characterization techniques are available to study solid surfaces and the properties of catalysts, and no single characterization method can be used to explain the basis for the catalyst deactivation phenomena of three-way catalysts (Wachs 1992, Niemantsverdriet 2000). In the proceeding sections, the characterization techniques utilized in this thesis are presented in detail.
4.3.1. Scanning Electron Microscopy
Scanning Electron Microscopy (SEM) was used in this thesis for high magnification imaging and elemental analysis. A Jeol JSM-6400 scanning electron microscope equipped with an energy dispersive spectrometer (EDS) was used for the analysis. In the pretreatment stage, flat pieces of fresh and aged catalysts were cut, and either potted in epoxy or fastened with a carbon tape in order to obtain side or top views of the catalyst respectively. Catalysts were polished down to 1 µm using diamond paste and coaled prior the analysis to avoid the accumulation of charge. The accelerating voltage and current in the measurements were 15 kV and 12 nA, respectively, and the resolution of the instrument was 3.5 nm (35 kV). SEM-EDS resources were used in the Institute of Electron Optics at the University of Oulu.
4.3.2. Physisorption analyses
Measurements of gas adsorption isotherms are widely used for determining the surface area and pore size distribution of solids. The use of nitrogen as the adsorptive gas is recommended if the surface areas are higher than 5 m2/g (Serwicka 2000). The first step in the interpretation of a physisorption isotherm is to identify the isotherm type. This in turn allows for the possibility to choose an appropriate procedure for evaluation of the textural properties. Non-specific Brunauer-Emmett-Teller (BET) method is the most commonly used standard procedure to measure surface areas, in spite of the oversimplification of the model on which the theory is based. The BET equation is applicable at low p/p0 range and it is written in the linear form (Wachs 1992):
where
p is the sample pressure,
p0 is the saturation vapour pressure,
na is the amount of gas adsorbed at the relative pressure p/p0,
nam is the monolayer capacity, and
C the so-called BET constant.
Equation (8) can be applied for determining the surface areas and pore volumes from adsorption isotherms, if the adsorption isotherms are of type IV according to IUPAC classification. The pore size distributions can be calculated from desorption isotherms. The pores are usually classified according to their widths as micropores (diameter less than 2 nm), mesopores (diameter between 2 and 50 nm) and macropores (diameter exceeding 50 nm) (Hayes & Kolaczkowski 1997). Several approaches have been developed to assess the micro- and mesoporosity, and to compute pore size distribution from the adsorption-desorption data. All of these involve a number of assumptions, e.g. relating to pore shape and mechanism of pore filling. (Serwicka 2000)
In this thesis, physisorption measurements were carried out to characterize catalysts before and after the ageings. Specific surface areas (m2/g) and pore volumes were measured according to the standard BET method, as described above, by using a Coulter Omnisorp 360CX. The specific surface areas and pore volumes were obtained from N2 adsorption isotherms at -196°C by assuming the cylindrical shape of pores. Catalysts were outgassed in a vacuum at 140°C overnight before the measurements. All the BET values in this study were measured within a precision of ± 5%. Pore size distributions for micropores as well as meso- and macropores were calculated from N2 -desorption isotherms by differential HK (Horvath-Kawazoe) and BJH (Barrett-Joyner-Hallender) methods respectively (see Anon 1992). Since the monoliths showed systematically lower BET values than the crushed samples after similar ageing procedures, all the BET values presented in this thesis have been determined for the metallic monoliths with a standard shape and mass.
4.3.3. Chemisorption analyses
Chemisorption measurements were carried out in order to determine the dispersions of Pd and Rh metal particles, monolayer capacities and the amount of active metal in the catalysts. Hydrogen and carbon monoxide were used as the adsorbate gases. H2-chemisorption and CO-chemisorption experiments were carried out close to room temperature (30°C) by volumetric adsorption method by using a Coulter Omnisorp 360CX (at the University of Oulu) and a Sorptomatic 1900 (at Kemira Metalkat Oy), respectively. The accuracy of the measurements was estimated to be better than ± 5%. The experimental procedure for the H2-chemisorption measurements is presented in Table 8. In the chemisorption procedure, the temperature ramping rate of the furnace was 10°C/min. As shown in Table 8, the adsorption of H2 was measured twice. The difference between these two measurements was assumed to be the amount of irreversibly adsorbed H2, which is further used to calculate the dispersion values.
Table 8. The experimental procedure for the H2-chemisorption measurements at the University of Oulu (Anon 1992).
| 1. Flow of He at 150°C for 5 minutes followed by 10 minutes at 375°C |
| 2. Evacuation at 375°C for 10 minutes |
| 3. Reduction in flowing H2 at 375°C for 10 minutes followed by 5 minutes at 400°C |
| 4. Evacuation at 400°C for 20 minutes followed by 10 minutes at 30 °C |
| 5. Leak test at 30 °C |
| 6. First analysis with H2 at 30 °C |
| 7. Evacuation at 30 °C for 30 minutes |
| 8. Second analysis with H2 at 30 °C |
Chemisorption has long been employed as a valuable technique for rapid evaluation of the active metal dispersions and hence the particle sizes of supported metals (Gasser 1985). This method has, however, undergone severe criticism, since the underlying assumptions of the stoichiometry between adsorbate gas and precious metal and the particle geometry may not be true, in particular in the case of small particles (Di Monte et al. 2000). Furthermore, in the case of metal oxides (such as CeO2 and Ce-Zr-mixed oxides) in contact with active metals, adsorbed H2-molecules can also diffuse from the active metal particles to the washcoat. This spillover effect can be reduced by lowering the adsorption temperature, as has been reported by Bernal et al. (1993) and Fornasiero et al. (1995).
Active metal dispersions and particle sizes are calculated by assuming the stoichiometry factor between chemisorbed gas molecules and surface metal atoms. In this thesis, chemisorption measurements are based on the assumption of the stoichiometry of 2:1 for H2 and the stoichiometry 1:1 for CO adsorption, respectively, and regardless of the particle size. The stoichiometric ratio may depend on the precious metal particle size, a reason why caution should be exercised when comparing the dispersion values of different catalysts. However, it is assumed that all the aged catalysts as well as the fresh catalyst exhibit rather large metal particle sizes due the low dispersion values (below 30%). Therefore, the changes in dispersion values presented in this thesis reflect the structural changes induced by ageings, such as the sintering of the precious metals. As well, the chemical correctness behind the stoichiometry assumptions is not relevant because, in this case, the relative dispersion values are more interesting than the absolute ones.
4.3.4. Temperature-programmed techniques
Transient techniques are powerful in the investigation of catalytic surface phenomena. Temperature-programmed methods can provide useful information on solid surfaces, their interactions with adsorbed gas molecules, and thermal stability of surface desorption states (Falconer & Schwarz 1983, Malet 1990, Salvador & Merchán 1998). In this research, temperature-programmed desorption (TPD) of NO is used to obtain information on the ageing-induced changes in the adsorption-desorption behaviour of NO, and to evaluate how these changes are associated with the deactivation of catalysts. NO was chosen as the adsorbate gas molecule because it is the key compound in the purification process of exhaust gases of gasoline engines. (Armor 1992, Taylor 1993)
NO-TPD measurements were carried out in a vacuum or under the carrier gas flow by using NO as the adsorbate gas. NO/Ar (5%) gas mixture and carrier gas Ar (99.998%) were supplied by AGA Ab, Sweden. The gas flows were regulated with mass flow controllers (Bronkhorst High-Tech B.V., The Netherlands). In the pretreatment stage, catalysts were evacuated at two hours and then reduced under hydrogen flow for 10 minutes at 500°C, followed by 15 minutes at 550°C. The catalysts were then cooled down by evacuation before the measurements. NO/Ar (5%) was adsorbed on the catalyst surface at room temperature for 10 minutes. The NO-TPD measurements were carried out in a quartz chamber at temperatures 30°–800°C. The volume of the catalyst monolith was 1.4 cm3 and the mass of the washcoat about 250 milligrams. Pressure in the reactor chamber was below 10-4 mbar (down to 10-7 mbar) during the vacuum measurements, whereas the carrier gas measurements were performed at atmospheric pressure. A small portion of the product flow was taken through a capillary into a quadrupole mass spectrometer (Carlo Erba Instruments Q.T.M.D., Italy) (see Anon 1988). The mass numbers of interest molecules were monitored and the results were stored in a PC. The sensitivity factors of the corresponding molecules were determined by using appropriate gas mixtures. A scheme of the reactor system for TPD measurements is presented in Fig. 16.
The TPD profiles were measured as a function of temperature at a linear heating rate (30°C/min). Prior to every TPD run, the excess of gas was removed by evacuation until no residual gases were detected. A West TP190 temperature controller was used to control the heating rate. The heating rate was found to affect the position of the peak maximum. A higher temperature of maximum desorption was observed by increasing the heating rate (Acke 1998, Niemantsverdriet 2000). Therefore, the heating rate was kept constant during the experiments. Furthermore, blank tests were carried out with the uncoated metal foil and no desorption or reactions were detected.
4.3.5. X-ray diffraction
X-Ray diffraction (XRD) was used to investigate the bulk phases present in the sample and to determine the ageing-induced solid-solid phase transformations. X-rays are energetic enough to penetrate into the material and their wavelengths are of the same order of magnitude as interatomic distances in solids. Thus, a collimated beam of X-rays is diffracted by the crystalline phases in the sample according to Bragg’s Law (Atkins 1995):
where
λ is the wavelength of the X-rays,
d is the distance between two atomic planes in the crystalline phase,
n is the order of the diffraction, and
θ the incoming diffraction angle.
The XRD diffractograms presented in this study were recorded on a Siemens D5000 diffractometer employing nickel-filtered Cu Kα radiation (λ = 1.5406 Å, 40 kV, 30 mA) at 0.020° intervals in the range 20° ≤ 2θ ≤ 75° with 1s count accumulation per step directly from the catalyst foils. The aged catalysts were prepared as mentioned above (see section 4.2). Diffraction patterns were assigned using the PDF database supplied by the International Centre for Diffraction Data (PDF2- Diffraction Database File). XRD resources of the Institute of Electron Optics at the University of Oulu were used.
4.3.6. X-ray Photoelectron Spectroscopy
While XRD gives information on the bulk phases present in the sample, X-ray Photoelectron Spectroscopy (XPS) can be used to study the sample surface. In this thesis, XPS was mainly utilized to study the ageing-induced changes in the chemical states of active metals. The XPS measurements were carried out at Tampere University of Technology, Finland, by using a Kratos XSAM 800 electron spectrometer with a base pressure less than 1×10-8 Torr (1.33×10-6 Pa). An analytical chamber was connected to a preparation chamber in which the catalysts could be heated up to 300°C in H2 atmosphere (99.9999%, AGA Ltd.). Non-monochromatic Mg Kα X-rays were used as a primary excitation. No severe charging of the catalysts was observed. The hemispherical energy analyzer was run in a fixed analyzer transmission (FAT) mode with the pass energy of 38 eV. Due to the known difficulties in referencing the binding energies to the C 1s line (Bhattacharya et al. 1997), the Al 2p line at 74.2 eV was used as a reference together with the C 1s line at 284.6 eV. All binding energies quoted in this study were measured within a precision of ±0.2 eV. The reducibility of the Rh oxide phases formed in the different ageing procedures was investigated by annealing the catalysts in situ in 400 mbar of static H2 at 300°C for 30 minutes followed by the XPS measurement. The adequacy of the reduction treatment time was verified by repeating reduction-measurement cycles several times. No remarkable additional changes in the XPS spectra were found after the first reduction treatment. The intention was to discover the relative amount of easily reducible Rh oxide species, and not to try to reduce all oxide species to metallic. A detailed description of the theory behind the XPS and the measurement system can be found in Suhonen (2002).
4.3.7. Activity measurements
Catalytic activities were determined by laboratory scale light-off experiments to compare the catalysts after the ageings. Catalyst light-off is determined as the temperature of 50% conversion, which is used to indicate the efficiency of an automotive exhaust gas catalyst (the lower the light-off temperature, the more active the catalyst is). In addition to light-off temperatures, the conversions of CO and NO at 400°C were also determined.
The experimental set-up for the activity measurements is presented in Figures 17 and 18. Catalytic activities were determined by using a simple model reaction: the reduction of NO by CO in lean and rich conditions. The composition of test gas mixture is presented in Table 9. Before the measurements, the catalysts were reduced in a hydrogen flow (99.98%, AGA Ltd.) at 500°C for 10 minutes, followed by 15 minutes at 550°C. Activity measurements were carried out at atmospheric pressure by using a cylindrical catalyst with a volume of 1.4 cm3 (length 28 mm and diameter 8 mm). In the measurements, the gas-solid reactor system equipped with mass flow controllers (Brooks 5850TR), magnetic valves for flow selection, tubular furnace with a quartz reactor and analysis instruments were used. The total gas flow during the experiments was 1 dm3/min corresponding to the feed gas hourly space velocity (GHSV) of 43 000 h-1. The temperature of a catalyst was increased from room temperature up to 400°C, with a linear heating rate of 20°C/min. The concentrations of CO, NO, CO2, N2O and NO2 as a function of temperature were measured every 5 seconds by an FTIR gas analyser (GasmetTM CR2000) and the gas flow was controlled by mass flow controllers (Brooks 5850TR). Furthermore, the effect of poisoning on the catalytic activity was evaluated by changing the flow direction in the catalyst. Blank tests were carried out with the uncoated metal foil to ensure the inactivity of metal foil in the thermal treatments. In the following discussion, differences larger than ± 5°C in the light-off temperatures and ± 1% in the conversions of CO and NO can be regarded as statistically significant.
Table 9. Composition of the test gas mixture for the activity measurements.
| Component | Lean | Rich |
|---|---|---|
| CO | 800 ppm | 1200 ppm |
| NO | 1000 ppm | 1000 ppm |
| N2 | Balance | Balance |
Figure 18. Activity measurement system equipped with the GASMETTM gas analyzer (University of Oulu).
The activity of some aged pre-catalysts and main catalysts was also tested at Kemira Metalkat Oy, Finland. In these measurements, the conversions of CO, HC and NOX were measured as a function of catalyst’s temperature using a test gas mixture simulating the real exhaust gas composition. Catalyst light-off temperatures (T50 values) as well as conversions at 400°C were determined. Furthermore, OSC was measured for fresh and aged monoliths by CO-O2 exchange experiments at constant adsorption temperatures of 450°C, 600°C and 750°C. The consumption and the adsorption of O2 and CO were determined by mass spectrometer. (Härkönen et al. 2001)
4.3.8. Chemical analyses
Chemical analysis provides the information on elemental composition of the catalyst. The ‘wet’ and ‘dry’ analyses were performed due to low quantities of poisons present in the catalysts. The dry analysis was carried out beyond the SEM-EDS sensitivity, as described in section 4.3.1. In a wet analysis, the fresh and aged catalysts were dissolved in an acidic solution in order to determine the quantities of the most important catalytic poisons (Ca, P, S, Pb, Mg and Zn) and the amounts of active metals (Pd and Rh). These elements were typically present in small quantities, and unevenly distributed in the catalyst. Therefore, separated samples were prepared from the inlet and outlet parts of the engine-aged and vehicle-aged catalysts. For the chemical analysis, 0.010 to 0.050 g of the washcoat (scraped from the monolith) was dissolved in an acidic solution (HNO3, HCl, HF and H3BO3) and subjected to the digestion of the sample in the microwave oven (Milestone MLS 1200). This resulted partly in an incomplete dissolution of the analysed solids. The decomposed sample was analysed quantitatively by plasma atomic emission spectrometry (Pye Unicam 7000 ICP-AES) in the Trace Element Laboratory at the University of Oulu.