Dehydration Reaction Over G-Alumina

Applied Catalysis A: General 172 (1998) 97±106 In-situ characterisation of the surface intermediates for the ethanol dehydration reaction over g-alumina under dynamic conditions Serge Golay, Ralf Doepper, Albert Renken* Institute of Chemical Engineering, Federal Institute of Technology, 1015 Lausanne, Switzerland Received 2 February 1998; received in revised form 20 March 1998; accepted 20 March 1998 Abstract The dynamic behavior of the ethanol adsorption on g-alumina were investigated at 180 and 2008C by the transient-response method coupled with FT±IR data of the catalyst surface. The existence of three adsorbates was demonstrated: a reacting species which is the precursor for the formation of the gas-phase ethene; an inhibiting species responsible for the low steadystate reaction rate; and a spectator species accumulating on the catalyst surface. Their infrared spectra indicate an ethoxidelike structure for the three adsorbates. Their C±H stretching bands can be depicted by four Lorentzians whose parameters indicate different surface environments. The surface concentration of the reacting species was determined on the basis of the transient ethene response. A value of 0.770.07 mol/kgcat was found at 1808C. The surface concentration of the spectator species was determined by ex-situ thermogravimetric experiments. A value of 0.130.01 mol/kgcat was found at 1808C. The most likely structure of this species corresponded to an ethanol molecule coordinated in a bidentate manner on Al3‡ cations, with stabilisation of the alcoholic-hydroxyl group via lateral hydrogen bonding with an adjacent surface hydroxyl. # 1998 Elsevier Science B.V. All rights reserved. Keywords: Ethanol dehydration; FT±IR; Adsorbed species; Dynamic conditions 1. Introduction The stop-effect, a dynamic phenomenon ®rst described by Koubek et al. [1,2], is observed for the reactions of deamination and alcohol dehydration over amphoteric metal-oxide catalysts, such as alumina. It consists in a temporary increase of the reaction rate when the reactant feed is switched off and replaced by an inert gas stream with the total ¯ow rate kept constant. This phenomenon was also observed by *Corresponding author. Tel.: 0041 21 693 3181; fax: 0041 21 693 3680; e-mail: [email protected] 0926-860X/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. PII S0926-860X(98)00109-4 Makarova et al. [3] with the dehydration of n-butanol on H-ZSM-5 and amorphous aluminosilicate, and by Moravek and Kraus [4] with the ethanol dehydration on alumina. Two basic models were discussed by Thullie and Renken [5,6] to describe the stop-effect. The ®rst model considers the adsorption of the reactant on two different sites S1 and S2 with a consecutive reaction involving an adsorbed intermediate (on S1) and a free site, S2. The reactant A is only weakly adsorbed on the site S2 and desorbs rapidly when the feed is switched off, increasing the number of free sites S2 available for the reaction. Therefore, the stopeffect is provoked by an inhibiting effect of the 98 S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 reactant which is suppressed by its replacement by an inert gas. The second model considers the chemisorption of the reactant on a single site with a consecutive adsorption of the reactant on the ®rst adsorbed layer. Again, the second layer is only weakly adsorbed and by switching off the reactant feed the concentration of the chemisorbed complex necessary for the reaction is increased. In that case, the stopeffect is explained by a blocking phenomenon occurring at steady-state due to the formation of a nonreactive complex. It was previously shown with the dehydration of ethanol-to-ethene on g-alumina [7] that the ®rst model (Eqs. (1)±(3)), considering the existence of two types of sites, gives a better description of the stop-effect than the second model. This model was further simpli®ed in order to decrease the number of parameters required for the modelling. The adsorption on S1 is considered to be irreversible (Eq. (1)), whereas it is weak on S2 (Eq. (2)). The exact physical nature of S1 and S2 is not completely clear. By comparison with the explanation proposed by Koubek et al. [1] for the stop-effect which is based on the mechanism for the amines elimination reactions on alumina, S1 can be associated with an acid site and S2 with a basic site. The diethylether formation is described, for simplicity, by a quasi-homogeneous reaction term (Eq. (4)). k1 A ‡ S1 ! AS1 K2 A ‡ S2 $ AS2 A : ethanol (1) E : ethene (2) AS1 ‡ S2 ! E ‡ W ‡ S1 ‡ S2 2A ! D ‡ W 2. Experimental The apparatus, catalyst and gases employed, as well as the infrared bands attribution, have been previously described in detail [7]. 2.1. Apparatus The ¯ow-apparatus consisted of two feed sections converging to a four-way valve, a ®xed-bed tubular reactor, an infrared transmission cell and a gas chromatograph. The ®xed-bed reactor, used to achieve a suf®cient ethanol conversion, was placed just in front of the infrared cell. RTD experiments showed that it behaves like a plug-¯ow reactor, and the infrared cell like a continuous stirred tank reactor. The tubular reactor was described by the tanks-in-series model using nine tanks for the catalyst compartment. 2.2. Gases and catalyst k3 k4 kinetic behaviour is responsible for the particular shape of the observed infrared signal. In this work, we paid close attention to the dynamic behaviour of the different reaction adsorbates. Their nature and formation pathway are studied by the transient-response method coupled with in-situ infrared data of the catalyst surface. They were further characterised by the calculation of their spectral line shapes using non-linear least squares ®tting of Lorentzians. Finally, thermogravimetric experiments are used to determine the surface concentration of the most stable species. D : diethylether W : water (3) (4) The adsorbed intermediates, identi®ed as surfacebound ethoxides, were monitored under dynamic conditions using in-situ infrared spectroscopy. Their similar structure produce similar infrared spectra with no characteristic band for either AS1 or AS2. These two species can only be distinguished by their different dynamic behaviour during the inert gas purge. The AS2 concentration decreases immediately, whereas AS1 is slowly consumed to form ethene. This different The carrier gas was argon (>99.99%, Carbagas, Lausanne, Switzerland) and the ethanol (>99.8%, No. 02860, Fluka Chemie AG, Buchs, Switzerland) feed was provided by a temperature-controlled bubble column fed by argon. As catalyst, g-alumina (Type Al-3982, Engelhard De Meern B.V., De Meern, The Netherlands) was used in this study. The speci®c area and pore-size distribution were determined via Sorptomatic (Type 1900, Carlo Erba Instruments, Milano, Italy). The speci®c area is 1632 m2/g and the average pore diameter 7.40.1 nm. The impurities present in the S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 catalyst are Fe2O3 (0.07%), SiO2 (0.01%) and Na2O (0.06%) [8]. A self-supporting wafer, 13 mm in diameter, was prepared by pressing alumina powder for the infrared transmission cell (7108 N/m2). Other wafers were pressed and then grinded to particles of 0.35±0.16 mm diameter for the ®xed-bed reactor. A total of 483 mg of the catalyst were used in the reactor and a 27 mg wafer was placed in the cell. 2.3. Procedure The stop-effect was measured at 180 and 2008C and a pressure of 140 kPa according to the following sequence: 108 min ethanol/830 min argon. The initial ethanol concentration was 0.25 mol/m3 and the total ¯ow rate was 200 ml (NTP)/min. Before each experiment, the catalyst was pre-treated at 4158C for 1 h under inert gas in order to remove the adsorbed ethoxide species. In addition, an infrared background spectrum was measured at the reaction temperature under inert gas. This procedure leads to a reproducible surface state for each measurement. 99 140 ml (NTP)/min. The catalyst was placed in the balance immediately after the transient experiments. The mass loss and catalyst temperature were measured as a function of time. Approximately 40 mg of catalyst were used in each experiment. 3. Results and discussion The behaviour of the infrared signal in a stop-effect experiment can be interpreted on the basis of the simultaneous gas phase and surface responses. The ethene concentration is measured at the outlet of the system ®xed-bed/infrared cell, whereas the 2970 cmÿ1 absorbance is determined in the cell. A typical example for a reaction temperature of 2008C is presented in Fig. 1, after an ethanol feeding time of 108 min. Three distinct kinetic phases (1, 2 and 3) are observed on the surface after the stop, corresponding 2.4. In-situ infrared spectra Three types of strongly adsorbed species have been already identi®ed [7]: surface hydroxyl groups, surface-bound ethoxides and surface acetate groups. The acetates are formed by the adsorption of ethanol when the catalyst is used for the ®rst time. As they do not desorb during the catalyst pre-treatment, they remain on the surface throughout all the subsequent experiments. The ethoxides are thought to be the reaction intermediates, corresponding to AS1 and AS2 in the two sites model (Eqs. (1)±(3)). Their concentration is proportional to the calculated height of the 2970 cmÿ1 absorbance band with a baseline correction between 3100 and 2600 cmÿ1 [7]. This measurement technique gave similar results to those obtained by calculating the area under the C±H stretching bands. The measured ethoxides concentration corresponds to the sum of the concentrations of AS1 and AS2. 2.5. Thermogravimetric experiments A STA 625 Rheometrics balance was used for the thermogravimetric analysis with a nitrogen ¯ow of Fig. 1. Typical responses of the gas phase and the surface for a stop-effect experiment. (TRˆ2008C, CA,0ˆ0.25 mol/m3, Qˆ200 ml (NTP)/min). 100 S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 to three different species: a rapid decrease, lasting 5± 10 min (phase 1); followed by a slow diminution; and a plateau at the end of the transient. The ®rst decrease is due to the desorption of the inhibiting species adsorbed on the basic sites and coincides with the simultaneous sharp increase of ethene formation. The second diminution (phase 2) results from the consumption of the reacting intermediates adsorbed on the acid sites to form ethene [7]. During this period, another rate enhancement is observed in the gas phase 60 min after the stop, which can be described by adding a new reaction step for the reacting species to the already developed model [9]. Due to the fact that the infrared signal is measured through the catalyst wafer, after the ®xed bed, the sensitivity of the infrared signal is relatively low. Therefore, the surface behaviour does not re¯ect the second rate enhancement. The origin of the plateau (phase 3) cannot be explained directly: an adsorbate seems to be stuck on the surface when no more ethene is produced in the gas phase. The formation of this third species is probably also responsible for the increase of the infrared signal during the ethanol feeding, observed when the gas-phase signal is already stationary. A similar phenomenon of dissymmetry between the gas phase and surface behaviour was also observed under periodic operation of the system [7]: the periods invariance was obtained for the ethene after a different ®rst cycle, whereas the infrared signal was not reproducible. Its value still increased during the next two periods before being stable. The existence of a slow parallel adsorption of ethanol under a non-reacting form was supposed to account for this difference in the dynamic behaviour of the gas phase and the surface. A different experimental procedure was applied to study the in¯uence of this third species which consisted of stopping the reactant feed before a stable surface signal was obtained. The Fig. 2 shows the transient responses after feeding times of 18, 36 and 108 min for a reaction temperature of 1808C. The temperature in¯uence on the ethene and infrared transients is important. Approximately 200 min after the stop, no more ethene is produced at 2008C, Fig. 1, whereas it is still detected 800 min after stopping the ethanol feed at 1808C, Fig. 2. From simulated curves [10], an activation energy of ca. 100 kJ/mol was estimated for the surface reaction step (Eq. (3)). The intensity of the 2970 cmÿ1 absorption band Fig. 2. Influence of the ethanol feeding time on the gas phase and the surface responses. Open symbols: 18 min, dot symbols: 36 min, solid symbols: 108 min feeding. (TRˆ1808C, CA,0ˆ0.26 mol/m3, Qˆ200 ml (NTP)/min). obtained before stopping the feed decreases in parallel with the ethanol exposition time due to the surface unsaturation. During the inert purge, the three infrared signals decrease regularly without intersecting each other. In fact, the difference between the three curves remains constant throughout all the transient. The absorbance increase compared to the 18 min ethanol feeding experiment is 0.04 for 36 min feeding and 0.07 for 108 min. Comparatively, the simultaneous gas-phase responses are unaffected by the variation of the feeding time: the same amount of ethene is produced at steady-state, and the transient responses after the stop are identical. As there is no relationship between the ethanol accumulation on the surface and the gas-phase products formation, it can be postulated that ethanol accumulates on g-alumina under the form of a non-reacting species. It is formed by the direct adsorption of ethanol on the catalyst, as the migration 101 S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 Fig. 3. In-situ infrared spectra of the catalyst surface. Spectrum (a) steady-state; (b) 12 min after the stop; (c) end of the transient. (TRˆ2008C, CA,0ˆ0.25 mol/m3, Qˆ200 ml (NTP)/min). or transformation of adsorbed ethanol would have implied the same infrared absorbance at the end of all the transients. This supposition was further con®rmed by carrying out two consecutive experiments (not shown here) with a feeding time of 36 min, without catalyst pre-treatment in-between. The resulting infrared signal is located between the ones observed for the 36 and 108 min feeding experiments, corresponding to an effective time of 72 min under ethanol. Hence, the presence of three different species on the catalyst surface can be assumed. The fact that they exhibit a completely different dynamic behaviour can be advantageously used to determine their own infrared spectra. In Fig. 3, the infrared spectra at steadystate (spectrum (a)), after 12 min of inert gas purge (spectrum (b)), and at the end of the transient (spectrum (c)) are presented. At steady-state all the adsorbates are present, whereas the inhibiting species is completely desorbed after 12 min under inert conditions, and only the spectator species remains on the surface at the end of the transient. On each spectrum, a continuous absorption is observed over the entire accessible spectral range. It has already been reported by KnoÈzinger and StuÈbner [11] and Travert et al. [12], who explained it as the consequence of very easily polarizable hydrogen-bond formation. The bands attribution, Table 1, was made by analogy with the gas-phase spectrum, and by comparison with previous works dealing with alcohols adsorption on metaloxide catalysts. The O±H group vibration at 3725 cmÿ1 belongs to a g-alumina hydroxyl group, Table 1 Infrared bands attribution for adsorbed ethanol on g-alumina Wave number (cmÿ1) Vibration Reference Wave number (cmÿ1) Vibration Reference 3725 3550 3510 2970 2930 2900 2870 (OH) (OH) (OH)  a(CH3)  a(CH2)  s(CH2)  s(CH3) [11±13] [11] [11] [11,15,16] [11,15,16] [16] [11,15,16] 1560 1480 1450 1390 1170 1115 1070  a(OCO)  s(OCO) a(CH3) s(CH3) (CO), (CC) (CO), (CC) (CO), (CC) [11,14] [11,14] [11±13,15] [11±13,15] [12,15] [12,15] [12,15] 102 S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 whereas the two bands ca. 3500 cmÿ1 belong to stabilised alcoholic-hydroxyl group [11,13]. The two bands at 1480 and 1560 cmÿ1 are those of an acetate structure. Their peculiar appearance is due to the presence of this species in the background spectrum. All the other vibrations are attributed to an ethoxide species, generally thought to be the reaction intermediate. Chemisorbed water formation, with a bending vibration ca. 1630 cmÿ1 [15], is not observed. The general shapes of the spectra are the same and only slight differences in the relative intensities of the various vibrations are noticed. Therefore, it can be stated that the three species have an ethoxide-like structure. The surface hydroxyl groups are completely restored at the end of the transient, except for the 3720 cmÿ1 band which is still present. The stabilised alcoholic-hydroxyls groups ca. 3550 cmÿ1 do not diminish to the same degree as the other bands. Hence, it can be stated that the spectator species possesses a stabilised alcoholic-hydroxyl group and interacts with a surface hydroxyl group. KnoÈzinger and StuÈbner [11] postulated an adsorption structure for the most strongly held alcohol consistent with the above observations. The oxygen atom of the alcohol molecule is coordinated in a bidentate manner to Al3‡ cations and the alcoholic-hydroxyl group is hydrogen bonded to a surface hydroxyl. This is the most likely structure of the spectator species. The measured infrared spectra are linear combinations of those of the adsorbed species. Therefore, the real spectra of these adsorbates can be calculated by judiciously subtracting them from each other (Fig. 4). The spectrum of the spectator species (spectrum (c)) is measured at the end of the transient when no more ethene is produced. The spectrum of the inhibiting species (spectrum (b)) is obtained by subtracting the spectrum (b) from the spectrum (a), and this of the reacting species (spectrum (a)) by subtracting the spectrum (c) from (b). These calculations are made for the C±H stretching region, with a baseline correction between 3200 and 2600 cmÿ1 and an arti®cial shift to zero absorbance, in order to account for the previously described continuous absorption. The resulting spectra show important differences in the relative intensities and in the positions of their bands. In order to quantify these differences, curve-®tting calculations by the non-linear least squares method were carried out on these spectra. Lorentzian functions are chosen, as they represent the natural line shapes in infrared spectroscopy [17]. Each C±H stretching group can be ®tted with four Lorentzians whose Fig. 4. C±H stretching bands of the infrared spectra. Spectrum (a) steady-state; (b) 12' after the stop; (c) end of the transient; aˆ(b)±(c); bˆ(a)±(b). (TRˆ2008C, CA,0ˆ0.25 mol/m3, Qˆ200 ml (NTP)/min). 103 S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 Table 2 Parameters of the Lorentzians peak functions Species Parameter ÿ1 Peak 1 Peak 2 Peak 3 Peak 4 Reacting (Spectrum a) Position (cm ) Height (ÿ) Relative height (ÿ) FWHH (cmÿ1) 2972 0.22 1 28 2934 0.076 0.35 22 2904 0.036 0.16 28 2873 0.11 0.50 72 Inhibiting (Spectrum b) Position (cmÿ1) Height (ÿ) Relative height (ÿ) FWHH (cmÿ1) 2970 0.10 1 30 2927 0.038 0.38 22 2901 0.011 0.11 11 2871 0.077 0.77 70 Spectator (Spectrum c) Position (cmÿ1) Height (ÿ) Relative height (ÿ) FWHH (cmÿ1) 2977 0.071 1 23 2935 0.024 0.34 21 2904 0.012 0.17 12 2892 0.025 0.35 73 parameters are given in Table 2. Due to different surface environments and adsorption sites, the relative heights and the full width of the peaks at half height vary for the three ethoxide species. The 2970 cmÿ1 peak is the highest and best described peak. Its position is similar for the reacting and the inhibiting species, but is shifted to higher values for the spectator species. Porchet [8] and Kiwi-Minsker et al. [18] have reported a similar shift of the C±H stretching vibration in the case of adsorbed methanol over selectively Mg2‡-modi®ed g-alumina which could be correlated to the strength of adsorption. Assuming this would indicate a stronger adsorption of the spectator species in comparison to the other two. A concomitant switch of the other CH3 stretching vibration (Peak 4, Table 2) to higher values is also observed. The calculated spectra can be added together to simulate the expected experimental spectrum. An excellent agreement is achieved with the measured spectrum (see Fig. 5). This shows that the spectra of Fig. 5. Comparison between measured and calculated spectra. Solid lines: experimental spectra; dashed lines: calculated spectra. Spectrum a: reacting species; b: inhibiting species; c: spectator species. (TRˆ2008C, CA,0ˆ0.25 mol/m3, Qˆ200 ml (NTP)/min). 104 S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 the adsorbed species can be well described with four Lorentzians. The surface concentrations of the reacting and the spectator species can also be determined on the basis of their dynamic behaviour. As it was shown in the Section 1, the adsorption of the reacting species is considered to be irreversible (Eq. (1)). Therefore, its consumption is only due to the ethene formation. By integrating the complete ethene molar ¯ow after stopping the ethanol feed, the number of ethene moles produced during the stop-effect is obtained, which is equal to the number of moles of the reacting species. Hence, the surface concentration of this species is calculated by dividing the integrated molar ¯ow by the total catalyst mass in the system. A value of 0.770.07 mol/kgcat is found at 1808C, which corresponds to the surface concentration of site 1, assuming a complete steady-state surface coverage at this temperature. Considering that the cross-sectional area of the ethanol molecule adsorbed on a solid surface is Ê 2 [19], this represents ca. 80% of a monolayer. 28.3 A At 2008C the integration leads to a value of 0.690.09 mol/kgcat, re¯ecting the decrease of the surface coverage with the reaction temperature. In the background spectrum of the catalyst (Fig. 6) only surface hydroxyl groups, with vibrations between 3450 and 3800 cmÿ1, and surface-bound acetates, with two peaks at 1480 and 1560 cmÿ1, are present. As no C±H stretching vibrations ca. 2900±3000 cmÿ1 are detected, it is concluded that the spectator species is removed during the catalyst pre-treatment. Consequently, it can be postulated that this species is stable at the reaction temperature but can be removed at higher temperatures. Thermogravimetric experiments were made with the freshly used catalyst (Fig. 7). The stability of the spectator species at low temperature is suf®cient to ensure reproducible measurements. The catalyst was ®rst heated at 1808C for 1 h to remove water, and then its temperature was raised to 4158C with an heating rate of 108C/min. This temperature was maintained for another hour to desorb completely the spectator species. This cycle was repeated a second time to ensure a complete desorption. The mass loss is calculated by the mass difference measured at 1808C before and after the 4158C heating phase. The further mass decrease observed at 4158C is only due to the transient behaviour of the thermal analyser [20]. The measured weight of the probe is the difference between its real weight and other forces, such as the Archimedes force and the drag forces. A temperature increase causes an augmentation of the drag forces and a decrease of the Archimedes force, the result of which is a decrease of the apparent probe weight. Therefore, the mass decrease due to the desorption of the inhibiting species must be calculated at an identical catalyst temperature of 1808C. A sur- Fig. 6. Background spectrum of the catalyst surface at Tˆ2008C under inert gas. S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 105 Fig. 7. Typical example of thermogravimetric measurement. face concentration of 0.130.01 mol/kgcat is found at 1808C. 4. Conclusions By applying simultaneous measurements of the gas phase and of the catalyst surface for the ethanol dehydration on g-alumina, the existence of three adsorbed species with an ethoxide-like structure was demonstrated. These species are: ± a reacting species which is a precursor for the ethene formation; ± an inhibiting species which desorbs immediately under inert gas; and ± a spectator species which accumulates slowly on the catalyst surface during the ethanol feeding. This last species is stable under inert gas up to 2008C, but can be removed at temperatures higher than 4008C. As it absorbs in the same infrared region that the reaction intermediates postulated in the two sites model, its formation must be taken into account when dealing with the modelling of the surface behaviour during the stop-effect. The simplest way is to subtract an offset corresponding to the absorbance measured at the end of the transient from all the infrared data. 5. Nomenclature A D E FWHH FT±IR NTP S TR W Ethanol Diethylether Ethene Full width at half height Fourier Transform infrared spectroscopy Normal conditions of temperature and pressure (08C, 1.013105 Pa) Catalytic site Reaction temperature Water 5.1. Greek letters   Bending vibration Stretching vibration 5.2. Subscripts 1 2 a cat s Site type Site type Asymmetric Catalyst Symmetric 106 S. Golay et al. / Applied Catalysis A: General 172 (1998) 97±106 Acknowledgements The authors gratefully acknowledge the funding from the Swiss National Science Foundation and the Max Buchner Forschungsstiftung, Frankfurt, Germany. References [1] J. Koubek, J. Pasek, V. Ruzicka, in New Horizons in Catalysis, Elsevier-Kodansha, Amsterdam-Tokyo, 1980, p. 853. [2] J. Koubek, J. Pasek, V. Ruzicka, in: B. Delmon, G.F. Froment (Eds.), Catalyst Deactivation, Elsevier, Amsterdam, 1980, p. 251. [3] M.A. Makarova, E.A. Paukshtis, J.M. Thomas, C. Williams, K.I. Zamaraev, J. Catal. 149 (1994) 36. [4] V. Moravek, M. Kraus, J. Catal. 87 (1984) 452. [5] J. Thullie, A. Renken, Chem. Eng. Sci. 46 (1991) 1083. [6] J. Thullie, A. Renken, Chem. Eng. Sci. 48 (1993) 3921. [7] S. Golay, O. Wolfrath, R. Doepper, A. Renken, Model Discrimination for Reactions with Stop-effect (Studies in Surface Science and Catalysis, vol. 109), Elsevier, Amsterdam, 1997, p. 295. [8] S. Porchet, Ph.D. Thesis, Federal Institute of Technology, Lausanne, Switzerland, 1996. [9] S. Golay, R. Doepper, A. Renken, accepted for oral presentation, CHISA'98. [10] O. Wolfrath, Diploma Thesis, Federal Institute of Technology, Lausanne, Switzerland, 1997. [11] H. KnoÈzinger, B. StuÈbner, J. Phys. Chem. 82 (1978) 1526. [12] J. Travert, J.-C. Lavalley, O. Saur, J. Chim. Phys. 78 (1981) 27. [13] E.C. DeCanio, V.P. Nero, J.W. Bruno, J. Catal. 135 (1992) 444. [14] V. Moravek, M. Kraus, L.V. Malysheva, E.A. Paukshtis, E.N. Yurchenko, Collect. Czech. Chem. Commun. 53 (1988) 459. [15] G.A.M. Hussein, N. Sheppard, M.I. Zaki, R.B. Fahim, J. Chem. Soc. Faraday Trans. 1 87 (1991) 2661. [16] B. Schrader, in Infrared and Raman Spectroscopy, VCH, Weinheim, 1995, p. 192. [17] A.G. Marshall, F.R. Verdun, in Fourier Transforms in NMR, Optical, and Mass Spectrometry, Elsevier, Amsterdam, 1990, p. 10. [18] L. Kiwi-Minsker, S. Porchet, R. Doepper, A. Renken, Catalyst acid/base properties regulation to control the selectivity in gas-phase methylation of catechol (Studies in Surface Science and Catalysis, vol. 108), Elsevier, Amsterdam, 1997, p. 149. [19] A.L. McClellan, H.F. Harnsberger, J. Colloid Interface Sci. 23 (1967) 577. [20] W. Zimmerer, Ph.D. Thesis, Federal Institute of Technology, Lausanne, Switzerland, 1997.