Applied Catalysis A: General 237 (2002) 171–180 Dehydrogenation of methanol over copper-containing catalysts T.P. Minyukova a,∗ , I.I. Simentsova a , A.V. Khasin a , N.V. Shtertser b , N.A. Baronskaya a , A.A. Khassin a , T.M. Yurieva a a Boreskov Institute of Catalysis, 5, Pr. Lavrentieva, Novosibirsk 630090, Russia b Novosibirsk State University, 2, Ul. Pirogova, Novosibirsk 630090, Russia Received 18 October 2001; received in revised form 23 May 2002; accepted 28 May 2002 Abstract The catalytic properties in methanol dehydrogenation of copper metal formed as a result of reduction by hydrogen of copper-containing oxides with different structure: copper chromite (tetragonally distorted spinel), copper hydroxysilicate (Chrysocolla), and copper-zinc hydroxysilicate (Zincsilite) have been studied. This process proceeds via successive reactions: (I) 2CH3 OH = CH3 OOCH + 2H2 and (II) CH3 OOCH = 2CO + 2H2 . The methyl formate selectivity for the catalysts studied was close to 1.0 at low methanol conversion, X ≤ 0.1, where the dehydrogenation process is represented by reaction (I), occurring far from its equilibrium. At 0.2 ≤ X ≤ 0.55, the selectivity decreases with increasing conversion, and the ratio of the activities in successive reactions may serve as a comparative characteristic for the catalysts. At high conversions, when reaction (I) is close to its equilibrium, selectivity is independent of the properties of studied catalysts and depends on the methanol conversion. Reaction (I) shows low sensitivity to the state of metal copper of reduced catalysts and, hence, low sensitivity to the composition and structure of oxides-precursors. The catalysts’ activity in reaction (II) greatly depends on the state of metal copper in the catalysts. It was assumed that the catalyst activity in methyl formate conversion to CO and H2 and, hence, the selectivity of methanol dehydrogenation with respect to methyl formate in the region of moderate methanol conversion depends on the strength of interaction between metal copper particles and catalyst oxide surface, which is determined by the composition and structure of oxide-precursor. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Methanol dehydrogenation; Methyl formate; Cu-containing catalysts 1. Introduction CH3 OOCH = 2CO + 2H2 . The consecutive reactions of methanol dehydrogenation yielding methyl formate as an intermediate were shown to occur over the copper-containing catalyst of methanol synthesis [1,2]: In addition, a direct decomposition of methanol into CO and H2 is likely to occur (the reaction reverse of methanol synthesis): 2CH3 OH = CH3 OOCH + 2H2 , ∗ Corresponding author. Tel.: +7-3832-34-41-09; fax: +7-3832-34-30-56. E-mail address: min@catalysis.nsk.su (T.P. Minyukova). (I) CH3 OH = CO + 2H2 . (II) (III) The reactions differ essentially in their thermodynamic properties. According to Stull et al. [3], at 473 K the values of equilibrium constants, Kp , for these reactions are Kp (I) = 3.4 × 10−2 , Kp (II) = 0926-860X/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 2 ) 0 0 3 2 8 - 9 172 T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 2.0 × 104 , Kp (III) = 2.6 × 10. It means that the equilibrium conditions are most unfavorable for reaction (I), significantly limiting its occurrence. The equilibrium conditions are more favorable for reaction (III), and reaction (II) has practically no thermodynamic restriction. Studies on dehydrogenation of methanol are of practical interest since methyl formate is a valuable product used for synthesis of various organic compounds [4]. Studies in this field are also important for the theory of catalytic syntheses from CO and H2 . Recently, many studies were performed on dehydrogenation of methanol over copper oxide-containing catalysts, among them copper-silicon [5–8], coppertitanium and copper-zirconium [6], copper-zinc and copper-cerium [9], copper-chromium [10], copperaluminum [11], and copper-containing catalysts of methanol synthesis [1,2]. These studies provided the data on the catalysts’ activity and selectivity with respect to the formation of methyl formate, and on the kinetics and mechanism of the methanol dehydrogenation reactions. Copper oxide-containing catalysts are active in the methanol dehydrogenation reaction only after their reductive activation with hydrogen. The opinions on the nature of active component are different, e.g. the authors of [11] believe all three copper states, Cu2+ , Cu+ , and Cu0 , to be active. At the same time, many researchers [5–10] supposed that the activity is determined by the particles of metal copper forming at the oxide surface as a result of hydrogen activation. In the latter case, no exhaustive explanation of the dependence of catalytic properties of copper on its dispersion and morphology is made in the literature. The data on the effect of oxide-precursor structure and composition on the catalytic properties of copper are also lacking. In the present work, a study of the catalytic properties was performed for copper, obtained via the reductive activation with hydrogen of coppercontaining oxide compounds of various composition, namely: • copper chromite (CuCr2 O4 ) with a tetragonally distorted spinel structure; • copper hydroxysilicate (Cu:Si = 0.14:0.86) with Chrysocolla mineral structure; • copper-zinc hydroxysilicate (Cu:Zn:Si = 0.13:0.30: 0.57) with Zincsilite mineral (TOT trioctahedral phyllosilicate) structure. The structure and composition of reduced catalysts were studied previously [12–15] using various physico-chemical methods. The resulting particles of metal copper in these catalysts were shown to have different morphology and dispersion. 2. Experimental procedure 2.1. Catalyst preparation The samples of copper hydroxysilicate with Cu: Si = 0.14:0.86 and copper-zinc hydroxysilicate with Cu:Zn:Si = 0.13:0.30:0.57 were prepared by the deposition–precipitation method at a gradually increasing pH from the appropriate mixtures of aqueous solutions of copper and zinc nitrates with aerosol suspended in urea, and calcined at 723 K in flowing air [16]. The copper chromite sample CuCr2 O4 was prepared via thermal decomposition of coprecipitated copper-chromium hydroxycarbonate carried out in air at 1173 K. Then the catalysts were reduced in flowing hydrogen at the following temperatures: 533 K for copper hydroxysilicate (CS-0.14), 533 K for copper-zinc hydroxysilicate (CZS-0.13-1), 653 K for copper-zinc hydroxysilicate (CZS-0.13-2), and 573 K for copper chromite (Cu-Cr). 2.2. Catalyst characterization Thermal analysis (STA), infrared spectroscopy (IRS), X-ray diffraction analysis (XRD), and electron microscopy (EM) were used at every preparation step to control the conformity of the obtained samples’ composition and structure to the results of previous studies on the catalysts forming under the given conditions [12–15]. Thermal analysis was made with a Netzsch STA409 with 50–100 mg samples in platinum crucibles in hydrogen flowing at 150 ml/min and increasing the temperature at 5 K/min. Infrared transmission spectra were recorded in the region of 400–4000 cm−1 with a Bomem MB-102 T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 Fourier spectrometer. The samples were prepared as pellets with CsI. The X-ray diffraction study was performed with a Siemens D-500 diffractometer under monochromatic Cu K␣ radiation. The electron microscopy study was conducted using a JEM-100CX transmission electron microscope. The samples for the study were prepared as an alcohol suspension supported over perforated carbon film-supports after ultrasound dispersion. The surface area of metal copper in the reduced samples was determined via titration with nitrous oxide using pulse chromatography [17]. The surface area was calculated from the amount of fast-interacted N2 O for exclusion of copper metal bulk oxidation influence in accordance with [18]. The calculation was performed using the assumption that one oxygen atom is bound by two surface copper atoms, and the number of copper atoms/1 m2 is 1.46 × 1019 [19]. The metal copper surface area was evaluated also from the XRD and EM data. vity detector. No analytic measurement of the hydrogen concentration CH2 was made, it was calculated from the methyl formate and carbon monoxide concentrations on the basis of stoichiometric ratios according to Eq. (1): CH2 = 2Cmf + 2CCO . The catalytic properties of the samples were studied by the continuous flow method at atmospheric pressure in a temperature range of 423–543 K. Catalyst grains of size 0.25–1.0 mm, mixed with the quartz particles, were loaded in the reactor. Before the reaction, the initial samples were activated in a hydrogen flow at the temperatures indicated above. The temperature was elevated at the rate of 2 K/min, and then the catalyst was kept in hydrogen flow for 2 h at a specified temperature. After the activation, the initial mixture, containing 0.14–0.16 molar fractions of methanol in helium, was fed to the catalyst. The temperature of experiment was set, and after an hour the composition of the reaction mixture at the reactor outlet was successively analyzed several times (to the point of establishing a constant 0 (in composition). The concentrations of methanol, Cm the initial mixture) and Cm (in the reactor outlet mixture), of methyl formate, Cmf , and carbon monoxide, CCO , were determined chromatographically and expressed in terms of their molar fractions. No substantial amounts of other carbon-containing products were found in the experiments. The analysis was made with a flame-ionization detector and a thermal conducti- (1) Conversion degree (X) and selectivity (S) were calculated by Eqs. (2) and (3), respectively: X= 2Cmf + CCO , 0 (1 − C Cm mf − 2CCO ) (2) S= 2Cmf . 2Cmf + CCO (3) Contact time (t) was determined as a ratio between the catalyst volume loaded to the reactor (v) and adjusted to normal conditions volume of the initial steam-gas mixture fed in a unit time (V): t= 2.3. Catalytic properties 173 v . V (4) Changing the amount of catalyst loaded and the rate of steam–gas mixture feeding allowed to vary the contact time from 0.02 to 3 s. The catalyst volume was found as the ratio of the catalyst mass to its bulk weight. The reaction rates were calculated from the expression, which follows from the differential material balance for the gas flow. In the case of low concentrations of the reaction mixture components it may be represented by the following approximation: W = aNa dC × , Vm ρ dt (5) where W is the reaction rate (molecule/g s); a the stoichiometric coefficient; C the concentration of component being consumed or formed (molar fraction); Na the Avogadro number; Vm = 2.24 × 104 cm3 /mol the volume of 1 mol gas under the normal conditions; ρ the catalyst bulk weight (g/cm3 ) and t is the contact time (s). Note that, despite the presence of the catalyst bulk weight in Eq. (5), the calculated rate values are actually independent of ρ and relate to the catalyst weight unit, since implicitly ρ enters also into denominator of the t value. 174 T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 Fig. 1. Experimental DTG (solid line) and DTA (dash line) profiles of the temperature programmed reduction in the hydrogen flow: (1) sample CS-0.14, uncalcined; (2) sample CZS-0.13, calcined at 723 K in the inert gas flow. 3. Results and discussion 3.1. Catalyst composition and structure 3.1.1. Copper hydroxysilicate of Chrysocolla structure Chrysocolla mineral with the Cu:Si ratio 1:1 contains a jelly-like sediment with crystal admixtures, in which the presence of Si4 O10 layers was shown [16]. The structure of hydroxysilicate retains with a decrease in the Cu:Si ratio to 1:6 [14,16]. Chrysocolla remains stable up to 1200 K in air and in an inert medium. Reduction of copper hydroxysilicate with hydrogen yielding the metal copper proceeds with exothermic effect (the maximum at 455 K) accompanied by a weight loss. The observed effect of weight loss at 520 K characterizes a structural rearrangement with evolution of water. After the reduction at this temperature, the sample consists of silicon oxide, with copper particles located on its surface [15]. The STA, IRS, XRD, and EM studies of the CS-0.14 sample, obtained in the present work, showed the sample to be copper hydroxysilicate of Chrysocolla structure. The specific surface area of the CS-0.14 sample calcined at 723 K was 390 m2 /g. Fig. 1 shows DTG and DTA curves for the CS-0.14 sample reduction with hydrogen (curve 1). As expected, in accordance with [14,15], the sample reduction occurs at 455 K, and the structure is ruptured at 520 K. The 533 K was chosen as the temperature of sample activation. After the reduction at 533 K, the sample constitutes silicon oxide, with metal copper particles of size 30–50 Å located at its surface, as it was determined from XRD. According to the results of N2 O titration, the surface area of metal copper comprised 31.0 m2 /g catalyst (Table 1). Table 1 Characteristics of reduced samples Sample CS-0.14 CZS-0.13-1 CZS-0.13-2 Cu-Cr Cu content in precursor (at.%) Cu0 content (at.%) Size of particles (Å) Cu0 surface area (m2 /g catalyst)a Cu0 surface area (m2 /g catalyst)b 14.0 14.0 30–50 ∼27 31.0 13.0 4.3 30–70 ∼7 8.0 13.0 8.4 30–70 ∼13 7.0 33.3 14.0 50 × 100 × 100 ∼8 6.0 a b Copper surface area was calculated from the structural data. Surface area was determined by N2 O titration. T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 3.1.2. Copper-zinc hydroxysilicate of Zincsilite structure Zincsilite has a layered structure and belongs to the Smectite group. Zincsilite is represented by the formula Znx (Zn3−x 䊐x )[Si4 O10 ](OH)2 × nH2 O, where Zn3−x are the zinc ions located at octahedral positions of the layers formed by [Si4 O10 ] tetrahedrons, Znx are the zinc ions located in water interlayer, and 䊐x are the vacancies [20]. Copper ions partially substitute zinc ions in each structural position, being distributed among the ‘layer’ and ‘interlayer’ in a 2:1 ratio [14,15]. Copper ions are reduced with hydrogen in the following way [15]: (a) Copper ions located between [Si4 O10 ] layers (hereinafter referred to as copper ions located in the Zincsilite ‘interlayer’) are reduced at a considerable rate in the temperature range of 493–533 K producing metal copper particles. Copper ions from the ‘interlayer’ are reduced completely. (b) Copper ions at octahedral positions of the layers formed by [Si4 O10 ] tetrahedrons (hereinafter referred to as copper ions located in the Zincsilite ‘layer’) are reduced giving metal copper as the temperature increases from 553 K up to the point of rupturing the structure at 753 K. At 653 K, ca. 50% of these ions are reduced, with the structure being retained. The STA, IRS, XRD, and EM studies of the CZS-0.13 sample, obtained in the present work, showed the sample to be copper-zinc hydroxysilicate of Zincsilite structure. The specific surface area of the CZS-0.13 sample calcined at 723 K was 420 m2 /g. Fig. 1 shows DTG and DTA curves for the CZS-0.13 sample reduction (curve 2). The sample reduction producing metal copper proceeds in two steps: at 530 and 610 K, which completely agree with the results of [15]. So, the activation temperatures were chosen at 543 K (CZS-0.13-1 sample) and 653 K (CZS-0.13-2 sample). According to IRS, XRD and EM data, the reduced samples represented hydroxysilicates of Zincsilite structure, with metal copper particles of size 30–70 Å located at its surface. According to the results of N2 O titration, the surface area of metal copper amounted 8.0 m2 /g for the CZS-0.13-1 sample, and 7.0 m2 /g for the CZS-0.13-2 sample (Table 1). 175 3.1.3. Copper chromite Copper chromite CuCr2 O4 has a spinel-like structure with tetragonal distortion of the lattice (a = 8.537 and c = 7.792 Å) and normal distribution of cations [21]. As a result of interaction with hydrogen at 453–643 K, a partial (ca. 40% at 573 K) redox substitution of copper ions for protons occurs. In this case, the spinel structure remains, and the metal copper forms the plane particles of size 50 Å × 100 Å × 100 Å at the spinel surface. Copper particles are well edged and linked epitaxially with the spinel surface [12]. At higher temperatures, chromite is destructed in hydrogen to give metal copper at the surface of chromium oxide [22]. The STA, IRS, XRD, and EM studies of the Cu-Cr sample, obtained in the present work, showed the sample to be a spinel with the lattice parameters indicated above. The specific surface area of the Cu-Cr sample was 3.3 m2 /g catalyst. The sample under study was reduced at 573 K. According to the data of physico-chemical studies, its composition corresponded completely to that of the sample obtained in [12]. N2 O titration of the Cu-Cr sample reduced at 573 K showed the surface area of copper to be 6.0 m2 /g. Table 1 lists the surface area values for copper produced as a result of catalyst reduction with hydrogen. These values were obtained by titration with nitrous oxide and calculated on the basis of structural studies. One may see that the surface area values for copper hydroxysilicate and copper chromite obtained by different methods agree satisfactorily. In the case of copper-zinc hydroxysilicate, an agreement between the surface area values found by different methods is observed only after the reduction at 543 K (CZS-0.13-1 sample). As a result of catalyst reduction at 653 K (CZS-0.13-2 sample), the amount of metal copper increases up to 8.4 at.%, while the surface area determined by N2 O titration shows no increase (it even decreases to some extent), the particles size remains in the same range. These facts are discussed below. 3.2. Catalytic properties First of all, it is noteworthy that all the catalysts under study exhibit catalytic activity in methanol dehydrogenation only after reductive activation with 176 T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 Fig. 2. The contact time dependencies of methanol (䊉), methyl formate (䉱) and carbon monoxide (䊏) concentrations in the reactor outlet mixture and virtual methanol concentration (䊊) at 473 K for the Cu-Cr catalyst. Fig. 4. The contact time dependencies of methanol (䊉), methyl formate (䉱) and carbon monoxide (䊏) concentrations in the reactor outlet mixture and virtual methanol concentration (䊊) at 473 K for the CZS-0.13-1 catalyst. hydrogen. Figs. 2–5 show the contact time dependencies of the methanol, methyl formate, and carbon monoxide concentrations in the reaction mixture at the reactor outlet, obtained for catalysts pre-reduced with hydrogen: for Cu-Cr (Fig. 2), CS-0.14 (Fig. 3), and CZS-0.13-1 (Fig. 4) at 473 K, and for CZS-0.13-2 at 543 K (Fig. 5). Note that the observed dependencies reflect the changing concentrations of components over the whole length of the catalyst bed at a certain linear rate of the reaction mixture flow. At a flow rate Fig. 3. The contact time dependencies of methanol (䊉), methyl formate (䉱) and carbon monoxide (䊏) concentrations in the reactor outlet mixture and virtual methanol concentration (䊊) at 473 K for the CS-0.14 catalyst. Fig. 5. The contact time dependencies of methanol (䊉), methyl formate (䉱), and carbon monoxide (䊏) concentrations in the reactor outlet mixture and virtual methanol concentration (䊊) at 543 K for the CZS-0.13-2 catalyst. T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 of 1 cm/s, the dependencies will accurately reflect the dependence of component concentrations on the catalyst bed length. For all catalysts, at very short contact time, the methanol consumption is accompanied by methyl formate accumulation with no CO formation. Thus, the selectivity of methyl formate formation approaches unity. As the contact time increases, so does the methyl formate concentration. Then, after reaching its maximum, it decreases to a rather low value. CO formation becomes noticeable at certain values of contact time, which differ with the catalysts, and the CO concentration increases with increasing contact time. This function has an inflection point at the maximal concentration of methyl formate. Such a character of kinetic dependencies is clearly shown for catalysts Cu-Cr, CS-0.14, and CZS-0.13-1 at 473 K, while at this temperature in the case of the CZS-0.13-2 catalyst, its lower activity at the studied contact times does not allow to reach the maximum of the methyl formate concentration function and the inflection point of the CO concentration versus contact time curve. For the CZS-0.13-2 catalyst, the character of these dependencies is revealed completely at a higher reaction temperature, 543 K. Thus, for all the catalysts studied, the kinetic regularity typical of successive reactions is observed: the inflection point for the contact time function of the end product (CO) corresponds to the maximum of the intermediate product (methyl formate) concentration curve. Hence, under the given conditions, methanol dehydrogenation proceeds stepwise by reactions (I) and (II), and no direct dehydrogenation of methanol to CO and H2 by reaction (III) is observed. The dotted line in Figs. 2–5 shows the contact time ∗. dependence of the virtual methanol concentration Cm Virtual concentration is the proposed methanol concentration, which would provide, for the observed process, the methanol equilibrium with methyl formate and hydrogen at their observed concentrations. The ∗ values were found as methanol virtual pressures, Cm Pm∗ , from the expression for equilibrium constant of reaction (I): 1/2 Pm∗ = Pmf PH2 1/2 , (6) Kp where Kp is the equilibrium constant of reaction (I), 177 Pmf and PH2 are the partial pressures of methyl formate and hydrogen expressed as fractions of atmosphere and, at atmospheric pressure, numerically equal to the experimental concentrations of components expressed in molar fractions. This relationship shows how closely reaction (I) approaches its equilibrium under the conditions of successive occurrence of reaction (II). The analysis of reaction kinetics in terms of virtual partial pressure was many times used before [23]. The virtual concentration of methanol increases rapidly in the range of short contact times and low methanol conversions, and attains a value close to the experimental concentration of methanol at t = 0.1–0.5 s for various catalysts. Further, with increasing contact time, its value closely follows the experimentally observed dependence of methanol concentration on the contact time. The catalysts’ activity in reaction (I) was characterized by the rate of methyl formate production, W1 , corresponding to methanol concentration in the initial mixture, via differentiation of the contact time dependence of methyl formate concentration, and calculated according to formula (5) (in this case, a = 1). This variable may be considered as the “initial rate” of reaction (I) in the sense that its value characterizes the rate of the process at the beginning of the catalyst bed and corresponds to the initial composition of the reaction mixture. As evident from experimental data, at low conversion no distorting effect of the reverse reaction or the successive conversion of methyl formate by reaction (II) on the determined rate of forward reaction (I) takes place. The catalysts’ activity in reaction (II) was characterized by the maximum value of methyl formate conversion rate W2 . The W2 rate was determined via differentiation of the contact time dependence of carbon monoxide concentration at the inflection point and calculated by formula (5) (in this case, a = 1/2). Using the carbon monoxide concentration versus contact time dependence allows to exclude the effect of methyl formate generation and consumption in reversible reaction (I) on the determined value of W2 . The results of the study are summarized in Table 2, which lists the rates of reactions (I) and (II) in reference to unit catalyst weight and to unit surface area of metal copper, at 473 K and molar fractions of methanol Cm = 0.14–0.16 and methyl formate Cmf = 0.018. 178 T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 Table 2 Catalysts’ activity in methanol dehydrogenation at 473 K Sample W1 (molecule/g s) W2 (molecule/g s) W1 (molecule/m2 s)a W2 (molecule/m2 s)a W1 /W2 CS-0.14 3.0 1.3 9.7 4.2 ∼2 1019 × × 1019 × 1017 × 1017 CZS-0.13-1 CZS-0.13-2 1018 1018 4.5 6.8 5.6 8.5 ∼7 × × 1017 × 1017 × 1016 3.3 × 1.6 × 1017 4.7 × 1017 2.3 × 1016 ∼20 Cu-Cr 7.7 8.0 1.3 1.3 10 × × × × 1018 1017 1018 1017 See text for details. a The values of copper surface area determined by N O titration were used (see Table 1). 2 These conditions were taken as standard for the comparison of catalyst activity. The obtained experimental data allow only an approximate evaluation of apparent activation energies for reactions (I) and (II). For all catalysts, the values of activation energy of reaction (I) were shown to vary in the range of 75–90 kJ/mol, while for reaction (II) in the range of 70–90 kJ/mol. Note that in [7] the apparent activation energy of reaction (I) for a copper-silicon catalyst with a copper content of 4 wt.%, obtained by ion exchange, was found to be 63 kJ/mol, while the apparent activation energy of reaction (II) for a copper-containing methanol synthesis catalyst was found in [2] to be equal to ca. 60 kJ/mol. Since all the studied catalysts exhibit catalytic activity in methanol dehydrogenation only after their reductive activation with hydrogen, it seems reasonable to compare the catalysts properties by the values of their specific activity, i.e. the reaction rate referred to unit surface area of metal copper. One may see from Table 2 that the catalysts’ specific activities in reaction (I) differ no more than by a factor of 2.5. Thus, with the activation energies also not being much different, we may conclude that reaction (I) shows low sensitivity to the state of metal copper in reduced catalysts and, hence, low sensitivity to the composition and structure of the oxide-precursor. Specific activities in reaction (II) for the studied catalysts show a more considerable variation: the extreme values differ ca. 20-fold. It means that reaction (II) is sensitive to the state of metal copper and, hence, sensitive to the composition and structure of oxides-precursors. Fig. 6 shows the selectivity of methyl formate generation versus methanol conversion. In the region of low conversion values, X ≤ 0.1, the selectivity for all studied catalysts are close to unity. In this region, the methanol dehydrogenation is represented mainly by reaction (I) proceeding far from its equilibrium. The catalysts’ activities in reaction (II) are not high enough to exert a considerable negative effect on selectivity in this range of X values. As the conversion degree increases, the selectivities decrease, and their values diverge for different catalysts first gradually and then, in the X range from 0.3 to 0.55, very sharply. In this range, reaction (I) approaches its equilibrium, while the methyl formate concentration, and hence the rate of its conversion, attain the maximum values. One can see from Fig. 6 that at X = 0.3 the selectivity for all the catalysts is 0.9 and higher; at X = 0.45 it differs significantly in the sequence CZS-0.13-2, CZS-0.13-1, Cu-Cr and CS-0.14 and is equal 0.8, 0.7, 0.7 and 0.5, correspondingly; at X = 0.5 the selectivity is equal, 0.7, 0.3, 0.3 and 0.2, correspondingly. In the region of Fig. 6. Methyl formate selectivity vs. methanol conversion at 473 K: ( ) for the CS-0.14 catalyst, (䊏) for the Cu-Cr catalyst, (䉱) for the CZS-0.13-1 catalyst and (䊊) for the CZS-0.13-2 catalyst. T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 methanol conversion 0.2 ≤ X ≤ 0.55, the catalyst selectivity may be compared by the ratio of catalyst activities in successive reactions, W1 /W2 , with its values for the studied reactions listed in Table 2. Indeed, as the X value is increasing the range of 0.2 ≤ X ≤ 0.55, the decrease of the selectivity to methyl formate for CZS-0.13-2 catalyst (with W1 /W2 = 20.4) is not so dramatic as for the other catalysts. The most sharp decrease of the selectivity takes place for CS-0.14 catalysts having W1 /W2 = 2.3. In the X > 0.55 region, where reaction (I) is close to its equilibrium, shifting with the X increase, the selectivities are low and practically independent on the catalyst properties. The methyl formate concentration in this region agrees satisfactorily with the expression for the equilibrium constant of reaction (I) at the concentrations of methanol and hydrogen varying due to occurrence of reaction (II). In this case, the selectivity is determined by a degree of methanol conversion. Consider the dependence of metal copper activity in the reaction of methyl formate conversion to CO and H2 on the structure of oxide-precursor. The CS-0.14 sample was shown to be the most active in reaction (II) and, hence, the least selective in methyl formate generation. After activation with hydrogen, the catalyst constitutes silicon oxide, with metal copper distributed over its surface as particles of size 30–50 Å and regular spherical shape. Copper chromite (Cu-Cr sample) exhibits a lower activity in reaction (II) and a higher selectivity in methyl formate generation. As a result of sample activation with hydrogen, metal copper forms as plane particles of size 50 Å × 100 Å × 100 Å, linked epitaxially with the spinel surface. Evidently, the interaction of copper particles with the surface of reduced chromite is stronger than the interaction of copper particles with the surface of silicon oxide. This may lead to a decrease in the rate of reaction (II), probably, due to the decreasing strength of methyl formate adsorption. For Zincsilite, the particle shape was not studied, but copper ions in Zincsilite are known to occupy two structural positions (in the ‘interlayer’ and in the ‘layer’) being reduced at various temperatures. At 543 K, copper ions from the ‘interlayer’ are reduced, and the CZS-0.13-1 catalyst shows an activity in reaction (II) close to that of copper chromite. In addition, at 653 K (CZS-0.13-2 catalyst), copper ions from the ‘layer’ are reduced. This considerably decreases the activity in reaction 179 (II) and, hence, considerably increases the methyl formate selectivity: the W1 /W2 ratio for CZS-0.13-1 catalyst is ∼7, while for the CZS-0.13-2 catalyst it is ∼20. To explain this result, it should be assumed that copper particles, formed from the ‘interlayer’ copper ions, change their properties as a result of activation at a higher temperature. For example, one may assume that copper atoms formed by the reduction of copper ions from the Zincsilite ‘layer’ crystallize mainly at the copper particles emerged after reduction of copper ions from the ‘interlayer’, so that the copper surface area does not increase (Table 1). The observed decrease in specific activity in reaction (II) may be explained by a partial decoration of copper particles, probably, with Si–Ox clusters. The decoration indicates a strong interaction of metal particles with the Zincsilite surface (strong metal–support interaction (SMSI)), which may change the electronic properties of copper, and hence, its catalytic properties. Thus, we are inclined to assume that the selectivity of methyl formate generation depends on the strength of the interaction between metal copper particles and the catalyst oxide surface due to the interaction effect on the catalyst activity in reaction (II). The character of the metal particle interaction with the oxide surface of reduced catalysts depends on the composition and structure of the oxide-precursor. Note that in [7,8], in the region of much higher dispersity of metal copper as compared to the catalysts studied in the present work, the structural sensitivity of methanol dehydrogenation to methyl formate was observed over copper-silicon catalysts, obtained by ion exchange. 4. Conclusions 1. Methanol dehydrogenation using copper-containing catalysts of various nature, namely, copper chromite with the structure of tetragonally distorted spinel, copper hydrosilicate of Chrysocolla structure, and copper-zinc hydroxysilicate of Zincsilite structure, proceeds only after activation with hydrogen, which results in the formation of metal copper of various dispersity and morphology at the oxide surface. 2. Dehydrogenation of methanol over the studied catalysts proceeds via successive reactions: (I) 2CH3 OH = CH3 OOCH + 2H2 and (II) 180 T.P. Minyukova et al. / Applied Catalysis A: General 237 (2002) 171–180 CH3 OOCH = 2CO + 2H2 . No simultaneous reaction of direct dehydrogenation of methanol to CO and H2 is observed. 3. In the region of low methanol conversion, X ≤ 0.1, where the dehydrogenation process is represented mainly by reaction (I), occurring far from its equilibrium, the methyl formate selectivity for all the catalysts studied were close to 1.0. At 0.2 ≤ X ≤ 0.55, the selectivity decrease with increasing conversion, and the catalyst activity ratio in successive reactions may serve as their comparative characteristic. At high conversion values, when reaction (I) is close to its equilibrium, the selectivity is independent of catalyst properties studied and depends only on the methanol conversion. 4. Reaction (I) shows low sensitivity to the state of metal copper in reduced catalysts and, hence, low sensitivity to the composition and structure of oxides-precursors. The catalysts’ activity in reaction (II) greatly depends on the state of metal copper in the catalysts. 5. It was assumed that the catalyst activity in methyl formate conversion to CO and H2 and, hence, the selectivity of methanol dehydrogenation with respect to methyl formate generation in the region of moderate methanol conversion depend on the strength of the interaction between metal copper particles and the catalyst oxide surface, which is determined by the composition and structure of the oxide-precursor. Acknowledgements The authors are grateful to G.N. Kustova, L.M. Plyasova, and V.I. Zaykovskii for structural and spectroscopic studies of the catalysts. References [1] L. Shlegel, D. Gutshik, A.Ya. Rozovskii, Kinet. Katal. 31 (4) (1990) 1000 (in Russian). [2] S.V. Gorshkov, G.I. Lin, A.Ya. Rozovskii, Kinet. Katal. 40 (3) (1999) 334. [3] D.R. Stull, E.F. Westrum, G.C. Sinke, The Chemical Thermodynamics of Organic Compounds, Wiley, New York, 1969, pp. 450, 643. [4] J.S. Lee, J.C. Kim, Y.G. Kim, Appl. Catal. 57 (1990) 1. [5] T. Sodesawa, React. Kinet. Catal. Lett. 24 (3/4) (1984) 259. [6] T. Sodesawa, React. Kinet. Catal. Lett. 32 (1) (1986) 63. [7] T. Sodesawa, M. Nagacho, A. Onodera, F. Nozaki, J. Catal. 102 (1986) 460. [8] E.D. Guerreiro, O.F. Gorriz, G. Larsen, L.A. Arrua, Appl. Catal. A: Gen. 204 (2000) 33. [9] A.L. Lapidus, S.N. Antonyuk, V.D. Kapkin, I.A. Bruk, S.D. Sominskii, N.S. Nechuro, Neftechim 25 (1985) 761 (in Russian). [10] Y. Wang, R. Gang, S. Han, React. Kinet. Catal. Lett. 67 (2) (1999) 305. [11] S. Sato, M. Iijima, T. Nakayama, T. Sodesawa, F. Nozaki, J. Catal. 169 (1997) 447. [12] L.M. Plyasova, L.P. Solovyeva, T.A. Kriger, O.V. Makarova, T.M. Yurieva, Kinet. Katal. 37 (4) (1996) 584 (English). [13] T.M. Yurieva, L.M. Plyasova, O.V. Makarova, T.A. Krieger, J. Mol. Catal. A: Chem. 113 (1996) 427. [14] T.M. Yurieva, G.N. Kustova, T.P. Minyukova, E.K. Poels, A. Bliek, M.P. Demeshkina, L.M. Plyasova, T.A. Krieger, V.I. Zaikovskii, Mat. Res. Innov. 5 (1) (2001) 3. [15] T.M. Yurieva, T.P. Minyukova, G.N. Kustova, L.M. Plyasova, T.A. Krieger, M.P. Demeshkina, V.I. Zaikovskii, V.V. Malakhov, L.S. Dovlitova, Mat. Res. Innov. 5 (2) (2001) 74– 80. [16] C.J.G. Van Der Grift, P.A. Elberse, A. Mulder, J.W. Geus, Appl. Catal. 59 (1990) 275. [17] J.J.F. Scholten, J.A. Konvalinka, Trans. Faraday Soc. 65 (1969) 2465. [18] S. Sato, R. Takahashi, T. Sodesawa, K. Yuma, Y. Obata, J. Catal. 196 (2000) 195. [19] J.W. Evans, P.S. Casey, M.S. Wainwright, D.L. Trimm, N.W. Cant, Appl. Catal. 7 (1983) 75. [20] Minerals, Issue 2, Vol. IV, The Layered Silicates, Nauka, Moscow, 1992 (in Russian) 101. [21] V.M. Ust’yantsev, V.P. Mar’evich, Izv. Akad. Nauk Neorg. Mater. 9 (2) (1973) 336 (in Russian). [22] E.E. Rachkovskii, G.K. Boreskov, Kinet. Katal. 9 (4) (1970) 1249 (in Russian). [23] M. Boudart, in: Proceedings of the Boreskov Conference on Catalysis on the Eve of the XXI Century: Science and Engineering, Memorial G.K., Abstracts, Novosibirsk, 1997, Part 1, p. 5.