- Email: [email protected]
Catalytic ceramic membrane in a three-phase reactor for the competitive hydrogenation–isomerisation of methylenecyclohexane
Separation and Purification Technology 34 (2004) 239–245
Catalytic ceramic membrane in a three-phase reactor for the competitive hydrogenation–isomerisation of methylenecyclohexane Aldo Bottino a , Gustavo Capannelli a,∗ , Antonio Comite a , Adriana Del Borghi b , Renzo Di Felice b a
Dipartimento di Chimica e Chimica Industriale, Università degli Studi di Genova, Via Dodecaneso 31, 16146 Genova, Italy b Dipartimento di Ingegneria Chimica e di Processo “G.B. Bonino”, Università degli Studi di Genova, Via all’Opera Pia 15, 16145 Genova, Italy
Abstract Hydrogenation and oxidation reactions can be carried out in very mild conditions using a three-phase catalytic reactor. The challenge is to overcome the diffusion resistance that affects this type of reactor. The catalytic membrane reactor (CMR), where the membrane is used not only as physical selective barrier but also as a chemical reactor, can be an efficient alternative to more conventional systems in improving the contact among solid catalyst, gas and liquid. The performances of different catalytic membranes were explored in the hydrogenation–isomerisation of methylenecyclohexane, in a temperature range between 288 and 343 K. Comparisons between a classic batch stirred tank and CMR were also carried out. Various characteristics of the reacting system, such as the overall process rate, the effect of temperature, the reaction order with reference to the substrate and the hydrogen and reaction selectivity were studied. © 2003 Elsevier B.V. All rights reserved. Keywords: Three-phase catalytic membrane reactor; Membrane; Hydrogenation; Isomerisation; Multiphase reactors
1. Introduction Membranes are becoming an increasingly popular alternative to traditional systems in the separation of multicomponent mixtures, either in the gas or in liquid phase. The principal characteristic of a membrane is its structure (e.g. a porous structure), which selectively allows components to pass from one side to the other. Typical examples are the separation of gas mixtures, ∗ Corresponding author. Tel.: +39-10-3536197; fax: +39-10-3536199. E-mail address: [email protected] (G. Capannelli).
where the smaller gas molecules such as hydrogen are separated from larger components, or the production of freshwater from the seawater, where the membrane in this case let the solvent molecules through while retaining most of the solute. This specific membrane behaviour has another obvious application in the field of reaction engineering. The idea here is very simple and one that chemical engineers are still researching: a single unit which would carry out reaction and product separation at the same time. Gas-phase reactions, where the physical barrier is used to keep the reactant and products separated as much as possible, are a possible exploitation of this
1383-5866/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S1383-5866(03)00196-5
240
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
membrane characteristic [1]. In spite of their great potential, the industrial application of membrane reactors is not so common although their use is certainly on the increase. There is, finally, the most recent application of membrane reactors: the three-phase catalytic membrane reactor (CMR), where a gas and a liquid stream are brought in intimate contact using the membrane with the catalyst deposited on its surface. As in previous cases, the membrane can keep the different streams separate (even though gas and liquid are immiscible). When the system is compared to classical arrangements, such as slurry or three-phase fluidised bed reactors (where the catalyst is suspended in the fluid), other great potential advantages are revealed. The first advantage is related to catalyst deposition on the membrane active layer, which eliminates any catalyst loss due to elutriation from the fluid flow in the reactor. The second advantage concerns the path followed by the reactant species to reach the active catalysts [2]. We will show here that resistance to mass transfer is greatly reduced in the case of membrane reactors when compared to suspended solid reactors, with obvious benefits to the overall process. Thus, the main aim of this work is to experimentally verify the advantages of using membrane reactors in three-phase reacting systems. We chose methylenecyclohexane hydrogenation as a case study since it gives only one isomer: 1-methylcyclohexene. The choice is not linked to the industrial importance of these products, but rather to the analytical simplicity of the substrate, its similarity with fine chemical products and to the possibility of giving two parallel reactions characterised by a different dependence on hydrogen concentration. Commercial ceramic membranes with the ␥-Al2 O3 or TiO2 selective layer were used as a support for platinum (Pt), palladium (Pd) or ruthenium (Ru) catalysts. This process can proceed either by direct hydrogenation or by isomerisation to methylcyclohexene and subsequent hydrogenation to methylcyclohexane. Kinetic parameters (partial reaction order with respect to hydrogen, apparent activation energy) along with the effect of operating conditions were determined and the performance of the CMR was compared with that of a slurry reactor, for the same catalyst and the same support (␥-Al2 O3 ).
Table 1 Geometric characteristics of the membranes Geometric features
Length (mm)
Total membrane Membrane without enamelling Outer diameter Inner diameter
150 100–120 10.7 6.7
2. Experimental 2.1. Supports and preparation of the catalytic systems Commercial ceramic membranes (SCT-US Filter and Schumacher) were used as supports for the deposition of the catalyst. These membranes were tubes of macroporous ␣-Al2 O3 internally coated with a thin mesoporous layer of ␥-Al2 O3 or TiO2 . Table 1 reports the geometrical characteristics of the commercial membranes used. Both ends of these ceramic tubes were enamelled in order to keep the inside separate from their outside during the execution of the activity tests. The ␥-Al2 O3 powders used as support for the catalyst were supplied by SCT-US Filter and were obtained with the same preparation procedure of the ␥-Al2 O3 that constitutes the selective layer of the employed membranes. In Table 2, the catalyst deposition conditions are reported. Platinum and palladium were deposited onto the supports (membranes and powders) by the ion exchange technique. 40 ml of a diluted precursor (PdCl2 or H2 PtCl6 ) solution was kept in contact with the support for 24 h after adjusting pH with HCl to about 4. Both powders (after filtration) and membranes were washed with deionised water in order to remove the unbonded precursor. For the deposition of ruthenium, an impregnation procedure without pH correction and final washings was adopted. In each case after the catalyst deposition, the supports (membranes or powders) were dried at 100 ◦ C overnight and reduced with H2 at 450 ◦ C for 2 h. 2.2. Characterisation techniques The catalytic membranes were morphologically characterised by a scanning electron microscope (SEM, Leo Steroescan 440) equipped with an EDX
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
241
Table 2 Catalytic systems prepared along with their preparative conditions Acronym
Type/support
Deposition technique
Catalyst precursor
Concentration (g/l)
Pd/A Pt/A Ru/A Pd/T Pd/␥-Al2 O3 Pt/␥-Al2 O3 Ru/␥-Al2 O3
Membrane/␥-Al2 O3 Membrane/␥-Al2 O3 Membrane/␥-Al2 O3 Membrane/TiO2 Powder/␥-Al2 O3 Powder/␥-Al2 O3 Powder/␥-Al2 O3
Ion exchange Ion exchange Impregnation Ion exchange Ion exchange Ion exchange Impregnation
PdCl2 + HCl H2 PtCl6 +HCl RuCl3 PdCl2 +HCl PdCl2 + HCl H2 PtCl6 + HCl RuCl3
0.67 0.1 1 0.67 0.67 0.1 1
for microanalysis. N2 adsorption/desorption at 77 K measurements were made with a Micromeritics ASAP 2010 instrument determining the specific surface area (BET method) and the pore dimension distribution (BJH method). Catalyst loading was determined by inductively coupled plasma (ICP, Varian Vista Pro) after the catalyst samples were chemically attacked with aqua regia. Water permeability and gas permeance measurements were carried out, respectively, by forcing microfiltered water or pure gases (H2 , He or N2 ) through the membrane at room temperature and measuring the pressure inside and outside the membrane. 2.3. Catalytic runs Fig. 1 shows the scheme of the pilot plant used to evaluate catalytic activity of the prepared membranes. The reactor, R, was made of stainless steel
Fig. 1. Three-phase CMR. Schematic diagram of the pilot plant: (g) valves (M1, M2) mass flow controller, (F) bubble flow meter, (N) condenser, (C) liquid sampling, (V) volumetric pump and (S) tank.
and the catalytic membrane was assembled inside it by means of vyton o-rings in order to keep the internal side of the membrane separate from the external side. In the feeding configuration employed gas flows on the inside of the membrane and liquid on the outside [3]. Gases (hydrogen and/or an inert gas such as helium) fed to the reactor were regulated by means of mass flow meters (Brooks instruments). The flow of gas going out from the reactor was measured using a bubble flow meter. The liquid phase (ethanol with methylenecyclohexane and the products that are formed during the runtime) was recirculated from the tank, S, to the reactor with a flow of 15 ml/min. The absolute pressure of the liquid phase was set at 1.5 bar while that of the gas line was fixed at 2 bar. The total liquid volume in the plant was 60 ml. The catalytic activity of the powders was also evaluated in a slurry reactor [4]. Catalytic powders were stirred in 60 ml of a solution of methylenecyclohexane (Co = 0.055 mol/l) in ethanol in which H2 was bubbled. In both cases, the composition of the liquid phase was checked over time at a constant temperature (the range investigated lay between 288 and 343 K) by analysing a very small amount of the reaction solution with gas chromatography (FID, packed column SE 30 10% on Chromosorb W). Before each run, the catalytic powder or membrane was activated by flowing H2 at 50 ◦ C for 2 h in the reactor. The hydrogenation of methylenecyclohexane follows the pathway illustrated in Fig. 2. The process involves consecutive and parallel reactions. The initial reactant, methylenecyclohexane, can be either hydrogenated to methylcyclohexane or isomerised to methylcyclohexene, which can subsequently be hydrogenated to methylcyclohexane.
242
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
Fig. 2. Reaction scheme. Fig. 3. SEM micrograph of the Pt/A membrane cross-section.
3. Results and discussion Table 4 Transport properties of the membranes used at 23 ◦ C
3.1. Characterisations The porous structure of these membranes has been already shown in a previous work [6]. These membranes have an asymmetric porous structure with a pore diameter decreasing from the outer to the inner diameter. Schematically, the membrane cross-section can be divided into two main regions: an ␣-Al2 O3 macroporous support (with a thickness of about 2 mm) and a mesoporous thin top layer (Fig. 3). From these morphological observations, the thickness of the top layer was measured (as reported in Table 3). The Pd/T membrane has the thinnest top layer and some defects on its surface were also encountered (cracks with a length of about 2 m). From N2 adsorption measurements, the pore size distribution was calculated. It agrees with the supplier’s specification: the nominal pore diameter of ␥-Al2 O3 and TiO2 membrane layers was about 5 and 10 nm, respectively. EDX measurements revealed that the catalyst in the membranes was located in the top layer and thus the
System
Pd/A and Ru/A Pt/A Pd/T
Water permeability (m3 m−2 s−1 bar−1 )
0.33 1.8 3.48
Gas permeance (×107 mol s−1 m−2 Pa−1 ) H2
He
N2
0.25 3.8 –
0.14 2.40 4.53
0.072 0.90 2.48
catalyst percentage referring to the top layer alone was assessed. The amount of catalyst was between 0.5 and 1%, similar to the values found for the catalytic powders. Table 4 reports the permeability of gases and water of the membranes. Pd/A and Ru/A membranes show lower permeability values due to their thicker top layers, while the Pd/T membrane gives a higher permeability that is clearly related to the presence of the surface defectivity cited above. Moreover, in this membrane, gas permeance dependence on the pressure (an index of the Poiseuille contribute to the permeating flow) was more marked.
Table 3 Summary of the characterisation results of the catalytic membranes Acronym
Layer
Thickness (m)
Specific surface areaa (m2 /g)
Pore diameter (nm)
Catalyst amount (mg)
Catalysta (% wt./wt.)
Pd/A Pt/A Ru/A Pd/T
␥-Al2 O3 ␥-Al2 O3 ␥-Al2 O3 TiO2
2.4 1.8 2.4 0.7
263 253 260 200
4.8 4.8 4.8 10
0.15 0.19 0.11 0.71
0.8 1.13 0.5 6
a
Referred to the amount of the mesoporous layer alone.
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
243
Fig. 4. Behaviour of composition with time for (a) Pt/␥-Al2 O3 and (b) Pd/␥-Al2 O3 systems: (䊐) MLENE, (䉱) MNANO, (䉬) MLANO.
3.2. Catalytic runs Two typical runs are depicted in Fig. 4. In Fig. 4(a) (platinum catalyst), the reactant concentration decreases as the reaction time increases and only one product, MLENE, is formed, with the isomer concentration being quite negligible. On the other hand, in Fig. 4(b), relative to palladium catalyst, both reaction products are obtained. From our experimental runs, not described here for the sake of brevity, we found that Pt or Ru should be used as catalysts if methylcyclohexane is the target product. In fact, in this case, only the hydrogenation of methylenecyclohexane occurred in spite of the presence of ␥-Al2 O3 (␥-Al2 O3 acid properties favour isomerisation). Indeed, Pd promotes the migration of the double bond, forming the isomer compound, MLENE. Pd behaviour seems to be related to its ability to dissolve H2 with formation of a stable hydride [5]. From Fig. 4, reaction rates can be easily obtained from the slope of the measured concentration of the reactant or of the product. Due to the large experimental error towards the end of reaction, it was decided to use only the initial reaction rates. To have a qualitative picture of the behaviour of the membrane when used as a three-phase catalytic reactor, a series of experiments was carried out and compared with a more classical slurry reactor working under equivalent conditions (temperature, pressure, reactant concentration, etc.). The first series of comparative experimental runs investigated the effect of the temperature on the over-
all process rate, measured here as the rate of disappearance of methylenecyclohexane per unit catalyst weight. It is well known that when the logarithm of the overall process rate is plotted as a function of the inverse of temperature (with all other parameters kept at a fixed value), the slope yields the value of the apparent activation energy and from the magnitude of this parameter important deductions about the prevailing regime can be made. Fig. 5 depicts the measured reaction rate when the Pt catalyst was used in a temperature range 288–343 K. It can be seen that at lower temperatures, the reaction rates are very similar for the two systems, suggesting that the same conditions
Fig. 5. Arrhenius plot of the system Pt/␥-Al2 O3 in a (䊉) CMR and in a (䉱) slurry reactor.
244
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
Fig. 6. Diffusive and chemical steps occurring during hydrogenation reactions in a CMR: (1) H2 dissolution; (2) diffusion in liquid; (3) adsorption and chemical reaction; (4) desorption and back-diffusion in liquid.
could be governing the phenomenon. However, as the temperature increases above 315 K, the reaction rates for the slurry reactor tend to level off, whereas the membrane reactor keeps following the same trend. The analysis of this behaviour is rather straightforward: the slurry reactor undergoes an evident change of reaction regime, moving from a kinetic-controlled regime to diffusion-controlled regime whereas the membrane reactor displays a kinetic-controlled regime in the whole temperature range of temperature investigated, with the calculated activation energy in the region of 60 kJ/mol. This agrees with typical published values for the hydrogenation reaction [7]. This very interesting behaviour can be qualitatively justified if we consider the model of the membrane reactor, as depicted in Fig. 6, and compare it with the case of catalysts supported on a spherical porous particle. In the slurry reactor, the porous catalytic particle is completely enveloped in the liquid. Therefore, the gaseous H2 has to dissolve in the liquid phase, diffuse in the liquid film resistance around the particle and in the porous solid structure before reaching the catalytic site. Diffusion-controlled regime is consequently expected. Since liquid is present only on one side of the very thin catalytic layer in a CMR, it has a very short path before reaching the catalysts. Thus, diffusion is less likely to be the controlling factor of the process. An indirect confirmation of the above suggested mechanism has been obtained by comparing product selectivity for the two reactor arrangements with the palladium catalysts: it has been found that the ratio (hydrogenation rate/isomerisation rate) is about 0.25
for the membrane reactor whereas the same ratio drops to less than 0.001 for the slurry reactor. Obviously, in the second case hydrogen is much less available compared to the membrane arrangement, confirming the advantage of the CMR when good fluid-catalyst contact is sought. The dependence of the hydrogenation reaction rate on the reactant concentration was investigated. A zero apparent reaction order with reference to MLANO was found. In order to evaluate the influence of H2 on CMR, some runs were carried out by gradually increasing the H2 partial pressure in a He stream with gas total pressure at about 2 bar. For higher H2 partial pressure, no carrier He stream was used. It was found (Fig. 7) that the rate of MLANO formation proportionally increases with the H2 partial pressure until a value of partial pressure of about 0.8 bar with a partial reaction
Fig. 7. Pd/T system: specific rate of MLANO formation as a function of the H2 partial pressure. Symbols same as in Fig. 4.
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
245
4. Conclusions
Fig. 8. Pd/T system: composition vs. time. He was fed at interval B, while H2 was fed at intervals A and C.
order of 0.76 was reached. Over a partial pressure of 0.8 bar, the MLANO formation rate becomes independent of H2 , suggesting that H2 is always available on catalytic sites. In both cases, it seems that the transfer of H2 from the gas phase to the catalytic site is not rate limited. Finally, the presence of H2 and its influence in forming the isomer compound, MLENE, was tested by carrying out runs without feeding H2 or substituting H2 with He during the course of the reaction. The evidence is that without an initial supply of H2 neither the hydrogenation nor isomerisation starts. Hence, the presence of H2 is necessary also for the startup of the isomerisation reaction. Nevertheless, as can be seen from Fig. 8, an interruption of the H2 feeding slows down the rate of formation of the hydrogenated product, MLANO, and has no effect on the rate of formation of the isomer compound, MLENE. Further analysis of this behaviour is in progress and will be reported in due time.
A CMR applied in a three-phase process allows for better control of the operative conditions and to exploit a wide range of process temperatures, since the overall process rate is controlled by a kinetic regime. The advantage over a classic slurry reactor is the possibility of carrying out hydrogenation while avoiding the use of higher H2 pressure to overcome its mass-transfer limitations. As the range of operating conditions in which the reactor works in a kinetic regime is expanded, higher overall reaction rates can be obtained by increasing the temperature. Moreover, reaction selectivity can also be easily controlled.
References [1] J. Zaman, A. Chakma, Inorganic membrane reactors, J. Membr. Sci. 92 (1994) 1. [2] R. Dittmeyer, V. Hollein, K. Daub, Membrane reactors for hydrogenation and dehydrogenation processes based on supported palladium, J. Mol. Catal. A 173 (2001) 135. [3] O. Monticelli, A. Bezzi, A. Bottino, G. Capannelli, A. Servida, Hydrogenation of cinnamaldehyde: the use of three-phase catalytic membrane reactors, in: Proceedings of the Fourth Workshop on Optimisation of Catalytic Membrane Reactor Systems, Oslo, Norway, 1997, pp. 125–134. [4] G. Biardi, G. Baldi, Three-phase catalytic reactors, Catal. Today 52 (1999) 223. [5] P. Rylander, Catalytic Hydrogenation in Organic Syntheses, Academic Press, London, 1979. [6] G. Vitulli, A. Verrazzani, E. Pitsali, P. Salvadori, G. Capannelli, G. Martra, Pt/␥-Al2 O3 catalytic membranes vs. Pt on ␥-Al2 O3 powders in the selective hydrogenation of p-chloronitrobenzene, Catal. Lett. 44 (1997) 205. [7] J.M. Winterbottom, Z. Khan, A.P. Boyes, S. Raymahasay, Catalytic hydrogenation in a packed bed bubble column reactor, Catal. Today 48 (1999) 221.
Catalytic ceramic membrane in a three-phase reactor for the competitive hydrogenation–isomerisation of methylenecyclohexane Aldo Bottino a , Gustavo Capannelli a,∗ , Antonio Comite a , Adriana Del Borghi b , Renzo Di Felice b a
Dipartimento di Chimica e Chimica Industriale, Università degli Studi di Genova, Via Dodecaneso 31, 16146 Genova, Italy b Dipartimento di Ingegneria Chimica e di Processo “G.B. Bonino”, Università degli Studi di Genova, Via all’Opera Pia 15, 16145 Genova, Italy
Abstract Hydrogenation and oxidation reactions can be carried out in very mild conditions using a three-phase catalytic reactor. The challenge is to overcome the diffusion resistance that affects this type of reactor. The catalytic membrane reactor (CMR), where the membrane is used not only as physical selective barrier but also as a chemical reactor, can be an efficient alternative to more conventional systems in improving the contact among solid catalyst, gas and liquid. The performances of different catalytic membranes were explored in the hydrogenation–isomerisation of methylenecyclohexane, in a temperature range between 288 and 343 K. Comparisons between a classic batch stirred tank and CMR were also carried out. Various characteristics of the reacting system, such as the overall process rate, the effect of temperature, the reaction order with reference to the substrate and the hydrogen and reaction selectivity were studied. © 2003 Elsevier B.V. All rights reserved. Keywords: Three-phase catalytic membrane reactor; Membrane; Hydrogenation; Isomerisation; Multiphase reactors
1. Introduction Membranes are becoming an increasingly popular alternative to traditional systems in the separation of multicomponent mixtures, either in the gas or in liquid phase. The principal characteristic of a membrane is its structure (e.g. a porous structure), which selectively allows components to pass from one side to the other. Typical examples are the separation of gas mixtures, ∗ Corresponding author. Tel.: +39-10-3536197; fax: +39-10-3536199. E-mail address: [email protected] (G. Capannelli).
where the smaller gas molecules such as hydrogen are separated from larger components, or the production of freshwater from the seawater, where the membrane in this case let the solvent molecules through while retaining most of the solute. This specific membrane behaviour has another obvious application in the field of reaction engineering. The idea here is very simple and one that chemical engineers are still researching: a single unit which would carry out reaction and product separation at the same time. Gas-phase reactions, where the physical barrier is used to keep the reactant and products separated as much as possible, are a possible exploitation of this
1383-5866/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S1383-5866(03)00196-5
240
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
membrane characteristic [1]. In spite of their great potential, the industrial application of membrane reactors is not so common although their use is certainly on the increase. There is, finally, the most recent application of membrane reactors: the three-phase catalytic membrane reactor (CMR), where a gas and a liquid stream are brought in intimate contact using the membrane with the catalyst deposited on its surface. As in previous cases, the membrane can keep the different streams separate (even though gas and liquid are immiscible). When the system is compared to classical arrangements, such as slurry or three-phase fluidised bed reactors (where the catalyst is suspended in the fluid), other great potential advantages are revealed. The first advantage is related to catalyst deposition on the membrane active layer, which eliminates any catalyst loss due to elutriation from the fluid flow in the reactor. The second advantage concerns the path followed by the reactant species to reach the active catalysts [2]. We will show here that resistance to mass transfer is greatly reduced in the case of membrane reactors when compared to suspended solid reactors, with obvious benefits to the overall process. Thus, the main aim of this work is to experimentally verify the advantages of using membrane reactors in three-phase reacting systems. We chose methylenecyclohexane hydrogenation as a case study since it gives only one isomer: 1-methylcyclohexene. The choice is not linked to the industrial importance of these products, but rather to the analytical simplicity of the substrate, its similarity with fine chemical products and to the possibility of giving two parallel reactions characterised by a different dependence on hydrogen concentration. Commercial ceramic membranes with the ␥-Al2 O3 or TiO2 selective layer were used as a support for platinum (Pt), palladium (Pd) or ruthenium (Ru) catalysts. This process can proceed either by direct hydrogenation or by isomerisation to methylcyclohexene and subsequent hydrogenation to methylcyclohexane. Kinetic parameters (partial reaction order with respect to hydrogen, apparent activation energy) along with the effect of operating conditions were determined and the performance of the CMR was compared with that of a slurry reactor, for the same catalyst and the same support (␥-Al2 O3 ).
Table 1 Geometric characteristics of the membranes Geometric features
Length (mm)
Total membrane Membrane without enamelling Outer diameter Inner diameter
150 100–120 10.7 6.7
2. Experimental 2.1. Supports and preparation of the catalytic systems Commercial ceramic membranes (SCT-US Filter and Schumacher) were used as supports for the deposition of the catalyst. These membranes were tubes of macroporous ␣-Al2 O3 internally coated with a thin mesoporous layer of ␥-Al2 O3 or TiO2 . Table 1 reports the geometrical characteristics of the commercial membranes used. Both ends of these ceramic tubes were enamelled in order to keep the inside separate from their outside during the execution of the activity tests. The ␥-Al2 O3 powders used as support for the catalyst were supplied by SCT-US Filter and were obtained with the same preparation procedure of the ␥-Al2 O3 that constitutes the selective layer of the employed membranes. In Table 2, the catalyst deposition conditions are reported. Platinum and palladium were deposited onto the supports (membranes and powders) by the ion exchange technique. 40 ml of a diluted precursor (PdCl2 or H2 PtCl6 ) solution was kept in contact with the support for 24 h after adjusting pH with HCl to about 4. Both powders (after filtration) and membranes were washed with deionised water in order to remove the unbonded precursor. For the deposition of ruthenium, an impregnation procedure without pH correction and final washings was adopted. In each case after the catalyst deposition, the supports (membranes or powders) were dried at 100 ◦ C overnight and reduced with H2 at 450 ◦ C for 2 h. 2.2. Characterisation techniques The catalytic membranes were morphologically characterised by a scanning electron microscope (SEM, Leo Steroescan 440) equipped with an EDX
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
241
Table 2 Catalytic systems prepared along with their preparative conditions Acronym
Type/support
Deposition technique
Catalyst precursor
Concentration (g/l)
Pd/A Pt/A Ru/A Pd/T Pd/␥-Al2 O3 Pt/␥-Al2 O3 Ru/␥-Al2 O3
Membrane/␥-Al2 O3 Membrane/␥-Al2 O3 Membrane/␥-Al2 O3 Membrane/TiO2 Powder/␥-Al2 O3 Powder/␥-Al2 O3 Powder/␥-Al2 O3
Ion exchange Ion exchange Impregnation Ion exchange Ion exchange Ion exchange Impregnation
PdCl2 + HCl H2 PtCl6 +HCl RuCl3 PdCl2 +HCl PdCl2 + HCl H2 PtCl6 + HCl RuCl3
0.67 0.1 1 0.67 0.67 0.1 1
for microanalysis. N2 adsorption/desorption at 77 K measurements were made with a Micromeritics ASAP 2010 instrument determining the specific surface area (BET method) and the pore dimension distribution (BJH method). Catalyst loading was determined by inductively coupled plasma (ICP, Varian Vista Pro) after the catalyst samples were chemically attacked with aqua regia. Water permeability and gas permeance measurements were carried out, respectively, by forcing microfiltered water or pure gases (H2 , He or N2 ) through the membrane at room temperature and measuring the pressure inside and outside the membrane. 2.3. Catalytic runs Fig. 1 shows the scheme of the pilot plant used to evaluate catalytic activity of the prepared membranes. The reactor, R, was made of stainless steel
Fig. 1. Three-phase CMR. Schematic diagram of the pilot plant: (g) valves (M1, M2) mass flow controller, (F) bubble flow meter, (N) condenser, (C) liquid sampling, (V) volumetric pump and (S) tank.
and the catalytic membrane was assembled inside it by means of vyton o-rings in order to keep the internal side of the membrane separate from the external side. In the feeding configuration employed gas flows on the inside of the membrane and liquid on the outside [3]. Gases (hydrogen and/or an inert gas such as helium) fed to the reactor were regulated by means of mass flow meters (Brooks instruments). The flow of gas going out from the reactor was measured using a bubble flow meter. The liquid phase (ethanol with methylenecyclohexane and the products that are formed during the runtime) was recirculated from the tank, S, to the reactor with a flow of 15 ml/min. The absolute pressure of the liquid phase was set at 1.5 bar while that of the gas line was fixed at 2 bar. The total liquid volume in the plant was 60 ml. The catalytic activity of the powders was also evaluated in a slurry reactor [4]. Catalytic powders were stirred in 60 ml of a solution of methylenecyclohexane (Co = 0.055 mol/l) in ethanol in which H2 was bubbled. In both cases, the composition of the liquid phase was checked over time at a constant temperature (the range investigated lay between 288 and 343 K) by analysing a very small amount of the reaction solution with gas chromatography (FID, packed column SE 30 10% on Chromosorb W). Before each run, the catalytic powder or membrane was activated by flowing H2 at 50 ◦ C for 2 h in the reactor. The hydrogenation of methylenecyclohexane follows the pathway illustrated in Fig. 2. The process involves consecutive and parallel reactions. The initial reactant, methylenecyclohexane, can be either hydrogenated to methylcyclohexane or isomerised to methylcyclohexene, which can subsequently be hydrogenated to methylcyclohexane.
242
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
Fig. 2. Reaction scheme. Fig. 3. SEM micrograph of the Pt/A membrane cross-section.
3. Results and discussion Table 4 Transport properties of the membranes used at 23 ◦ C
3.1. Characterisations The porous structure of these membranes has been already shown in a previous work [6]. These membranes have an asymmetric porous structure with a pore diameter decreasing from the outer to the inner diameter. Schematically, the membrane cross-section can be divided into two main regions: an ␣-Al2 O3 macroporous support (with a thickness of about 2 mm) and a mesoporous thin top layer (Fig. 3). From these morphological observations, the thickness of the top layer was measured (as reported in Table 3). The Pd/T membrane has the thinnest top layer and some defects on its surface were also encountered (cracks with a length of about 2 m). From N2 adsorption measurements, the pore size distribution was calculated. It agrees with the supplier’s specification: the nominal pore diameter of ␥-Al2 O3 and TiO2 membrane layers was about 5 and 10 nm, respectively. EDX measurements revealed that the catalyst in the membranes was located in the top layer and thus the
System
Pd/A and Ru/A Pt/A Pd/T
Water permeability (m3 m−2 s−1 bar−1 )
0.33 1.8 3.48
Gas permeance (×107 mol s−1 m−2 Pa−1 ) H2
He
N2
0.25 3.8 –
0.14 2.40 4.53
0.072 0.90 2.48
catalyst percentage referring to the top layer alone was assessed. The amount of catalyst was between 0.5 and 1%, similar to the values found for the catalytic powders. Table 4 reports the permeability of gases and water of the membranes. Pd/A and Ru/A membranes show lower permeability values due to their thicker top layers, while the Pd/T membrane gives a higher permeability that is clearly related to the presence of the surface defectivity cited above. Moreover, in this membrane, gas permeance dependence on the pressure (an index of the Poiseuille contribute to the permeating flow) was more marked.
Table 3 Summary of the characterisation results of the catalytic membranes Acronym
Layer
Thickness (m)
Specific surface areaa (m2 /g)
Pore diameter (nm)
Catalyst amount (mg)
Catalysta (% wt./wt.)
Pd/A Pt/A Ru/A Pd/T
␥-Al2 O3 ␥-Al2 O3 ␥-Al2 O3 TiO2
2.4 1.8 2.4 0.7
263 253 260 200
4.8 4.8 4.8 10
0.15 0.19 0.11 0.71
0.8 1.13 0.5 6
a
Referred to the amount of the mesoporous layer alone.
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
243
Fig. 4. Behaviour of composition with time for (a) Pt/␥-Al2 O3 and (b) Pd/␥-Al2 O3 systems: (䊐) MLENE, (䉱) MNANO, (䉬) MLANO.
3.2. Catalytic runs Two typical runs are depicted in Fig. 4. In Fig. 4(a) (platinum catalyst), the reactant concentration decreases as the reaction time increases and only one product, MLENE, is formed, with the isomer concentration being quite negligible. On the other hand, in Fig. 4(b), relative to palladium catalyst, both reaction products are obtained. From our experimental runs, not described here for the sake of brevity, we found that Pt or Ru should be used as catalysts if methylcyclohexane is the target product. In fact, in this case, only the hydrogenation of methylenecyclohexane occurred in spite of the presence of ␥-Al2 O3 (␥-Al2 O3 acid properties favour isomerisation). Indeed, Pd promotes the migration of the double bond, forming the isomer compound, MLENE. Pd behaviour seems to be related to its ability to dissolve H2 with formation of a stable hydride [5]. From Fig. 4, reaction rates can be easily obtained from the slope of the measured concentration of the reactant or of the product. Due to the large experimental error towards the end of reaction, it was decided to use only the initial reaction rates. To have a qualitative picture of the behaviour of the membrane when used as a three-phase catalytic reactor, a series of experiments was carried out and compared with a more classical slurry reactor working under equivalent conditions (temperature, pressure, reactant concentration, etc.). The first series of comparative experimental runs investigated the effect of the temperature on the over-
all process rate, measured here as the rate of disappearance of methylenecyclohexane per unit catalyst weight. It is well known that when the logarithm of the overall process rate is plotted as a function of the inverse of temperature (with all other parameters kept at a fixed value), the slope yields the value of the apparent activation energy and from the magnitude of this parameter important deductions about the prevailing regime can be made. Fig. 5 depicts the measured reaction rate when the Pt catalyst was used in a temperature range 288–343 K. It can be seen that at lower temperatures, the reaction rates are very similar for the two systems, suggesting that the same conditions
Fig. 5. Arrhenius plot of the system Pt/␥-Al2 O3 in a (䊉) CMR and in a (䉱) slurry reactor.
244
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
Fig. 6. Diffusive and chemical steps occurring during hydrogenation reactions in a CMR: (1) H2 dissolution; (2) diffusion in liquid; (3) adsorption and chemical reaction; (4) desorption and back-diffusion in liquid.
could be governing the phenomenon. However, as the temperature increases above 315 K, the reaction rates for the slurry reactor tend to level off, whereas the membrane reactor keeps following the same trend. The analysis of this behaviour is rather straightforward: the slurry reactor undergoes an evident change of reaction regime, moving from a kinetic-controlled regime to diffusion-controlled regime whereas the membrane reactor displays a kinetic-controlled regime in the whole temperature range of temperature investigated, with the calculated activation energy in the region of 60 kJ/mol. This agrees with typical published values for the hydrogenation reaction [7]. This very interesting behaviour can be qualitatively justified if we consider the model of the membrane reactor, as depicted in Fig. 6, and compare it with the case of catalysts supported on a spherical porous particle. In the slurry reactor, the porous catalytic particle is completely enveloped in the liquid. Therefore, the gaseous H2 has to dissolve in the liquid phase, diffuse in the liquid film resistance around the particle and in the porous solid structure before reaching the catalytic site. Diffusion-controlled regime is consequently expected. Since liquid is present only on one side of the very thin catalytic layer in a CMR, it has a very short path before reaching the catalysts. Thus, diffusion is less likely to be the controlling factor of the process. An indirect confirmation of the above suggested mechanism has been obtained by comparing product selectivity for the two reactor arrangements with the palladium catalysts: it has been found that the ratio (hydrogenation rate/isomerisation rate) is about 0.25
for the membrane reactor whereas the same ratio drops to less than 0.001 for the slurry reactor. Obviously, in the second case hydrogen is much less available compared to the membrane arrangement, confirming the advantage of the CMR when good fluid-catalyst contact is sought. The dependence of the hydrogenation reaction rate on the reactant concentration was investigated. A zero apparent reaction order with reference to MLANO was found. In order to evaluate the influence of H2 on CMR, some runs were carried out by gradually increasing the H2 partial pressure in a He stream with gas total pressure at about 2 bar. For higher H2 partial pressure, no carrier He stream was used. It was found (Fig. 7) that the rate of MLANO formation proportionally increases with the H2 partial pressure until a value of partial pressure of about 0.8 bar with a partial reaction
Fig. 7. Pd/T system: specific rate of MLANO formation as a function of the H2 partial pressure. Symbols same as in Fig. 4.
A. Bottino et al. / Separation and Purification Technology 34 (2004) 239–245
245
4. Conclusions
Fig. 8. Pd/T system: composition vs. time. He was fed at interval B, while H2 was fed at intervals A and C.
order of 0.76 was reached. Over a partial pressure of 0.8 bar, the MLANO formation rate becomes independent of H2 , suggesting that H2 is always available on catalytic sites. In both cases, it seems that the transfer of H2 from the gas phase to the catalytic site is not rate limited. Finally, the presence of H2 and its influence in forming the isomer compound, MLENE, was tested by carrying out runs without feeding H2 or substituting H2 with He during the course of the reaction. The evidence is that without an initial supply of H2 neither the hydrogenation nor isomerisation starts. Hence, the presence of H2 is necessary also for the startup of the isomerisation reaction. Nevertheless, as can be seen from Fig. 8, an interruption of the H2 feeding slows down the rate of formation of the hydrogenated product, MLANO, and has no effect on the rate of formation of the isomer compound, MLENE. Further analysis of this behaviour is in progress and will be reported in due time.
A CMR applied in a three-phase process allows for better control of the operative conditions and to exploit a wide range of process temperatures, since the overall process rate is controlled by a kinetic regime. The advantage over a classic slurry reactor is the possibility of carrying out hydrogenation while avoiding the use of higher H2 pressure to overcome its mass-transfer limitations. As the range of operating conditions in which the reactor works in a kinetic regime is expanded, higher overall reaction rates can be obtained by increasing the temperature. Moreover, reaction selectivity can also be easily controlled.
References [1] J. Zaman, A. Chakma, Inorganic membrane reactors, J. Membr. Sci. 92 (1994) 1. [2] R. Dittmeyer, V. Hollein, K. Daub, Membrane reactors for hydrogenation and dehydrogenation processes based on supported palladium, J. Mol. Catal. A 173 (2001) 135. [3] O. Monticelli, A. Bezzi, A. Bottino, G. Capannelli, A. Servida, Hydrogenation of cinnamaldehyde: the use of three-phase catalytic membrane reactors, in: Proceedings of the Fourth Workshop on Optimisation of Catalytic Membrane Reactor Systems, Oslo, Norway, 1997, pp. 125–134. [4] G. Biardi, G. Baldi, Three-phase catalytic reactors, Catal. Today 52 (1999) 223. [5] P. Rylander, Catalytic Hydrogenation in Organic Syntheses, Academic Press, London, 1979. [6] G. Vitulli, A. Verrazzani, E. Pitsali, P. Salvadori, G. Capannelli, G. Martra, Pt/␥-Al2 O3 catalytic membranes vs. Pt on ␥-Al2 O3 powders in the selective hydrogenation of p-chloronitrobenzene, Catal. Lett. 44 (1997) 205. [7] J.M. Winterbottom, Z. Khan, A.P. Boyes, S. Raymahasay, Catalytic hydrogenation in a packed bed bubble column reactor, Catal. Today 48 (1999) 221.