Ind. Eng. Chem. Res. 2010, 49, 12371–12378
12371
Kinetics of Methanol Transformation into Hydrocarbons on a HZSM-5 Zeolite Catalyst at High Temperature (400-550 °C) Andre´s T. Aguayo, Diana Mier,* Ana G. Gayubo, Mo´nica Gamero, and Javier Bilbao Departamento de Ingenierı´a Quı´mica, UniVersidad del Paı´s Vasco, Apartado 644, 48080 Bilbao, Spain
A kinetic model of seven lumps has been established which allows the quantification of the product distribution (oxygenates, n-butane, C2-C4 olefins, C2-C4 paraffins (without n-butane), C5-C10 fraction, methane) in the transformation of methanol into hydrocarbons at high temperature (400-550 °C) on a HZSM-5 zeolite catalyst (SiO2/Al2O3 ) 30) with high acidic strength (>120 kJ (mol of NH3)-1) and agglomerated with bentonite and alumina. The kinetic model fits well the experimental data obtained in a fixed bed reactor, from small values of space time in which the formation of hydrocarbons is incipient, to a space time of 2.4 (g of catalyst) h (mol CH2) -1 for a complete conversion of methanol. The rise in temperature increases the yield of C2-C4 olefins, so that the maximum value (∼50%) is obtained at the ceiling temperature for the hydrothermal stability of the HZSM-5 (550 °C) and space times between 0.6 and 1 (g of catalyst) h (mol CH2)-1. 1. Introduction The transformation of methanol into hydrocarbons is a well established route for the production of fuels and raw materials for synthesizing petrochemical products from alternative sources to oil.1,2 Methanol is obtained (via gasification) using wellestablished technologies,3 from natural gas, coal, and lignocellulosic biomass (in this case helping to reduce net CO2 emissions).4 While the original interest of the valorization of methanol was focused on obtaining automotive fuels by the MTG (methanol to gasoline) and MOGD (Mobil olefins to gasoline and diesel) processes on HZSM-5 zeolite catalysts,5,6 the growing demand for olefins has shifted the interest toward the MTO process (methanol to olefins), which has also been originally proposed on a HZSM-5 zeolite,7 and currently contributes to 10% of the world olefin production. The availability of raw material alternatives to oil in different geographical sites, the lowest energy consumption, lower CO2 emissions, and good prospects for technology development are advantages of the MTO process over other processes that contribute to the production of olefins: steam cracking (70% of the olefins produced), FCC (fluid catalytic cracking) (18%), and the dehydrogenation of paraffins (2%).8,9 The MTO process is performed on large scale, with natural gas as a source of syngas, in fluidized bed reactors, on a silicoaluminophosphate catalyst (SAPO-34) and with high olefin selectivity (>80%), but with a rapid deactivation by coke, which requires a fluidized bed regenerator interconnected to the reactor.10-13 Kinetic models have been determined for the SAPO-34 catalyst,14 and for the SAPO-18,15 which has a crystalline structure isomorphic to that of SAPO-34 with pores of 3.8 × 3.8 Å and has the advantage of a lower cost of preparation and a slower deactivation by coke, as its sites have lower acidic strength.16 The production of olefins by joint transformation of methanol and linear paraffins in the same reactor is a new route, known as the CMHC process (coupled methanol and hydrocarbons cracking),17-19 which allows the upgrading of paraffins (stream of secondary interest in refineries) and which has the advantage * To whom correspondence should be addressed. Tel: 34-94-6015501. E-mail:
[email protected].
of being an energy-neutral process, working under conditions in which the heat generated in the methanol transformation (highly exothermic reaction) is consumed in the endothermic cracking of paraffins. Furthermore, the saving of fixed assets and the performance of the two reactions in the same reactor lends interesting synergies between the reaction schemes.20 The discrimination of catalysts with different acidity and shape selectivity for the CMHC process has shown the suitability of HZSM-5 zeolite catalysts with high total acidity and sufficient site acid strength for the cracking of paraffins (>120 kJ (mol of NH3)-1), which give way to a suitable balance between conversion, selectivity of C2-C4 olefins, and deactivation by coke.21 The scaling up of the CMHC process requires a model to quantify the effect of process conditions on product distribution, and, accordingly, kinetic models for the two reactions that occur simultaneously are required for the establishment of this global model. A kinetic model has been proposed in a previous paper for the cracking of n-butane (taken as a paraffin model) in the 400-550 °C range (required for a high conversion of n-butane) on a HZSM-5 zeolite catalyst (SiO2/Al2O3 molar ratio ) 30).22 The kinetic model for the transformation of methanol in the same temperature range (400-550 °C) has been developed in this paper. The kinetic modeling of methanol transformation into hydrocarbons over HZSM-5 zeolite catalyst has been the target of numerous studies in the literature.23-30 However, most of the kinetic models proposed correspond to results below 450 °C, and, consequently, experimental research and kinetic modeling of methanol transformation at higher temperatures are needed, which are required for the cracking of paraffins in the CMHC process. Kaarsholm et al.31 have developed a kinetic model that fits well the distribution of each C2-C4 olefin at this high temperature range (400-550 °C) on a HZSM-5 zeolite catalyst modified with P in order to moderate the acid strength of the sites, which is an appropriate strategy to maximize the selectivity of olefins and mitigate coke deactivation.32,33 Furthermore, HZSM-5 zeolite has hydrothermal stability problems and dealumination is enhanced by water content in the reaction medium.34 In the CMHC process, the water in the reaction medium is only that formed by methanol dehydration, and, consequently, a limit temperature of 550 °C should be established to avoid irreversible deactivation of the catalyst.
10.1021/ie101047f 2010 American Chemical Society Published on Web 11/12/2010
12372
Ind. Eng. Chem. Res., Vol. 49, No. 24, 2010
2. Experimental Section 2.1. Reaction Equipment and Product Analysis. The runs have been carried out under normal pressure in an automated reaction equipment described in a previous paper.21 The reactor is made of 316 stainless steel with an internal diameter of 9 mm and 10 cm of effective length. It is located inside a ceramiccovered stainless steel cylindrical chamber, which is heated by an electric resistance and can operate up to 100 atm and 700 °C with a catalyst mass of up to 5 g. The bed consists of a mixture of catalyst and inert solid (carborundum with an average particle diameter of 0.16 mm) to ensure bed isothermality and attain sufficient height under low space time conditions. The temperature is controlled by a digital TTM-125 Series controller and measured by a thermocouple (K type) situated in the bed of catalyst. There are two other temperature controllers: one for the furnace chamber and the other for the transfer line between the reactor and the micro-GC. The operating variables are controlled by Bespoke software. Reaction product samples (diluted in a He stream of 17 cm3 min-1) are continuously analyzed in a micro-GC (Varian CP4900). The remaining stream of reaction products passes through a Peltier cell at 0 °C. The amount of liquid condensate is controlled by a level sensor, and the noncondensable gas flow is vented. The micro-GC (with Star Toolbar software) is provided with three analytical modules and the following columns: Porapak Q (PPQ) (10 m), where the lighter products are separated (CO2, methane, ethane, ethylene, propane, propylene, methanol, dimethyl ether, water, butanes and butenes); a molecular sieve (MS-5) (10 m), where H2, CO, O2 and N2 are separated; 5CB (CPSIL) (8 m), where the C5-C10 fraction is separated. The quantification and identification of the compounds was carried out based on calibration standards of known concentration. The balance of atoms (C, H, O) is closed in all runs above 99.5%. 2.2. Catalyst. The catalyst has been selected in a previous paper, based on the combination of different criteria (activity at zero time on stream, olefin selectivity, deactivation by coke deposition, and hydrothermal stability).21 The selected catalyst (HZ-30) has been prepared using a HZSM-5 zeolite, with SiO2/ Al2O3 ) 30, supplied by Zeolyst International in ammonium form, which has been calcined at 570 °C in order to obtain the acid form. The zeolite has been agglomerated with a binder (bentonite, Exaloid) (30 wt %) and alumina (Prolabo, calcined at 1000 °C to become inert) as inert charge (45 wt %). The catalyst particles have been obtained by wet extrusion, using a high-pressure hydraulic piston, through 0.8 mm diameter holes. The extrudates are first dried at room temperature for 24 h, and then they are sieved to a particle diameter between 0.15 and 0.3 mm. The particles are dried in an oven at 110 °C for 24 h and then calcined at 575 °C for 3 h. This temperature is reached following a ramp of 5 °C min-1. The agglomeration of the active phases does not significantly reduce acidity, but improves the accessibility of the reactants (providing the catalyst with a matrix with mesopores and macropores), which is essential for reducing deactivation by coke deposition35,36 and increasing the hydrothermal stability in the regeneration step by coke combustion.37 The catalyst allows the performance of up to ten reactionregeneration cycles, without observing irreversible deactivation either in the reaction stage or in the regeneration stage, which is performed in situ by coke combustion with air in the reactor at 550 °C.38 Table 1 sets out the physical properties and the acidity values of the catalyst. The porous structure has been determined by
Table 1. Catalyst Properties acid strength, kJ (mol of NH3)-1 total acidity, (mmol of NH3) · g-1 dp, Å SBET, m2 g-1 Vm, cm3 g-1 Vp (17 < dp(Å) < 3000), cm3 g-1 pore volume distribution (%) 500 Bro¨nsted/Lewis site ratio at 150 °C
120 0.23 102 220 0.044 0.69 2.96/46.5/50.5 1.50
N2 adsorption-desorption (Micromeritics ASAP 2010) and Hg porosimetry (Micromeritics Autopore 9220). The micropore volume corresponds to the active phase, whereas the volume of meso- and macropores corresponds to the matrix of the catalyst (bentonite and alumina). The total acidity and acid strength of the catalysts have been determined by monitoring the adsorption-desorption of NH3, by combining the techniques of thermogravimetric analysisdifferential scanning calorimetry and temperature-programmed desorption using a Setaram TG-DSC calorimeter connected online with a Thermostar mass spectrometer from Balzers Instruments.39,40 The Bro¨nsted/Lewis (B/L) acid site ratio (Table 1) has been determined by analyzing the region of 1400-1700 cm-1 in the FTIR spectrum of adsorbed pyridine, which has been obtained using a Specac catalytic chamber connected online with a Nicolet 6700 FTIR spectrometer. Bro¨nsted/Lewis site ratio value at 150 °C has been determined from the ratio between the intensity of pyridine adsorption bands at 1545 cm-1 and 1450 cm-1 and taking into account the molar extinction coefficients of both adsorption bands (εB ) 1.67 cm µmol-1 and εL ) 2.22 cm µmol-1). 3. Results 3.1. Effect of Operation Conditions. Experiments have been carried out at normal pressure under the following operating conditions: in the 400-550 °C range; space time, up to 2.4 (g of catalyst) h (mol CH2)-1; time on stream, 5 h. The products have been grouped into the following lumps: (i) light olefins (ethylene, propylene, and butenes); (ii) paraffins (ethane, propane, and isobutene); (iii) C5+ aliphatics, which include all the olefins and paraffins with more than five carbon atoms; (iv) aromatics (benzene, toluene, and xylenes); (v) and methane, which is a minor compound. The insignificant formation of CO and CO2 has not been considered, given that their maximum yield is lower than 0.5% (of C transformed) at the maximun temperature, 550 °C, and for the minimun value of space time studied. The yield of each lump has been calculated as: Yi )
Fi F0
(1)
where Fi and F0 are the molar flow rates of the i lump in the product stream and of methanol in the feed, respectively, with both terms being expressed in CH2 equivalent units. The conversion of oxygenates has been defined as: X)
F0 - Fe F0
(2)
where Fe is the molar flow rate of oxygenates (methanol/ dimethyl ether) at the outlet of the reactor, expressed in CH2 equivalent units. Figure 1 shows the effect of space time on the yields of the different lumps of products (graph a) and on the yield of each
Ind. Eng. Chem. Res., Vol. 49, No. 24, 2010
12373
Figure 2. Effect of temperature on oxygenates conversion for different values of space time.
Figure 1. Effect of space time on product yields (graph a) and on individual olefin yields (graph b), at 500 °C.
individual olefin (graph b), at 500 °C. Figure 1a shows the absence of olefins and higher hydrocarbons in the stream of products for space time values lower than 0.25 (g catalyst) h (mol CH2)-1. A very rapid formation of hydrocarbons occurs at higher values of space time due to the autocatalytic nature of the reaction, which is a characteristic of the “hydrocarbon pool” mechanism.41-46 The yield of C2-C4 olefins peaks for a given value of space time, which indicates that they are intermediates in the kinetic scheme. For a space time higher than 1 (g of catalyst) h (CH2 equivalent mol)-1, the yield of olefins tends asymptotically toward constant values, higher than those corresponding to the thermodynamic equilibrium, which has also been observed by other authors.31 Figure 1b shows that propylene is the major olefin. The yield of the remaining products increases as the space time is increased and reaches constant values as in the case of olefins. The yield of C5+ aliphatics is lower than that corresponding to thermodynamic equilibruium, as a consequence of steric limitations (shape selectivity) of HZSM-5 zeolite, which hinders the formation of high molecular weight compounds, whose diffusion is restricted in the zeolite channels (5.4 × 5.6 Å). Figure 2 shows the significant effect of temperature on conversion. Thus, for a space time of 0.3 (g of catalyst) h (mol CH2)-1, the conversion is negligible below 450 °C, increases up to 0.06 at 500 °C, and above this temperature increases exponentially. For a space time of 0.56 (g of catalyst) h (mol CH2)-1, the exponential increase in the conversion with temperature occurs above 450 °C. For a space time of 2.4 (g of catalyst) h (mol CH2)-1, the conversion is complete in the whole temperature range studied. Figure 3 shows the effect of temperature on reaction product yields, for a space time of 2.4 (g of catalyst) h (mol CH2)-1, for which the conversion is complete above 400 °C, as shown in Figure 2.
Figure 3. Effect of temperature on product yields (graph a) and on individual olefin yields (graph b), for a space time of 2.4 (g of catalyst) h (mol CH2)-1.
The yield of light olefins is minimum at 450 °C (Figure 3a), as a consequence of their conversion to heavier hydrocarbons. Furthermore, the cracking reactions of heavy hydrocarbons to light olefins are enhanced above 450 °C. As a consequence, a progressive increase in ethylene and propylene yield is observed in Figure 3b, whereas the yield of butenes remains constant as temperature is increased, given that their cracking to light hydrocarbons is balanced by their formation from the cracking of heavier hydrocarbons. The yield of paraffins (80% isobutane) decreases above 450 °C, mainly due to the cracking capacity of isobutane and, to a lesser degree, of propene. The yield of aromatics is slightly influenced by temperature, with a very low crackability above 450 °C, whereas the yield of C5+ aliphatics steadily decreases as temperature is increased and significantly above 500 °C. Methane is a minor compound in the process,
12374
Ind. Eng. Chem. Res., Vol. 49, No. 24, 2010
established as the sum of squares of the differences between the experimental and calculated values of composition: nl
nl
OF )
∑wφ
i i
i)1
Figure 4. Contour map for C2-C4 olefins (expressed in percentage of CH2 equivalent units) for different combinations of temperature and space time.
although its yield increases significantly above 500 °C, as it is mainly formed by the thermal decomposition of oxygenates (methanol and dimetyl ether), which is enhanced by temperature. Figure 4 shows the contour curves for C2-C4 olefins (expressed in percentage of CH2 equivalent units) for different combinations of temperature and space time. The maximum olefin yield (∼50 wt %) is obtained at the maximum temperature studied (550 °C) and space time values between 0.6 and 1 (g of catalyst) h (mol CH2)-1. 3.2. Methodology for the Kinetic Study. Ideal flow (plug flow) has been assumed in the fixed bed reactor, and isothermal regime has also been assumed, since temperature differences at radial and longitudinal positions in the bed are lower than 1 °C. The reaction rate of each i compound at zero time on stream is calculated considering the different reaction steps in which this compound is involved: ri )
dXi ) d(W/F0)
j
∑ (υ ) r
i j j
(3)
where W is the mass of catalyst. Xi is the concentration of each lump, expressed as the molar fraction based on the organic compounds in the reaction medium, in CH2 equivalent units. With this definition, the results of the integration of the kinetic equations correspond to concentration units that are easy to understand because of their physical meaning and easy to relate to the yield and mass balance of the components. Moreover, (υi)j in eq 3 is the stoichiometric coefficient of component i in the kinetic step j, and rj is the reaction rate of the kinetic step j, with the concentration of each lump defined as the molar fraction based on all the components in the reaction medium (organics and inerts), yi. Thus, these units for concentration are equal to partial pressure, pi. This improves the physical meaning of concentration-dependent terms in the kinetic equations by adapting them to a meaning that is consistent with the mass law. The empiricism of the kinetic equation is lower than when using the concentration-dependent term based on mass fractions, as is common practice (to simplify calculations) in catalytic reactions with complex reaction schemes. The kinetic parameters for each kinetic model proposed have been calculated by multivariable nonlinear regression. Optimization has been carried out by minimizing an objective function
)
p
∑ w ∑ R (Xj* - X i
i)1
j
i,j
2 i,j)
(4)
j)1
where wi is the weight factor for each lump i of the kinetic scheme, φi is the sum of squares for the lack of fit, for each of these lumps, including the values obtained from the runs * j i,j is the repeated under the same experimental conditions, Rj; X average experimental value of composition of each lump i determined from the experiments repeated under the same conditions j, Xi,j is the corresponding value calculated integrating the mass balance for lump i, eq 3, nl is the number of lumps in the kinetic scheme, and p is the total number of variable combinations used in the experimentation. The kinetic parameters of best fit are the kinetic constants of each kinetic step j. In order to reduce the correlation between frequency factor and activation energy, the Arrhenius reparameterized equation has been used by expressing the kinetic constant, kj, as a function of its correspondent value, kj*, at a reference temperature, T*:
[ (
kj ) k*j exp -
Ej 1 1 R T T*
)]
(5)
The composition values of each lump for the proposed kinetic models have been calculated by integrating the expressions corresponding to eq 3, using a MATLAB program based on fourth-order finite-difference approximation. This program is combined with another one for multivariate nonlinear regression, also written in MATLAB, for calculating the kinetic parameters of best fit (the kinetic constants of individual reaction steps for a reference temperature (500 °C) and the corresponding activation energies, Ej). The significance and discrimination of the models proposed have been carried out following the procedure explained in detail in a previous work.22 3.3. Kinetic Model. Most of the kinetic models proposed in the literature to describe the transformation of methanol over catalysts prepared based on HZSM-5 zeolites are simplified models based on lumps, which have been proposed for MTG process conditions (
