LIQUID PHASE HYDROGENATION OF OLEFINS ON SUPPORTED PALLADIUM-ALUMINA CATALYST

Abstract

These investigations are concerned with the hydrogenation of some specific long chain olefins in liquid phase over a supported Pd-Al203 catalyst at moderate temperature and pressure conditions. As olefin feeds, a straight chain olefin, n-octene and a branched chain olefin isooctene (codimer of isobutene) were chosen for detailed investigations. Later, the hydrogenation of isododecene - a trimer of isobutene was also studied to know the effect of increasing chain-length on the reaction rate. The hydrogenation experiments were carried out in a —3 3 1.0 x 10 m stainless steel stirred autoclave with external heating and internal cooling arrangements. A commercial catalyst containing 0.5 % palladium on alumina support was used in the powdered form (-170 + 250 mesh) which was suspended in the liquid medium containing the olefin dissolved in n-heptane. Hydro genation was then carried out, with adequate stirring, under pressure varying the operating conditions in the following ranges to study their effects on the reaction rate: Reaction temperature: _ „ 3 03 -» 3 3 3 K Reaction pressure: 0.5 - 2.0 MPa - Olefin concentration: 0.19 -1.50 kmol .nT3 (ii) Catalyst loading: 5.56 . l9.50 kg.m~3 Stirring speed: 200 - 1200 r.p.m. In the beginning the effect of the stirrer speed was determined and then series of experiments were conducted at the stirring rate ensuring kinetic regime beyond the mass transfer limitation range. The conversion with time was followed by analysing olefin concentrations by bromine number. Complete analysis of the product was done by GLC. Using the experimental reaction rates and theoretical correlations the Thiele parameter for the powdered catalyst was computed and found to be about 10"1. This indicated that pore diffusion limitations with this catalyst were negligible. The influences of above mentioned operating conditions on the hydrogenation rate were studied for the three olefins separately. The reaction rate was observed to rise initially with increasing speed of the agitator and then levelled off at about 600 r.p.m. and above. The effects of other variables were, therefore, studied at the stirrer speed of 800 r.p.m. For the three olefinic feed stocks used in these studies the reaction rate was found to increase linearly with catalyst concentration. The rate of hydrogenation also increased with increasing hydrogen pressure,though (iii) non-linearly. Solubility of hydrogen in the reaction medium, which increases with pressure, was, therefore, considered for evaluating the rate dependence on hydrogen availability in the bulk liquid. Alongwith hydrogenation on this palladium catalyst some isomerisation of the double bond was also found to occur. By the time about 5 - 10 %conversion was attained, the distribution of olefins with terminal and internal double bonds almost reached the equilibrium composition. In general, the hydrogen concentration influenced the reaction rate more pronouncedly than olefin. In respect of olefin structures, the rate was found to decrease with branching in the molecule as well as with increase of the chain length. With the three olefins studied, the relative reaction rates were in the order: n-octene > isooctene > isododecene The results were then evaluated for the develop ment of kinetic models for the hydrogenation of these olefins. - From the effect of total pressure, initial olefin concentration and temperature on the initial reaction rate, it was found that surface reaction may be the controlling step. Hence several models (iv) were postulated for this controlling step and rate expressions formulated accordingly, m the development of these rate equations, the reverse reaction of olefin dehydrogenation was neglected since the thermodynamic calculations showed high equilibrium constant values. A non-linear least square analysis was then performed to compare the data obtained in experimental runs upto 90 minutes with the rate equations formulated. This covered the range of conversion upto about 94 %. Two independent techniques namely the Grid-Search and the Marquardt method were used for this analysis; computer programs were developed for use on an IBM 370/145 computer which minimized the sum of squares of residuals of experimental and predicted reaction rates. The rate equation which yielded no zero or negative constants and fitted the experimental data best was selected finally on the basis of an approximate statistical criteria and analysis of residuals. The results of the non-linear analysis led to the selection of the following model as the most satisfactory one to represent the experimental data for all the three olefins over the entire ranges of temperature and concentration studied. k K. L C C A B A B r = ~ ( 1*Jk£a )2 (1 ♦ K3Cb , (v) where k is the specific reaction rate constant; K and KB the equilibrium adsorption constants for hydrogen and olefin respectively; and C^ and CB are the hydrogen and olefin concentrations in the liquid phase. The average error between therates calculated from this model and those from the experimental data were + 1.32, + 1.69 and 3.01 % for isooctenes, n-octenes and isododecenes respectively. This model suggests that hydrogen in dissociated form and the olefin molecules are non-competitively adsorbed on the active sites of the catalyst. The constants in this rate equation are related to tempera ture in a similar way as Arrhenius relationship. The energies of activation were found to be 67.94, 69.65 and 64.42 kJ/mol for n-octene, isooctene and isododecene respectively. An attempt was made to correlate the rate of hydrogenation of these olefins with the molecular configuration of olefins. & energetic factors. The effects of diffusion in liquid filled pores on the rates of hydrogenation were also investigated by using-catalyst pellets instead of powder, in a —3 3 1.0 x 10 m stirred basket reactor. The same palladium catalyst originally obtained as cylindrical pellets of 4 mmx 4 mm size was used in these experi ments. With these pellets the reaction rates were found to be lower down to about 20 to 30 % of the rates (vl) observed with powdered catalyst; thus the effect of pore diffusion was quite significant. The effectiveness factor for these catalyst pellets and effective diffusivity of H2 were computed from the catalyst characterisation data using a simplified theoretical model. The catalytic effectiveness factors for the three olefins studied were dependent on the type of olefin and decreased in the order: isododecene > isooctene y n-octene The effective diffusivity of H,,, however, remained almost unaffected by the structural differences of these olefins.