Ind. Eng. Chem. Res. 1988,27, 1990-1995
1990
T, = m e a n temperature in t h e reactor (395 K), K Greek Symbols
used to express t h e decrease in adsorption enthalpy with surface occupancy for hydrogen isotopes y = power for deuterium in t h e power-law model a = power for hydrogen in the power-law model OH = fractional surface coverage for hydrogen 7 = parameter
Registry No. HP,1333-74-0; P d , 7440-05-3; D,, 7782-39-0.
Literature Cited Aldag, A. W.; Schmidt, L. D. “Interaction of Hydrogen with Palladium”. J . Catal. 1971,22, 260. Anderson, J. R. Structure of Metallic Catalysts; Academic: London, 1975; p 43. Bond, G. C. Catalysis by Metals; Academic: London and New York, 1962. Bonhoeffer, K. F.; Farkas, A. “On Adsorption and Reflection Processes in the Interaction of Hydrogen and Metals”. Trans. Faraday Soc. 1932,28, 242. Boreskov, G. K.; Vassilevitch, A. A. “ MBcanisme de L’echange Isotopique de L’hydrogBne sur de Films de Platine Evapores”. Actes. Congr. Znt. Catal., 2nd 1960,1, 1095. Boudart, M.; DjBga-Mariadassou, G. Kinetics of Heterogeneous Catalytic Reactions; Princeton University Press: Princeton NJ, 1984. Christman, K.; Ertl, G.; Schober, 0. ”LEED Intensities from Clean and Hydrogen Covered Ni(100) and Pd(ll1) Surfaces”. Surf. Sci. 1973, 40, 61. Conrad, H.; Ertl, G.; Latta, E. E. “Adsorption of Hydrogen on Palladium Single Crystal Surfaces”. Surf. Sci. 1974, 41, 435. Curtis Conner, W., Jr.; Cevallos-Candau, J. F.; Shaw, N.; Haensel, V. “Hydrogen Spillover and Surface Diffusion: Spillover from a Point Source”. In Spillover of Adsorbed Species; Pajonk, G . M., Teichner, S. J., Germain, J. E., Eds.; Elsevier: Amsterdam, 1983. Curtis Conner, W., Jr.; Pajonk, G. M.; Teichner, S. J. “Spillover of Sorbed Species”. Adv. Catal. 1984, 34, 1.
Eley, D. D. “The Absolute Rate of Conversionsof Parahydrogen by 1948, 44, 216. Metallic Catalysts”. Trans. Faraday SOC. Konvalinka, J. A.; Van Oeffelt, P. M.; Scholten, J. J. F. “Temperature Programmed Desorption of Hydrogen from Nickel Catalysts”. Appl. Catal. 1981, 1,41. Kramer, R.; Andre, M. “Adsorptionof Atomic Hydrogen on Alumina by Hydrogen Spillover”. J . Catal. 1979,58, 287. Lewis, F. A. The Palladium Hydrogen System; Academic: London, New York, 1967. Luft, G.; Romer, R.; Roder, H. “Kreislaufapparaturen fur reaktionskinetische Messungen“. Chem.-Zng.-Tech. 1973, 45, 596. Niklasson, C.; Andersson, B. “The Adsorption and Reaction of H2 and D, on a Ni/Si02 Catalyst” Znd. Eng. Chem. 1988a, 27,1370. Niklasson, C.; Andersson, B. “A Study of Hydrogen Spillover on Ni/SiOz under CSTR Conditions”. Submitted for publication in Appl. Catal. 1988b. Niklasson, C.; Smedler, G . “Kinetics of Adsorption and Reaction for the Consecutive Hydrogenation of 2-Ethylhexenal on a Ni/Si02 Catalyst”. Znd. Eng. Chem. Res. 1987, 26, 403. Rideal, E. K. “A Note on a Simple Molecular Mechanism for Heterogeneous Catalytic Reactions“. Proc. Cambridge Phil. SOC. 1939, 35, 130. Robell, A. J.; Ballou, E. V.; Boudart, M. “Surface Diffusion of Hydrogen on Carbon”. J . Phys. Chem. 1964,68, 2748. Scholten, J. J. F.; Konvalinka, J. A. “Hydrogen-Deuterium Equilibration and Parahydrogen and Orthodeuterium Conversion over Palladium: Kinetics and Mechanism”. J. Catal. 1966, 5, 1. Smedler, G. “Selective Hydrogenationof 2-Ethylhexenal”. Znd. Eng. Chem. Res. 1988, in press. Toyoshima, I.; Somorjai, G. A. “Heats of Chemisorption of 0 2 , Hz, CO, CO,, and Nz on Polycrystaline and Single Crystal Transition Metal Surfaces”. Catal. Rev.-Sci. Eng. 1979, 19, 1. Van Meerten, R. Z. C. “Gas-Phase Benzene Hydrogenation on a Nickel-Silica Catalyst”. Ph.D. Dissertation, Nijmegen, 1975. Vannice, M. A.; Hyun, S. H.; Kalpakci, B.; Liauh, W. C. “Entropies of Adsorption in Heterogeneous Catalytic Reactions”. J . Catal. 1979, 56, 358. Received for review February 16, 1988 Accepted June 28, 1988
Kinetics of Hydrogenation of 2-Ethylhexenal and Hydrogen/Deuterium Exchange on a Palladium/Silica Catalyst in a Continuously Stirred Tank Reactor Claes Niklasson Department of Chemical Reaction Engineering, Chalmers University of Technology, S-41296 Goteborg, Sweden
The kinetics of the consecutive hydrogenation of 2-ethylhexenal to 2-ethylhexanal and 2-ethylhexanol and the parallel H2/D2exchange were studied in a continuously stirred tank reactor (CSTR) in the presence of a Pd/Si02 catalyst. The kinetics for H2/D2exchange, during hydrogenation, were found to fit a Langmuir-Hinshelwood (L-H) type of model with equally competitive dissociative adsorption of hydrogen and associative adsorption of the unsaturated aldehyde on one and the same type of palladium site. The kinetics of the hydrogenation were described by a L-H type of kinetic rate model modified by decreasing enthalpy of adsorption for hydrogen with the square root of the total pressure of hydrogen isotopes. Natural hydrogen (H,) was found to hydrogenate the unsaturated aldehyde approximately twice as fast as the other hydrogen isotopes (HD, D2) did. The catalyst was found to be completely selective; i.e., no 2-ethylhexanol or 2-ethylhexenol was found. Selective hydrogenations of unsaturated aldehydes have been of g r e a t interest d u r i n g the last decades. Hydrogenations of crotonaldehyde (Noller and Lin, 1984) and 2-ethylhexenal (Collins et al., 1983) are two of the most studied processes a m o n g these. Since the industrial production of the saturated aldehyde and the alcohol from 2-ethylhexenal is performed in the liquid phase, the majority of p u b l i s h e d papers are concerned with studies of the reaction in t h i s phase. However, the results f r o m the
gas-phase hydrogenations under certain conditions can easily be transferred to comparable properties in the liquid-phase hydrogenations (Magnusson, 1983). Liquidphase hydrogenation of 2-ethylhexenal has been performed on different t y p e s of catalysts (Marcelin et al., 1984; Sousa-Aguiar and Schmal, 1980). H i g h selectivity with respect to 2-ethylhexanal has been reported when using palladium (Rylander and Himelstein, 1964; Mucho and Polievka, 1969) or nickel boride (Collins et al., 1983) as
0888-5885/88/2627-1990$01.50/0 0 1988 American Chemical Society
Ind. Eng. Chem. Res., Vol. 27, No. 11, 1988 1991 catalyst. The saturated aldehyde has in general been the most desirable product but the complete hydrogenation to the alcohol with very low yield of intermediate compounds could be achieved with a fused iron catalyst activated with V20, (Glebov et al., 1982). Most articles about hydrogenationof 2-ethylhexenal give only qualitative reasoning, and only a few discuss the adsorption or desorption of involved active species. Moreover, the influence of the catalytic surface on the chemical reaction steps has not been the subject of any study. Kinetic modeling of the hydrogenation of 2-ethylhexenal has been done by many authors (Ioffe et al., 1982; Collins et al., 1983; Bel'chikova et al., 1978; Palla Carreiro and Baerns, 1983; Tanaka and Yada, 1957). However the earlier literature is lacking a proposal of a kinetic rate equation with physical or chemical meaning to indicate how the hydrogenation is actually proceeding on the catalytic surface. Guidelines concerning dominant adsorption mechanisms, heats of adsorption, and competition for active sites are also desirable, and a discussion about what influence a carbonaceous deposit might have on the activity or selectivity of the catalyst would be useful. The objective of this paper is to propose a kinetic model of the Langmuir-Hinshelwood type that describes the hydrogenation process. From this model and other experimental results, conclusions can be drawn about adsorption mechanisms, competition of active species on the surface, dominating surface reaction step, and the ratedetermining step for this reaction. One aspect of the knowledge concerning this hydrogenation is the parallel hydrogen-palladium interaction. A method to measure this exchange during hydrogenation, originally presented by Magnusson (1983) and developed in this paper, includes the hydrogenation of the unsaturated compound with a mixture of Hzand D2. Measuring the rate of formation for HD as well as the different hydrogenation rates provides useful information. This information can then be investigated theoretically and lead to a better understanding of the underlying reaction mechanism for the hydrogenation process. Kinetic data from the H2/D2exchange reaction are also used to decrease the number of parameters in the total hydrogenation regression model. A review of the H2/D2exchange reaction occurring on different metal catalysts is given by Bond (1962).
Experiments The Catalyst. The catalyst was a Pd/Si02 catalyst prepared by a standard wet impregnation technique described elsewhere (Smedler, 1988). The support, supplied by Girdler (T-1571), consisted of porous silica pellets with a BET surface of 111.5 m2/g and a mean diameter of 5 X m. The catalyst content of palladium was 0.16% Pd ( U S , Perkin Elmer 370), with a total weight of 5.05 X kg, and a hydrogen uptake of 6.45 X mol of H/kg of catalyst (Standard H2 chemisorption Chemisorb 2800). The calculated dispersion was 42.9%. Experimental Procedure. The reactor used in all experiments was a continuously stirred tank reactor (CSTR) of Sunderland type described elsewhere (Niklasson and Smedler, 1987). The experimental devices, computer control, and experimental procedure are described in the same study. To avoid any serial correlation between the independent variables in the system, the time orders of reactor temperatures and flow combinations were randomized by the computer. For all experiments, the rate of hydrogenation, total pressure, and temperature in the reactor were measured. Parallel to these measurements, data on the exchange rate for H2/D2(partial pressures of hydrogen isotopes) were continuously sampled with the
mass spectrometer during the hydrogenation. Reproducibility experiments were performed for each temperature level in the reactor as well as in the evaporators (controlling the concentration of 2-ethylhexenal in the inflow) for selected flows of nitrogen, hydrogen, and deuterium. The overall stability of the catalyst was tested at the end of every experimental series with a duplication of the first 10 experiments. The partial pressures of hydrogen isotopes and the reactor temperature were chosen so as to avoid the palladium dispersed on the carrier being transformed to the &phase according to given directions (Scholten and Konvalinka, 1966). This control measure was necessary to avoid high ratios of H/Pd in the bulk phase of palladium. The 2-ethylhexenalwas delivered by BEROXO AB (Sweden) and had a purity of 98%. The impurities consisted of 2-ethylhexanal and 2-ethylhexanol in equal amounts and some traces of C4 aldehydes. All gases were of SR quality, and the nitrogen used was further purified in an oxygen trap (Alltech Associates). Analysis. The analysis of the hydrogen isotopes was performed in a quadrupole mass spectrometer (BALZERS QMG 311). The separation and analysis of aldehydes and alcohols were performed by means of a gas chromatograph supplied with a Supelco glass column (15% DEGS on 80/100 Chrom. WHP, 2 m). For each stationary experiment, five separate GC analyses were made.
Results Kinetics of the H2/D2Exchange Reaction under Hydrogenation. To understand the hydrogenation of 2-ethylhexenal in every detail and to get a complete description of the system, information about the underlying separate elementary steps is necessary. Therefore, data about the hydrogen adsorption and desorption on the fresh catalyst as well as on the hydrogenation catalyst are a useful tool in this investigation. The use of kinetic data from measurements of H2/D2exchange reaction on the fresh catalyst (Pd/Si02)and some assumptions about how the catalyst changes its behavior during hydrogenation make it possible to derive some plausible kinetic models. The kinetics of the Hz/Dzexchange reaction on this catalyst (Pd/Si02),prior to any hydrogenation occurring on the surface, are described elsewhere (Niklasson, 1988). The equilibrium reaction, given in eq 1, was studied kinetically during the hydrogenation of 2-ethylhexenal to 2-ethylhexanal:
*
Hz(d + Ddg) 2JWg) (1) Some typical experimental results are given in Table I. The reproducibility ranged from 0.2% to 4% and did not significantly depend on reactor temperature or concentration levels. No significant short- or long-time deactivation of the palladium catalyst was detected. Kinetic models for this exchange reaction are preferably Langmuir-Hinshelwood or Eley-Rideal types. A discrimination between proposed models was made statistically and physically. A Langmuir-Hinshelwood type of model with mutual competition for hydrogen isotopes and 2ethylhexenalon the metal surface was found to fit the data best, without any trends in the residuals: k+ rHD = T2[(2PHz + P H D ) ( 2 p D z + PHD)/Ptot - 2PHDI [l -I-exp(ASAv/R- AHAv/RT)pA' + (exp(ASH/R m ~ / R T ) p t o , ' ) ' / ~ I - (2) ~ where AHH= AHH' - vpF2/2is the enthalpy change upon adsorption for hydrogen isotopes, at zero surface coverage for hydrogen, corrected for a decrease with total partial
1992 Ind. Eng. Chem. Res., Vol. 27, No. 11, 1988 Table I. Typical Experimental Results for the Hydrogenation of 2-Ethylhexenal to 2-Ethylhexanal and the Parallel H2/D2 Exchange Reaction PA,
Torr 10.3 10.2 10.2 26.3 26.2 26.2 11.3 10.9 10.7 27.0 24.7 23.8 11.1
10.9 11.2 10.9 26.8 22.5 21.9 10.8 11.1 10.9 10.8 24.7 23.8 26.3 11.1 11.2 11.2 11.1 a
PH~I
Torr 114.0 30.6 40.2 112.0 39.6 91.8 22.3 105.9 78.3 22.2 77.4 104.4 102.3 71.1 18.6 102.2 18.5 70.3 100.9 85.6 26.4 109.8 85.7 26.2 108.4 84.5 92.9 30.5 40.1 92.9
Reproducibility experiments.
PDp
Torr 30.8 115.2 39.6 30.4 39.0 93.6 106.7 22.8 80.2 104.9 79.3 22.7 19.2 73.0 103.0 19.1 101.1 72.1 19.0 87.5 110.8 26.7 87.6 109.2 26.5 86.4 94.8 114.9 39.5 94.8
PHD, Torr 23.9 25.1 15.9 23.0 15.2 40.3 41.5 39.8 71.4 40.1 69.1 38.6 46.9 85.7 49.0 47.1 47.5 83.9 46.1 56.8 33.4 32.0 56.6 32.5 31.0 54.7 42.4 25.3 16.0 42.4
he, pmol/(s kg cat.) 915 1108 770 757 638 2129 1750 1743 3497 1640 3360 1520 2016 3941 2028 2015 1965 3872 1940 2763 1527 1533 2825 1366 1412 2674 2082 1135 770 2090
TABe,
pmol/(a kg cat.) 17.5 9.4 6.7 14.2 8.7 19.3 26.6 43.7 52.9 53.7 75.6 73.8 44.6 48.5 39.3 41.7 72.1 87.7 74.9 38.0 26.4 33.9 36.0 28.1 36.4 43.2 16.1 11.2
9.3 16.5
T,K 364.5 364.4 364.5 364.5 364.5 364.5 405.1 405.0 405.2 405.1 405.2 405.1 424.1 424.1 424.5 424.5O 425.0 425.1 425.2 385.6 385.3 385.4 385.2" 385.5 385.3 385.5 365.2b 365.2b 365.1* 365.1"~~
Long-time stability test experiments.
Table 11. Kinetic Parameters of the Langmuir-Hinshelwood Model Describing HD Exchange on Pd/Si02 95% conf. value limit units kt 3.8 X lot2 1.1 X lot1 pmol/(s kg of catalyst Torr) m A ' -3.8 X 3.0 X lot3 J/mol ASA" -7.2 X lo+' 8.5 X IOto J/(mol K)
pressure of hydrogen isotopes. The derivation procedure for this model is given elsewhere (Niklasson and Andersson, 1987). A small isotope effect was detected for the exchange on the fresh catalyst, but as the kinetic expression showed no significant trends with the individual partial pressures of hydrogen and deuterium, this could be disregarded. In the experimental data, the enthalpy of adsorption of hydrogen on the metal was found to depend on the total partial pressure of hydrogen isotopes. Usually the enthalpy of adsorption is correlated to the coverage of hydrogen, as recommended by Boudart and DjBga-Mariadassou (1984), but for small surface coverage the square root of ptotis a good approximation. This procedure is also statistically more convenient to use since it lowers the number of parameters and simplifies the regression procedure (notice the similarities with the Frumkin-Temkin isotherm where AH = AH0 - 78). Values of K A and KH,from the regression analysis, are given in Table 11; values for A H H o and 7 are given in the preceding paper in this issue (Niklasson (1988): AHHO = -6.8 X lo4 J/mol, 7 = -6.8 X lo2 J/(mol Torr@)). The proposed model explains 98.4 % of the total variation about the mean for the reaction rate (R2 = 0.984 as defined by Draper and Smith (1966)). According to the experimental results, the activation energy for hydrogen absorption was found to be not significantly far from zero. Therefore, this parameter was excluded from the regression. Table I11 also shows that the adsorption rate con-
Table 111. Mean Ratio between the Rate of H2/D2Exchange and the Rate of Hydrogenation of 2-Ethylhexenal (rHDe/rme), as a Function of Temperature and Partial and HD) Pressure of Hydrogen Isotopes (Hz,D2, an+: Torr T, K 90 170" 170b 230 370 82.9 52.2 109.6 117.6 61.8 43.7 68.8 390 78.5 410 49.6 40.0 65.4 72.7 430 42.7 45.2 48.3 81.1
'PD2