Hydrogenation of Vegetable Oils over Pd on Nanocomposite Carbon

Catalytic hydrogenation of vegetable oils was studied over Pd nanocomposite carbon catalysts. The mesoporous structure of the carbon support was benef...
5 downloads 0 Views 297KB Size
Ind. Eng. Chem. Res. 2008, 47, 7219–7225

7219

Hydrogenation of Vegetable Oils over Pd on Nanocomposite Carbon Catalysts Irina L. Simakova,*,† Olga A. Simakova,† Anatoliy V. Romanenko,† and Dmitry Yu. Murzin*,‡ BoreskoV Institute of Catalysis, NoVosibirsk, Russia, and Åbo Akademi UniVersity, Åbo/Turku, Finland

Catalytic hydrogenation of vegetable oils was studied over Pd nanocomposite carbon catalysts. The mesoporous structure of the carbon support was beneficial to achieve fast hydrogenation rates and the desired cis/trans ratio of monoenic products of edible oil hydrogenation. The preparation procedure developed for a slurry catalyst was successfully utilized for fixed bed applications. Industrial experience of Pd/nanocomposite carbon utilization in selective hydrogenation of edible and total hydrogenation of nonedible oils is described. Introduction Vegetable oil hydrogenation has been a subject of extensive research for many years.1-5 It is an important process due to its wide application in the production of edible fats and margarine. Partial hydrogenation of diunsaturated fatty acids and their esters improves stabilization toward oxygen. At the same time, since the melting point should not be increased, special care is usually taken on the ratio between Z (cis) and E (trans) isomers in monounsaturated acids.6 In some instances, complete hydrogenation of fatty acids is desired.7 Various catalysts were applied along the years for hydrogenation of vegetable fats and oils. Already in 1906 palladium catalysts were used for vegetable fat hydrogenation in the first margarine plants. In the late 1920s, nickel and nickel-copper catalysts were developed. These catalysts replaced palladium primarily due to lower costs, albeit higher temperatures were required for nickel.8 Another drawback of nickel catalysts is in their toxicity, which is important as the content of nickel according to regulations in, for example, Russia cannot exceed 10 mg of Ni/kg of hydrogenated fat.9 In addition, nickel catalysts are deactivated due to formation of nickel soaps, which is prominent in the presence of moisture and oxygen. Therefore, palladium catalysts made a comeback in 1960s. Palladium catalyst activity in processes of unsaturated CdC bond hydrogenation, positional and transisomerization was studied; characteristics of palladium catalysts supported on porous materials of different nature were investigated; and the influence of temperature, hydrogen pressure, catalytic concentration on the fatty acid, and isomeric composition of hydrogenated oils was determined.3,10,11 From the large body of studies, it was concluded that palladium catalysts are more effective than nickel ones at lower temperatures, which implies lower trans isomer content. The range of fatty acid products is similar for both catalysts.12 Control of catalyst-to- reactant ratio as well as hydrogen pressure is of crucial importance for palladium catalysts. Hydrogenation of fatty acids and triglycerides is a fast process; therefore, particular care should be taken with respect to mass and heat transfer. It means that a large interface between hydrogen, catalyst, and substrate is required, as transport limitations (gas-liquid, liquid-solid, and inside the catalyst particles) were reported to influence selectivity.13,14 * To whom correspondence should be addressed. E-mail: simakova@ catalysis.ru, [email protected]. † Boreskov Institute of Catalysis. ‡ Åbo Akademi University.

The engineering approaches to influence gas-liquid and liquid-solid mass transfer15 are well established also for hydrogenation of fatty acids.13 There is, however, still a quest for development of more selective catalysts, as the geometric isomerization producing trans-fatty acids can be influenced by the choice of support, its size, distribution along the catalyst particle, pore size, and other characteristics.16,17 For instance, according to ref 9, palladium supported on wideporous titanium dioxide and alumina is characterized by high initial activity, but is prone to fast deactivation, caused by active metal blocking with impurities from the feed and water adsorbing on the hydrophilic surface. Palladium catalysts on narrow porous activated carbon have lower hydrogenation rate and higher trans-isomerization rate than palladium oxides with wide pores.18 This implies that some optimal pore structure is required. Recently, palladium on structural mesoporous material (SBA-15) was utilized for hydrogenation of sunflower and canola oils.17 However, keeping in mind significant costs associated with synthesis of such mesoporous materials, it is not surprising that research is done continuously to improve the carbon materials, as catalyst supports, since they are very attractive due to tunable chemical and textural surface properties. Carbon-based monoliths were applied as catalyst supports for selective hydrogenation of edible oils.19 Despite the promising features of monolithic catalysts, they still have to make a breakthrough; therefore, in the present study, we were focusing on more conventional geometries. Summarizing the literature, it could be concluded that the following factors should be considered while developing active and selective catalysts for three-phase hydrogenation of vegetable oils: (a) high dispersion of palladium particles and resistance to sintering; (b) active metal accessibility for the reagent molecules (palladium particles should be inside the pores of the size larger than the size of the substrate, e.g., 1.5 nm; transport pores must exceed the reactant dimensions by one order of magnitude; the catalyst particle size should be less than 50 µm for slurry hydrogenations); (3) oil-support interactions (surface functional groups of the support should not lead to any decomposition or other side reactions of oil; adsorption of water and other polar impurities in oil should not result in catalyst deactivation); (4) resistance of the catalyst and the mixture with oil to spontaneous ignition. The paper describes the experience of the authors in utilization of carbon-supported catalysts in laboratory, pilot, and industrial

10.1021/ie800663j CCC: $40.75  2008 American Chemical Society Published on Web 08/23/2008

7220 Ind. Eng. Chem. Res., Vol. 47, No. 19, 2008 Table 1. Fatty Acids Composition of the Tested Feedstocks (Mass %) fatty acid

FFA

C14:0 C16:0 C16:1 C18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 melting point, ° C

0.3 17.2 0.5 7.6 36.6 30.0 1.7 1 2.5 0.3 28.4

rapeseed oil 5.0 1.1 0.5 62.9 21.8 6.8 1.6 0.3 10.2

sunflower oil

palm oil

0.09 6.43 0.14 3.31 25.55 64.3

1.35 64.88 0.4 4.6 22.72 4.43 0.8 0.48

0.17 nd

51

scale for vegetable oil hydrogenation. Composition of the tested feedstocks is presented in Table 1. Experimental Section Catalyst Preparation. The problem of active component accessibility for the reacting molecules can be solved by utilizing porous composite materials “Sibunit” with regulated porous size and volume.20,21 Sibunit is a class of porous carbon-carbon composite materials combining the advantages of graphite (chemical stability and electric conductivity) and active coals (high specific surface area and adsorption capacity). These composites are characterized by a high mesopore volume and a controllable narrow-pore-size distribution. Carbon composite material Sibunit is produced by pyrolitic carbon deposition on a granulated carbon black.21 Pregranulated carbon black is covered by a fixed amount of pyrolitic carbon through condensation or chemical vapor deposition. During a subsequent activation stage with steam at 700-850 °C, a part of the carbon is removed by gasification. A spongelike system including meso- and macropores whose dimensions depend on the dispersion of the initial carbon black is formed. When the carbon black is burnt out almost completely, the obtained granules lose their mechanical strength and are destroyed into shelllike fragments. Such a procedure allows control over a very wide range of structural and textural properties (the specific surface area can be varied from 0.1 to 800 m2/g, the pore volume from 0.1 to 2.0 cm3/g). Sibunit has found applications as adsorbents in medicine and the food-processing industry, water and sewage treatment, and nonferrous hydrometallurgy as well as catalyst supports in hydrogenation, hydrotreating, hydrodechlorination reactions, and more recently ammonia synthesis.21 In the case of hydrogenation of fatty acids, a system of wide pores with the size 50-80 nm and volume 0.2 cm3/g is able to provide efficient transport of oil and hydrogen molecules to palladium particles, which are situated on a well-developed surface (300-400 m2/g) with pores of the size 2-6 nm. For the slurry hydrogenation experiments, the fractional composition of 5-50 µm was selected with the majority of particles below 15 µm following the requirements of efficient mass transfer.15 The particle size of the Pd/C catalyst was measured by a Coulter Counter “TA” apparatus. The electron microscopy (TEM and HREM) data were obtained by electron microscopes JEM-100CX (JEOL) with a 4.5-Å resolution and JEM2010 with a 1.4-Å resolution. The surface of the carbon support surface is formed by basal facets of carbon microcrystals of low polarity and oxygencontaining groups (C-OH, CdO, COOH) located at edge faces.22 Because of low polarity, the catalyst surface under

testing condition is mainly covered by the reactants and is not blocked by water and other polar compounds contained in the oil. The method of palladium deposition on carbon supports was selected based on detailed investigation of palladium metal particles’ formation while depositing them from water solutions of chloride complexes.23 This method involves a preparation of a palladium hydroxide solution of colloids with a size of 2-3 nm and their adsorption on accessible carbon support surface. For synthesis of Pd/C catalyst via the control of palladium particle size and distribution through the support grain,24 any water-soluble salt of PdII is appropriate, but the most usable is H2PdCl4 or Na2PdCl4. Basic agents such as hydroxides, carbonates, and bicarbonates of alkali metals or nitrogen-containing bases, usually Na2CO3, are utilized for palladium deposition. The Pd/C catalyst preparation method based on hydrolysis of H2PdCl4 at pH e5-6 and deposition of palladium hydroxide onto carbon support was thoroughly studied in refs 25 and 26. In this work, Pd/C (Sibunit) catalysts with 0.5-5 wt % loading of Pd were prepared by hydrolysis of H2PdCl4 at pH 5-6, as well as at pH g7, which gives so-called polynuclear hydroxocomplexes of palladium. These complexes were generated by mixing aqueous solutions of Na2CO3 and H2PdCl4, with a Na/Pd atomic ratio ranging from 5 to 21. The precursor was adsorbed on carbon, and the pH of the slurry was increased up to a Na/Pd ratio 1:2. Different mixing sequences were utilized. One of the methods was addition of Na2CO3 solution to carbon powder suspended in aqueous H2PdCl4. The 0.2 g of Sibunit was suspended in 5 cm3 of distilled water on agitating for 30 min. The 1 cm3 of 0.019 M H2PdCl4 solution was added to the slurry with a peristaltic pump (0.33 cm3/min) followed by the addition of 2 cm3 of a Na2CO3 solution. Finally, the mixture was agitated for 6 h. Carbon samples with the supported palladium hydroxide were thoroughly washed by distilled water, dried at 373 K, and reduced in a hydrogen flow at 423 K during 1 h. Since the influence of diffusion on activity and selectivity is well documented in fatty acid hydrogenations for fixed bed applications, a procedure to prepare egg-shell palladium catalysts on nanocomposite carbon was utilized. Samples of Pd/C catalyst with palladium content from 0.5 to 1.0 wt % were prepared by an incipient wetness impregnation method, which includes spraying the solutions H2PdCl4 and Na2CO3 onto the carbon granules (spheres by diameter of 2-3 and 4-6 mm) agitated in a rotating cylinder. In the first method, both water solutions of H2PdCl4 and Na2CO3 with the molar ratio 1:2 are sprayed independently with the same flow rate at room temperature. Alternatively, the water solution of H2PdCl4 and Na2CO3 with the same molar ratio is ripened for 18 min to form polynuclear palladium(II) hydroxocomplexes.27 After metal impregnation and drying at room temperature and subsequently under vacuum at 70 °C, the catalyst is reduced in the flow of hydrogen at 250 °C. Washing with water was performed until no Cl- ions could be detected in wash water followed by drying under vacuum at 70 °C. Similar to catalysts for slurry hydrogenation, the activated carbon Sibunit has been chosen as a carbonaceous support because of its high surface area and chemical inertness in acidic and basic media.28 Preferable carbon granule size was determined to be less than 4 mm based on the pressure drop calculations for triglyceride and fatty acid feeds. Catalytic Experiments. Hydrogenation was performed in a three-phase 150-mL stainless steel autoclave. A stirring rate of

Ind. Eng. Chem. Res., Vol. 47, No. 19, 2008 7221

Figure 1. Reactor setup for fixed bed hydrogenation.

1100 rpm was used in the experiments to avoid mass-transfer limitations. The amount of catalyst was 100 mg, while the amount of substrate was equal to 20 g. The experiments were conducted at 371 K at hydrogen pressures ranging from 0.1 to 1.0 MPa. Total reaction time was 2-3 h, depending on the metal loading, the catalyst activity, and the catalyst amount. The samples were periodically taken during kinetic runs and analyzed chromatographically (“Tsvet -500”) using 15 m × 0.25 mm × 0.5 mm quartz capillary column, Carbowax-DVB, and a flame ionization detector operating at 250 °C. In the gas chromatographic analysis, diunsaturated acids, monounsaturated acids, and saturated acids were calculated separately. In addition, hydrogen consumption from a calibrated reservoir was measured. Stereoselectivity in hydrogenation was obtained by IR analysis of methyl esters.29 To this end, the samples were esterified by sodium methylate. Infrared transmission spectra were recorded in the 400-6000-cm-1 region with 4-cm-1 resolution using the Shimazdu FTIR 8300 spectrometer equipped with the diffusion reflection accessories DRS-8000. The catalytic hydrogenation experiments with several feedstocks (Table 1) were carried out also in a tubular fixed bed reactor (Figure 1). The reactor length and the inner diameter were 583 and 24 mm, respectively. The catalyst (∼14.15 g, with a maximum load of 50 g) in the form of spherical particles (fractions 1.5, 2-3, and 4-6 mm) was placed in the reactor (Figure 6) and diluted with quartz (20 mL with the size of 3-5 mm). The experiments were typically carried out in an upward mode with a feed flow of 10 g/h. Reagents passed through a preheated reservoir and then flowed over the fixed catalyst bed. The hydrogen gas flow metered into the system using a mass flow controller was 200 cm3/min. The experiments were performed using a stainless-steel up-flow reactor at a temperature range from 146 to 225 °C and hydrogen pressure up to 8 bar. Condensable products were collected in a water-cooled collector and analyzed by the GC method. Results and Discussion Catalysts. According to TEM (Figure 2), metallic palladium particles of the size 2-3 nm are uniformly distributed on the surface of the wide porous support. Profiling analysis (method of microprobing) was done for 1% Pd/Sibunit intended for fixed bed applications with the catalyst particle size of 1.5 mm (Figure 3) demonstrated that palladium

Figure 2. TEM of 0.5% Pd/Sibunit.

Figure 3. Palladium distribution on egg-shell type 1% Pd/Sibunit with the catalyst particle size of 1.5 mm. Description of the preparation methods I and II is given in the text.

is predominantly located within a distance of 18 µm from the outer surface for catalyst I. XRD of the spent catalyst and XRF analysis of the product indicated that there were no changes in the metal crystallite size; e.g., no metal sintering was present. The metal phase turned

7222 Ind. Eng. Chem. Res., Vol. 47, No. 19, 2008

Figure 4. Concentration vs time dependence for hydrogenation of linoleic acid triglyceride over 1% Pd/Sibunit (2.5 wt % catalyst concentration, 98 °C).

out to be stable in the experimental conditions used for hydrogenations. The fact that no sintering was obtained is consistent with the literature. In particular, investigation of palladium sintering on carbon supports under high temperature in hydrogen and vacuum30 showed that stability of the palladium particles is mainly determined by the structure, orientation, and surface chemical composition of the carbon microcrystals. Palladium on porous carbons was demonstrated to be rather stable without losing metal dispersion even at temperatures close to 300 °C.30 In addition, it should be noted no palladium leaching was determined even after 110 h of operation in a laboratory fixed bed reactor. Hydrogenation in Slurry System. The observed reaction rate was linearly dependent on the catalyst loading in the reactor until a plateau was reached at significant catalyst bulk densities. For hydrogenation of triglycerides of linoleic acid at 0.12 MPa of hydrogen pressure at 98 °C, the region of gas/liquid mass transfer was achieved at a catalyst concentration of 2.5 wt %. Kinetic regularities obtained for vegetable oil (triglyceride of linoleic acid) hydrogenation over 1 wt % Pd/C are demonstrated in Figure 4. The time dependence of consumed hydrogen volume (not shown here) demonstrates a monotonic increase approaching a plateau after significant reaction time. The analysis of the product composition indicates that the initial steep part of the curve corresponds to hydrogenation of the triglyceride of linoleic 2 ) 1 ) acid (C18 ) while the triglyceride of oleic acid (C18 ) hydrogenation occurs with some delay; in the break point for the hydrogen pressure domain of 0.1-1.0 MPa at conversion of 1 ) C218) being ∼90% the selectivity toward C18 formation is 95-96%. 2 ) Concentration of C18 on time dependence demonstrated in 2 ) Figure 4 indicates the first order of the reaction in C18 1 ) concentration. C18 concentration also decreases with time in an exponential fashion at high conversions pointing on firstorder dependence in monounsaturated acid hydrogenation as well. Addition of H2PdCl4 solution to carbon powder suspended in aqueous Na2CO3 with the different ratios between Na and Pd (χ ) 5-21) resulted in catalysts with somewhat different metal particle sizes, ranging from 1.6 to 2.4 nm. The values of TOF for rapeseed oil hydrogenation over 1% Pd on Sibunit at

Figure 5. Cis/trans ratio at complete conversion as a function of hydrogen pressure in sunflower oil hydrogenation in a slurry reactor over 1% Pd/ Sibunit, mcat ) 100 g, moil ) 20 g.

Figure 6. Cis/trans ratio at complete conversion as a function of reaction temperature in sunflower oil hydrogenation over 1% Pd/Sibunit, mcat )100 g, moil)20 g, and P ) 0.2 MPa.

373 K and a hydrogen pressure of 6 atm were in the range 2.5 -3.1 s-1. Similar albeit slightly lower values were achieved when, during catalyst preparation, Na2CO3 solution in carbon powder was suspended in aqueous H2PdCl4. For metal particles of the average size 1.8-2.5 nm, TOF values were ∼2.4 s-1, showing an insignificant dependence of the rate on metal dispersion at least within the studied domain. The vegetable oil hydrogenation over 1 wt % Pd/C proceeds with high rates already starting at 371 K. The reaction rate as expected increases with temperature, and in the domain 140-180 °C, the apparent activation energy in experiments of sunflower oil hydrogenation over 1% Pd/Sibunit, at 0.2 MPa, mcat )100 g, moil)20 g was 25.3 kJ/mol. With further temperature increase, the rate constants decrease gradually due to accumulation of thermal decomposition products in the batch reactor, eventually decreasing the catalyst activity. Interestingly enough, in a semibatch reactor with continuous bubbling of hydrogen through oil, the catalyst works in a stable mode up to 513-523 K. The original vegetable oils contain mainly unsaturated CdC bonds in the cis form. Hydrogenation of triglycerides containing di- and triunsaturated fatty acidic fragments results in triglycerides of monounsaturated fatty acids with both cis and trans

Ind. Eng. Chem. Res., Vol. 47, No. 19, 2008 7223 a

Table 2. Hydrogenation Data over 0.5 wt% Pd on Sibunit Catalysts in the Fixed Bed Reactor Operating a Semibatch Modeb fatty acid

sunflower oilc

sunflower oil

rapeseed oil

FFA

50%/50% rapeseed/Palm oil

reaction T, °C reaction time, h C12:0 C14:0 C16:0 C16:1 C18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 melting point, °C iodine number

106-130 0.33

155-170 3

155-175 3.5

155-175 2.5

0.09 6.33

5.38

155-175 4 1.47 0.73 12.95

89.58 2.73

84.20

60.51

6.3 0.07 6.79 45.02 41.63

92.97 0.23

0.80 37.34

0.18

0.38

2.41

0.64

1.14

nd 110

65.5 0.2

64 2.3

61 0

56.8 0

a Fraction size 2-3 mm, hydrogen pressure 0.6 MPa, catalyst mass 14.15 g, hydrogen flow rate 300 cm3/min, and charge 50 g. b Industrially required product specification: MP, 53-65 °C; iodine number, 2.5 -30.0. c Cis/trans ratio in C18:1 acids is equal to 4.4.

double bonds. It is well-known that the ratio between cis and trans isomers influences the physiological31 as well as the physical properties of the hydrogenated fat, such as melting temperature, hardness, ability to form plastic mixtures with other fats, etc.32 The factors influencing the cis/trans composition are hydrogen pressure, catalyst activity, and loading as well as process temperature. These parameters determine hydrogen concentration on the catalyst surface, which is crucial not only for hydrogenation of double bonds per se but for cis/trans isomerization as well. One of the mechanistic explanations for the enhancement of double bond isomerization in the presence of hydrogen33 is involvement of hydrogen in this reaction. Addition of the first hydrogen atom leads to surface intermediates, e.g., halfhydrogenated species, which are either reacting with the second hydrogen or alternatively hydrogen is subtracted from this intermediate giving finally an isomerization product. The relative ratio of these routes can depend on the surface concentration of hydrogen; in particular, cis/trans isomerization is more prominent under the hydrogen deficiency, while hydrogen addition and formation of the saturated C-C bond is predominant if surface hydrogen concentration is high.33 Investigation of product composition for three-phase slurry vegetable oil hydrogenation over 1 wt % Pd/C demonstrated that at hydrogen pressures above 0.1-0.2 MPa content of trans isomers in the product, which does not contain diunsaturated acids, is 15.5-18.5%. It corresponds to the level achieved on nickel and palladium catalytic systems without diffusion limitations of hydrogen.5 The values of cis/trans ratios are given in Figure 5, showing rather low trans selectivity. Further increase of hydrogen pressure up to 1.1 MPa allows for a decrease in the content of trans isomers to 4-5%. Experiments with the catalysts of different dispersion showed very similar stereoselectivity. It is important to note that palladium on the nanocomposite carbon is able to keep low trans-isomerizing ability in a wide range of reaction temperatures (Figure 6). In particular, in the temperature range of 170-200 °C, the final content of trans isomers, when dienic acids are no longer present in the reaction mixture, is in the range of 17-19%. It also clearly follows from Figure 6 that the cis/rans ratio does not change with temperature increase; which means, that when cis/trans ratio is conversion independent,29,34 activation energy in hydrogenation of diunsaturated acids toward cis and trans monoenic acids has similar values.

Figure 7. Hydrogenation of sunflower oil over 0.5 wt % Pd/ Sibunit catalysts in the fixed bed reactor operating in a semibatch mode (fraction size 2-3 mm, hydrogen pressure 0.6 MPa, catalyst mass 14.15 g, hydrogen flow rate 300 cm3/min, oil charge 50 g, and temperature 106 °C).

Fixed Bed Hydrogenation. Hydrogenation of edible oils over nickel and palladium catalysts has been mainly performed industrially using the concept of batch operation, introduced in industrial practice almost a century ago. Koetsier and Lok35 discussed the advantages and disadvantages of fixed bed technology for the hydrogenation of edible oils. It was stated that the use of a fixed bed is expected to be attractive only for full hydrogenation of adequately refined vegetable oils of significant annual capacity, as sufficiently long run lengths should be achieved. Even at high pressures of hydrogen, impact of internal diffusion could be substantial, since the catalyst particles should be large enough to avoid significant pressure drop. In essence, it means that the reaction would proceed only within a thin shell at the outer surface of the catalyst particle. It is thus important to prepare a catalyst when the impact of internal diffusion will be minimized. Preliminary experiments were performed in a fixed bed reactor operating in a semibatch mode (continuous flow of hydrogen only) over 0.5% Pd on Sibunit with various feedstocks. The results are summarized in Table 2. As can be seen, the specification requirements, imposed by industry, were achieved. Note that for the estimation of degree of CdC double bonds saturation the melting point of fatty acid hydrogenated products (MPFA) was determined. This value reflects the completeness of fatty acid hydrogenation (for example, complete CdC double bond saturation of free fatty acids (FFA) and rapeseed oil

7224 Ind. Eng. Chem. Res., Vol. 47, No. 19, 2008

Figure 8. Hydrogenation of free fatty acids (Table 1) over 0.5 and 1 wt % Pd/Sibunit catalysts in a continuous fixed bed reactor (fraction size 2-3 mm, hydrogen pressure 0.6 MPa, catalyst mass 14.15 g, hydrogen flow rate 300 cm3/min, and oil feed 10.0 g/h).

corresponds to 61 and 65 °C of MPFA, respectively). The composition of the final products is also illustrated in Table 2. For the sake of comparison, extended experimental data for selective hydrogenation (e.g., at milder temperature conditions) of sunflower oil are also presented in Figure 7. These data show the potential of egg-shell palladium on Sibunit to be utilized in fixed bed applications, when diunsaturated fatty acids should be selectively hydrogenated to monoenic ones. Time-on-stream behavior of 0.5 and 1% Pd/Sibunit in hydrogenation of free fatty acids in a continuous (both for hydrogen and oil) mode up-flow fixed bed reactor is depicted in Figure 8. If temperature increase is significant, as it was for 0.5% Pd/ Sibunit (from 170 to 225 °C during 70 h on stream), it can result in decarbonylation of fatty acids with formation of CO. At the same time, as it can be seen from Figure 8, that while activity of 0.5% Pd/Sibunit catalyst declined with time on stream in hydrogenation of free fatty acids, 1% Pd/Sibunit displayed very stable performance. In general, decarbonylation and decarboxylation reactions could be rather prominent over palladium catalysts.36,37 Besides olefins, which could serve as precursors for coke eventually leading to catalyst activity decline, CO is also known to be a poison for noble metals38 acting further in favor of catalyst deactivation. Industrial Tests. Due to the promising performance of Pd/ Sibunit catalysts in three-phase slurry reactors, they were tested in industrial conditions. The main aim in these tests was to selectively hydrogenate free fatty acids. Preliminary pilot tests at a margarine factory in Maslozhirprom, Moscow in sunflower oil hydrogenation over 1% Pd/Sibunit at 170 °C resulted in a product with a melting point of 43-45 °C, which did not contained diunsaturated (C18) acids, while 75% of monounsaturated (C18) acids and 25% of saturated (C16 + C18) acids were present. The concentration of trans products was in the range of 48-50%. Overall, 40 kg of margarine was produced. The catalyst 1% Pd/(Sibunit) charge was introduced in an industrial reactor of 6 m3 at a margarine plant in Lvov (Ukraine) for sunflower oil hydrogenation in the temperature range of 140-200 °C. Overall, 30 t of margarine was produced from 18 tons of the hydrogenated oil product. More comprehensive tests were carried out for sunflower oil hydrogenation at Zaporogskiy margarine plant (Ukraine) in a similar reactor of 6 m3 with 1 wt % Pd/Sibunit catalysts. Thirty batch operations were performed leading to 180 t of product with the total catalyst consumption of 60 kg. The melting point

of the product was in the range of 30.5-36.0 °C. It contained 4-17% diunsaturated (C18) acids, 57-80% monounsaturated (C18) acids, and 10-25% saturated (C16 + C18) ones. Complete hydrogenation of free fatty acids and rapeseed oil was also performed in an industrial upflow fixed bed reactor at Nefis Cosmetics (Kazan, Russia). The 1 wt % Pd/Sibunit was utilized. The total catalyst loading in the reactor was 1.2 t. The hydrogenation was carried out with the rapeseed or sunflower oil flow rate of 0.5-0.9 t/h in the temperature domain 180-230 °C. Conversion of the feedstock was in the range of 80-90%. The industrial-scale experiments performed at different conditions (temperature, hydrogen pressure and flow rate, oil residence time, feedstock) for the period of ∼500 h resulted in 320 t of product of commercial quality. Conclusions Catalytic hydrogenation of various vegetable oils was studied over Pd nanocomposite carbon (Sibunit) catalysts. In the case of selective hydrogenation of edible oils, the mesoporous structure of the carbon support provided high cis/trans ratio of monoenic products. The egg-shell catalyst, containing palladium on the same support, showed stable behavior in the continuous hydrogenations of free fatty acids and vegetable oils. Pd/ nanocomposite carbon catalyst demonstrated excellent behavior in industrial conditions for both three-phase slurry and up-flow fixed bed reactors being active and stable in respectively selective hydrogenation of edible and total hydrogenation of nonedible oils. Literature Cited (1) Bailey, A. E. Theory and mechanics of the hydrogenation of edible fats. J. Am. Oil Chem. Soc. 1949, 26, 596. (2) Albright, L. F. Quantitative measure of selectivity of hydrogenation of triglycerides. J. Am. Oil Chem. Soc 1965, 42, 250. (3) Veldsink, J. W.; Bouma, M. J.; Schoon, N. H.; Beenackers, A. A. C. M. Heterogeneous hydrogenation of vegetable oils: literature review. Catal. ReV. Sci. Eng. 1997, 39, 253. (4) Koetsier, W. T. Hydrogenation of edible oils, technology and applications. In Lipid Technologies and Applications; Gunstone, F. D., Padley, F. B.,. Eds.; Marcel Dekker, Inc.: New York, 1997; pp 265-303. (5) Ahmad, M. M.; Priestley, T. M.; Winterbottom, J. M. Palladiumcatalyzed hydrogenation of soybean oil. J. Am. Oil Chem. Soc. 1979, 56, 571. (6) Dijkstra, A. J. Revisiting the formation of trans isomers during partial hydrogenation of triacylglycerol oils. Eur. J. Lipid Sci. Technol. 2006, 108, 249.

Ind. Eng. Chem. Res., Vol. 47, No. 19, 2008 7225 (7) Ma¨ki-Arvela, P.; Kuusisto, J.; Sevilla, E. M.; Simakova, I.; Mikkola, J.-P.; Myllyoja, J.; Salmi, T.; Murzin, D. Yu. Catalytic hydrogenation of linoleic acid to stearic acid over different Pd and Ru supported catalysts. Appl. Catal., A. In press. (8) Mattil, K. F.; Norris, F. A.; Stirton, A. J.; Swern, D. Bailey’s Industrial Oil and Fat Products, 3 ed.; John Wiley and Sons; New York, 1964. (9) Semikolenov, V. A.; Simakova, I. L.; Sadovnichij, G. V. New catalyst for vegetable oil hydrogenation. Chem. Ind. (Moscow) 1996, N3, 40. (10) Grau, R. J.; Cassano, A. E.; Baltanas, M. A. Catalysts and network modeling in vegetable oil hydrogenation processes. Catal. ReV. Sci. Eng., 1988, 30, 1. (11) Grau, R. J.; Cassano, A. E.; Baltanas, M. A. Intrinsic selectivities and yields in the catalytic hydrogenation of polyenoic fatty acid methyl esters. J. Am. Oil Chem. Soc. 1990, 67, 226. (12) Savchenko, V. I.; Makaryan, I. A. Palladium catalyst for the production of pure margarine: catalyst and new non-filtration technology improve production and quality. Platinum Met. ReV. 1999, 43, 74. (13) Beenackers, A.A.C.M. Mass transfer in gas-liquid slurry reactors. Chem. Eng. Sci. 1993, 49, 3109. (14) Balakos, M. W.; Hernandez, E. E. Catalyst characteristics and performance in edible oil hydrogenation. Catal. Today 1997, 35, 415. (15) Murzin, D.; Salmi, T. Catalytic kinetics; Elsevier: Amsterdan, 2005. (16) Bechmann, H. J. Hydrogenation practice. J. Am. Oil Chem. Soc. 1983, 60, 282. (17) Belkacemi, K.; Boulmerka, A.; Arul, J.; Hamoudi, S. Hydrogenation of vegetable oils with minimum trans and saturated fatty acid formation over a new generation of Pd-catalyst. Top. Catal. 2006, 37, 113. (18) Hsu, N.; Dlosady, L. L.; Rubin, L. J. Catalytic behavior of palladium in the hydrogenation of edible oils. J. Am. Oil Chem. Soc. 1988, 65, 349. (19) Perez-Cadenas, A. F.; Kapteijn, F.; Zieverink, M. M. P.; Moulijn, J. A. Selective hydrogenation of fatty acid methyl esters over palladium on carbon-based monoliths. Catal. Today. 2007, 128, 13. (20) Yermakov, Yu.I.; Surovikin, V. F.; Plaksin, G. V.; Semikolenov, V. A.; Likholobov, V. A.; Chuvilin, A. L.; Bogdanov, S. V. New carbon materials as support for catalysts. React. Kinet. Catal. Lett. 1987, 33, 435. (21) Fenelonov, V. B.; Likholobov, V. A.; Derevyankin, A. Yu.; Mel’gunov, M. S. Porous carbon materials prepared from C1-C3 hydrocarbons. Catal. Today 1998, 42, 341. (22) Simonov, P. A.; Semikolenov, V. A.; Likholobov, V. A.; Boronin, A. I.; Yermakov, Yu. I. Palladium catalysts on carbon supports. Report 1. General peculiarities on H2PdCl4 adsorption over carbon supports. IzVestia AN USSR 1988, 12, 2719. (23) Semikolenov, V. A.; Lavrenko, S. P.; Zaikovskii, V. I. Formation of supported palladium particles in Pd/C catalysts. Kinet. Katal. 1994, 35, 622. (24) Harada, T.; Ikeda, S.; Niyazaki, M.; Sakata, T.; Mori, H.; Matsumara, M. A simple method for preparing highly active palladium

catalysts loaded on various carbon supports for liquid-phase oxidation and hydrogenation reactions. J. Molec. Catal., A 2007, 268, 59. (25) Card, R. J.; Schmitt, J. L.; Simpson, J. M. Palladium-carbon hydrogenolysis catalysts: The effect of preparation variables on catalytic activity. J. Catal. 1983, 79, 13. (26) Simonov, P. A.; Troitskii, S.Yu.; Likholobov, V. A. Preparation of the Pd/C catalysts: A molecular-level study of active site formation. Kinet. Catal. 2000, 41, 255. (27) Troitskii, S.Yu.; Chuvilin; A, L.; Bogdanov, S. V.; Moroz, E. M.; Likholobov, V. A. Structure of polynuclear palladium(II) hydroxocomplexes and mechanism of their adsorption by carbon materials. Russ. Chem. Bull. 1996, 45, 1296. (28) Pe´rez-Cadenas, A. F.; Zieverink, M.P.; Kapteijn, F.; Moulijn, J. A. Selective hydrogenation of fatty acid methyl esters on palladium catalysts supported on carbon-coated monoliths. Carbon 2006, 44, 173. (29) Deliy, I. V.; Maksimchuk, N. V.; Psaro, R.; Ravasio, N.; Dal Santo, V.; Recchia, S.; Paukshtis, E. A.; Golovin, A. V.; Semikolenov, V. A. Kinetic peculiarities of cis/trans methyl oleate formation during hydrogenation of methyl linoleate over Pd/MgO. Appl. Catal., A 2005, 279, 99. (30) Semikolenov, V. A.; Lavrenko, S. P.; Zaikovskii, V. I. Sintering of Pd particles on the surface of carbon supports in hydrogen. React. Kinet. Catal. Lett. 1993, 51, 507. (31) List, G. R. Processing and reformulation for nutrition labeling of trans fatty acids. Lipid Technol. 2004, 16, 173. (32) Stinglei, D. V.; Wrobel, R. J. The effect of trans-isomers on the physical properties of hydrogenated oils. J. Am. Oil Chem. Soc. 1961, 38, 201. (33) Bernas, A.; Kumar, N.; Ma¨ki-Arvela, P.; Kul’kova, N. V.; Holmbom, B.; Salmi, T.; Murzin, D. Yu. Isomerization of linoleic acid over supported metal catalysts. Appl. Catal., A 2003, 245, 257. (34) Murzin, D. Yu.; Simakova, I. L. Kinetic aspects of stereoselectivity in hydrogenation of fatty acids. J. Mol. Catal., A 2008, 286, 156. (35) Koetsier, W. T.; Lok, M. C. Trends in the development of edible oil hydrogenation catalysts. http://www.soci.org/SCI/publications/2001/pdf/ pb95.pdf, SCI lecture papers series, Publication 96/1998. (36) Snåre, M.; Kubickova, I.; Ma¨ki-Arvela, P.; Chichova, D.; Era¨nen, K.; Murzin, D. Yu. Catalytic deoxygenation of unsaturated renewable feedstocks for production of diesel fuel hydrocarbons. Fuel 2008, 87, 933. (37) Snåre, M.; Kubickova, I.; Ma¨ki-Arvela, P.; Era¨nen, K.; Murzin, D. Yu. Heterogeneous catalytic deoxygenation of stearic acid for production of biodiesel. Ind. Eng. Chem. Res. 2006, 45, 5708. (38) Ma¨ki-Arvela, P.; Kumar, N.; Era¨nen, K.; Salmi, T.; Murzin, D. Yu. Inverse temperature dependence due to catalyst deactivation in liquidphase citral hydrogenation over Pt/Al2O3. Chem. Eng. J. 2006, 122, 127.

ReceiVed for reView April 23, 2008 ReVised manuscript receiVed June 27, 2008 Accepted July 11, 2008 IE800663J