Novel catalyst and related hydrogenations

The present invention relates to a structured catalyst based on sintered metal fibers (SMF) coated by a ZnO layer impregnated with Pd-nanoparticles, reactions of organic starting material with hydrogen in the presence of said catalyst and vitamins, carotinoids, perfume ingredients, and/or food or feed ingredients prepared by using this reaction.

Selective catalytic hydrogenations of alkynols to alkenols are important processes in the fine chemicals industry. Pd-based catalysts are known to give the highest selectivity and yield. Preferential formation of olefinic alcohols is attributed to the stronger adsorption of acetylenic alchohols in comparison with the half-hydrogenation product. Catalytic performance of palladium is known to be strongly influenced by its dispersion, nature of support and the use of promoters and additives. Catalyst design taking into consideration these factors can allow the a yield increase of target product and catalyst reuse.

In general, Pd atoms of low coordination number present in small particles of 1-2 nm provide too strong alkynol adsorption diminishing turnover frequency and selectivity. This phenomena is known as a geometric or "ensemble" effect. Particles of 7-10 nm size demonstrate better catalytic performance in hydrogenations of 2-butyne-l,4-diol and 2-methyl-3-butyn-2-ol (MBY) as compared to highly dispersed Pd. For MBY hydrogenation, selectivity to 2-methyl-3-buten-2-ol (MBE) has been found to increase in the following order: Pd black < Pd/C < Pd/ Al ₂ O ₃ <

Pd/BaSO ₄ < Pd/MgO < Pd/ZnO ≤ Pd/CaCO ₃ . Basic supports are preferable also in terms of activity. Formation of alkenol can be further enhanced by addition of a second metal as a promoter. The promoting effect of co-metal in alkynol hydrogenations has been shown for Pb, Mn, Cu, Bi, Sn, Au, Zn, Cd, etc. Modified electronic and geometric properties of palladium affect the adsorption of alkynols and alkenols and suppress β-PdH phase formation known to catalyze

direct hydrogenation of alkynol to the saturated alcohol. Moreover, the alkenol yield increases considerably in the presence of additives to the reaction mixture such as ammonia, quinoline, pyridine and sulfur compounds. The mechanism of their influence on activity/selectivity is still under discussion.

In industry acetylenic alcohol hydrogenations are carried out in the stirred tank reactors with Lindlar catalyst, 5% Pd/CaCO ₃ modified by lead acetate and often with addition of quinoline.

Lindlar catalyst being a fine powder is difficult to handle and requires filtration after the reaction. In this respect, structured catalysts are beneficial for process intensification and safety. Monoliths, membranes, metallic grids, bidimensional glass and carbon were used as catalyst supports in liquid-phase hydrogenations. Monoliths showed similar selectivity but much lower activity per Pd loading in comparison with a slurry catalyst in 3-methyl-l-pentyn-3-ol and 2- butyne-l,4-diol hydrogenations. The use of highly-selective bimetallic Pd-Ru (9:1) H ₂ -permeable membrane in 2-methyl-3-butyn-2-ol hydrogenation is limited by the high content of noble metal and low productivity per gram of Pd. Metallic grids have a disadvantage of low geometric surface area of ~ 100 cm ² /g. Fabrics of activated carbon fibers used in 2-butyne-l,4-diol hydrogenation possess low mechanical strength. Therefore, there is a need for a structured catalyst which may overcome limitations of the above-mentioned materials.

This work aims at the development of a structured catalyst for hydrogenation of 2-methyl-3- butyn-2-ol (MBY) to 2-methyl-3-buten-2-ol (MBE, Fig. 1), which is an intermediate in the synthesis of vitamins A and E and perfumes. Three-dimensional sintered metal filters (SMF) consisting of metallic microfibers were chosen as a structured catalyst support. SMF have high thermal conductivity that is a great advantage in exothermic hydrogenations, high porosity and permeability. The metal fiber matrix also acts as a micron-scale static mixer eliminating channeling. In addition, high mechanical strength, chemical and thermal stability, easy shaping make SMF promising materials for intensification of catalytic hydrogenation.

SMF were coated with a thin layer of ZnO known as efficient support for 2-methyl-3-butyn-2-ol hydrogenation. Pd nanoparticles were deposited from the beforehand prepared sol, and the material was heated in hydrogen atmosphere to create PdZn intermediates. ZnO layer acts both as

a basic support and a Pd promoter. The PαVZnO/SMF material was tested for mechanical stability, and its catalytic behavior was studied in MBY hydrogenation.

The term "structured catalyst" as used herein refers to catalysts wherein the spatial position of the catalyst is controlled. Structured catalysts are known in the art, see, e.g., Chimia 56(4), 2002, 159-163. Examples of structured catalysts are ceramic carrier constructions and fibrous structures, especially filamentous woven cloths. All types of filamentous woven cloths can be used for use in the present invention. The fibers may be from organic or inorganic matter. Examples are fabrics from activated carbon fibers such as acrylonitril fibers, glass libers, ceramic fibers, metal fibers or fleece composite oxides of activated carbon fibers. Preferred are polyacrylonitril fabrics. The individual fibers of the filamentous woven cloth preferably have a diameter of about 2 μm to about 100 μm, especially a diameter of no more than about 20 μm. The fabrics are suitably be woven from threads consisting of a boundle of individual fibers, providing a porous size of the woven cloth of less than about 1 mm. They may be chemically treated, e.g., with nitric acid to modify the specific surface and may have a coating, e.g. of metals such as Al, Ti or Pb to improve selectivity.

Examples of alternatives to Pd nanoparticles include nanoparticles of noble metals such as platinum, iridium, rhodium, ruthenium or combinations thereof. The catalyst may be present on the carrier fabric in an amount up to about 10 mass%, suitably 1-10 mass%. The loading of the carrier fabric is accomplished by treating with a solution of a precursor of the catalyst, e.g. a salt of the catalyst metal and subsequent drying and heating in a hydrogen atmosphere and can be controlled by the concentration of the catalyst precursor in the loading solution.

The hydrogenation in accordance with the present invention can be carried out under conditions conventionally used for hydrogenations of 2-butin-2,4-diol to produce 2-butene-2,4-diol. Suitably, the hydrogenation is carried out at a pressure of about 0.1 to about 6 MPa and at a temperature of about 350K to about 500K. The hydrogenation can be carried out batch wise or in continuous mode.

The following Example illustrates the invention further without limiting it.

Example

2.1. Materials

SMF (Southwest Screens & Filters SA, Belgium) made of FeCrAl alloy fibers (Cr 20%, Al 4.75%, Y 0.27%, other elements ~ 1-2%, Fe balance) in the form of a uniform pore panel (0.29 mm thickness, 71% porosity, 20μ fiber thickness, 675 g/m ² ) were used as a structured support.

Zinc acetate dihydrate (puriss. p. a., >99.5%), monoethanolamine (ethanolamine, puriss. p. a., ≥99%), acetoin (3-hydroxy-2-butanone, purum, mixture of monomer and dimer), sodium molydbate dihydrate (puriss. p. a., >99%), palladium (II) chloride anhydrous (purum), nitric acid (puriss., p.a.), ethanol (purum, >99.8%), toluene (puriss. p. a., >99.7%), 2-propanol (puriss. p. a., >99.8%) were supplied by Fluka. Acetone (puriss., >99%) was purchased from Riedel-deHaen. Lindlar catalyst (5% Pd content as confirmed by AAS) was a gift of DSM Nutritional AG (Switzerland).

2-Methyl-3-butyn-2-ol (purum, >99%), 2-methyl-3-buten-2-ol (purum, >97%), 2-methyl-2- butanol (purum, >98%), 1-butanol (purum, >99.5%), and quinoline (purum, >97%) were purchased from Fluka and used as received. Hydrogen (>99.99% purity) was from Carbagas, Switzerland. Demineralized-bidistilled water was used throughout this work.

2.2. Preparation ofPd/ZnO/SMF catalyst (0.2 wt.% Pd {3 wt.% to ZnOj, 6 wt.% ZnO)

In order to remove contaminations the SMF panels were degreased with acetone, boiled in toluene for 0.5 h and air-dried. To improve further adhesion of ZnO, SMF were oxidized in air at 1373 K for 3 h to create an α- Al ₂ O ₃ surface layer. Such a temperature (1373-1473 K) for the treatment of FeCrAl alloy with little rare earth content is known to lead to the formation of a

structured alumina film, characterized by equiaxed grains on the outer surface, while lower temperature treatment gives oxide whiskers.

ZnO film was prepared by sol-gel method using zinc acetate dihydrate (0.3 M in isopropanol) and solubility enhancement additives (monoethanolamine MEA and acetoin AIN) at a molar ratio of MEA: AIN: Zn = 1 :0.5: 1. The additives were mixed in a solvent prior to the addition of zinc acetate with the help of ultrasound, the sol was colored reddish brown due to reaction between MEA and AIN yielding imine HO-CH(CH ₃ )-C(CH ₃ )=N-C ₂ H ₄ -OH. Coating was conducted by a dip-coating procedure. The gel film thus obtained was air-dried at 383 K for 10 min and then heated at 873 K for 30 min. Rapid heating was applied which is known to result in the formation of highly oriented crystals (slow heating, on the other hand, gives plicated structures). The coating-heating procedure was repeated 7 times to give the weight gain of 6 wt%. ZnO. The coating then was post-annealed at 1173 K for 15 min to promote the formation of island structure of ZnO grains with the increased the specific surface area.

Pd sol was prepared as described in via dissolution of PdCl ₂ in a boiling aqueous solution of sodium molybdate at a molar ratio Mo:Pd=1.2, followed by hydrogenation for 30 min at room temperature. After wet impregnation for 1 h, ZnO/SMF panels were washed copiously with water, dried at ambient conditions and subjected to high-temperature treatment in hydrogen atmosphere (H ₂ :Ar=l:9, total flow of 450 mL/mn, 2 h at room temperature, 10 7min to 773 K, hold for 2 h and cooled at the same flow). Catalyst was stored at ambient conditions.

2.3. Catalysts characterization

Pd and Zn amounts after dissolution in hot nitric acid were determined by atomic absorption spectroscopy via Shimadzu AA-6650 spectrometer with an air-acetylene flame. ZnO loading was also determined gravimetrically.

The BET specific surface area and pores size distribution (PSD) of the support and the catalyst were determined using a Sorptomatic 1990 (Carlo Erba) instrument via N ₂ adsorption- desorption at 77 K. PSD calculation was performed by Dollimore/Heal method.

The ultrasonic adherence test for the mechanical stability of the catalyst was carried out using an ultrasonic bath (Bransonic ultrasonic cleaner, Branson Ultrasonic Corp., USA). The catalyst was treated in water for 20 min totally, and after each 5 min the material was dried at 393 K and weighed.

The surface morphology of the samples was investigated by scanning electron microscopy SEM, using a JSM-6300F, JEOL. XRD analysis was carried out in a Siemens D 500 diffractometer using CuKa radiation. The spectra were recorded in a rapid scanning mode (4.0 s/step, 20 step size of 0.04°) in a 20 range of 30-50°.

2.4. Hydrogenation experiments

Hydrogenations were carried out in a batch stainless steel reactor (250 mL autoclave, Buchi AG, Uster, Switzerland) equipped with a heating jacket and a hydrogen supply system. The structured catalyst was placed between two metal gauzes (2 x 8.5 cm) fixed on the self-gassing hollow shaft stirrer. For the reaction with the powdered Lindlar catalyst, 8-blade disk turbine impeller was used. At the working temperature the reactor was filled with the reaction mixture and the catalyst, flushed with Ar and kept for 5 min under stirring to equalize the temperature. Then the reactor was flushed with hydrogen and pressurized. During the course of the reaction, the pressure in the reactor was maintained constant.

The experiments were carried out at 308 K and 5 bar H ₂ pressure at intensive stirring of 2000 rpm (if not mentioned otherwise). Demineralized-bidistilled water was used as a reaction medium containing 0.1 M of 2-methyl-3-butyn-2-ol in total volume of 200 mL. Catalyst quantity corresponding to 0.9 mg of Pd was used giving the substrate-to-palladium molar ratio of 2400. In some experiments quinoline was added to the reaction medium.

The samples of the reaction mixtures were periodically withdrawn from the reactor via a sampling tube and analyzed by GC. The GC analysis was performed using Auto System XL (Perkin Elmer) equipped with a 30 m Stabilwax (Crossbond Carbowax-PEG, Restek, USA) 0.32 mm capillary column with a 0.25 μ coating. The carrier gas (He) pressure was 101 kPa. Injector and FID temperatures were 473 K and 523 K, respectively. The oven temperature was hold for 4 min at 323 K, then increased to 473 K at a ramp of 30 7min. The GC analysis conditions allowed detecting also the dimerized by-products formed during the reaction. 1-Butanol was used as an internal standard (0.74 g), thus, actual quantity of the reaction mixture components could be determined.

MBE yield was defined as Y = IπOI _MBY (converted into MBE) / πIOI _MBY (introduced) ^• 100%, MBY conversion as X = πIOI _MBY (reacted) / πIOI _MBY (introduced) ^■ 100%, selectivity to MBE as S = Y / X - 100%.

2.5. Catalyst reuse and regeneration

Between the hydrogenation runs, catalyst was washed with water, dried at ambient conditions and stored in vacuum desiccator. Two regeneration procedures were applied: 1 ) calcination at 773 K for 2 h with subsequent reduction in hydrogen atmosphere at 773 K for 2 h as described in 2.2; 2) the catalyst was placed in a beaker with ethanol, ultrasonically treated for 10 min in the ultrasonic bath and washed with water.

Characterization ofPd/ZnO/SMF

BET specific surface area of the synthesized material was found to be ~ 0.7 m ² /g with a pore specific volume of 6- 10 ^"3 cm ³ /g and mean pore radius of 1.2 nm. This value is in agreement with the surface area of SMF of < 1 m ² /g and of ZnO powder (4 m ² /g).

SEM microphotographs of the calcined SMF, ZnO/SMF and Pd/ZnO/SMF panels are presented in Figs. 2a-c. The calcined SMF microfibres (~ 22 μ diameter) are fully coated with the grain-

structured ZnO layer (~ 1.5 μ thickness), which remains intact after Pd deposition and the temperature treatment. Neither detached ZnO grains, nor large aggregates at the points of fibres contact were observed in the coated samples. Coating of SMF with a thin catalytically active layer is believed to retain all advantages of this 3D material, especially its high permeability and low pressure drop.

The adherence test for the mechanical stability of ZnO layer was performed via ultrasonic treatment of the ZnO/SMF in water. Cumulative weight loss of ZnO after 20 min treatment was 0.25 wt% with respect to ZnO weight. Thus, the synthesized material possesses high mechanical stability and resistance in water media. This adherence could be attributed to the formation of a thin layer of mixed oxides between FeCrAl alloy and ZnO at 1173 K, like ZnAl ₂ O ₄ or ZnFe ₂ O ₄ . Oxide layer formation has been also reported for ZnO-coated ZrO ₂ and SiO ₂ .

XRD pattern of the calcined SMF, ZnO/SMF and Pd/ZnO/SMF materials are presented in Figs.3a-c. Calcination of FeCrAl alloy SMF filters resulted in the formation of the surface alumina layer in agreement with the reported data. Characteristic peaks of crystalline ZnO were revealed in the samples after ZnO deposition, while in the Pd-containing material these peaks were of lower intensity due to the partial ZnO reduction and formation OfPdZn ₂ alloy (2θ of

41.2°). It is known that high-temperature reduction of Pd supported on ZnO induces strong metal- support interaction and formation of various PdZn intermediates.

3.2. Catalytic performance ofPd/ZnO/SMF catalyst with and without quinoline addition

Concentration profiles of MBY and products vs. reaction time are presented in Fig. 4. MBY concentration decreases concomitant with the MBE concentration increase. MBA appears at the very beginning of the reaction, confirming the presence of a parallel path b (Fig. 1) of the direct MBY hydrogenation to MBA. Once MBY is consumed, the rate of MBA formation drops to zero. Therefore, for MBY hydrogenation over Pd/ZnO/SMF catalyst path d of MBE hydrogenation to MBA is suppressed contrary to monometallic Pd and Pd-Pb-catalysts. This is

ascribed to the presence of Pd-Zn-phase, which is known to decrease hydrogen and olefin adsorption strength providing high selectivity in hydrogenation reactions. Besides, Zn being an electron donor with respect to Pd, modifies its electronic properties and changes the alkynol/alkenol relative adsorption strength. The maximum MBE yield for Pd/ZnO/SMF catalyst was achieved as 94.5%. At this reaction time MBA and dimers concentrations were 3.9 mM and 2.3 mM, respectively. Dimers formation stops when all MBY as has been already reported for MBY hydrogenation.

Catalytic behavior of Pd/ZnO/SMF catalyst was compared to the one of industrially used Lindlar catalyst. Both catalysts exhibited initial selectivity of 97%. The initial activity of Pd/ZnO/SMF catalyst was 0.71 mol _MBY /molp _d /s at 308 K and 5 bar H ₂ pressure. This value is an order of magnitude higher than the one of Lindlar catalyst (0.04 mol _MB γ/molp _d /s).

In order to improve the MBE yield in the hydrogenations with the structured catalyst, quinoline was added to the reaction mixtures. The effect of the nitrogen organic bases is explained as i) preferential adsorption of the additive compared to alkenol, i.e., poisoning (site blocking) of the less selective sites, ii) decreasing the catalyst activity that provides extra time for the mass- transfer processes in the catalyst particles and higher selectivity, iii) electron donation from the N-atom to Pd and change the alkynol/alkenol relative adsorption strength ("ligand" effect), iv) formation of Pd ^δ+ -H ^δ" mode on Pd surface resulting in the preferred attack of the nucleophilic H on the triple bond, or v) catalyst surface rearrangement. The nitrogen base-to-Pd molar ratio used in hydrogenation reactions varies from 2 to ~ 1500, being 7 for the original Lindlar work.

The influence of quinoline concentration on the catalytic performance of Pd/ZnO/SMF catalyst was studied in the range of quinoline-to-Pd molar ratio (Q/Pd) of 7í480. The results are summarized in Fig. 5 and Table 1. Increase of Q/Pd ratio from 0 to 26 increases both selectivity and activity, and time to achieve the highest MBE yield diminishes by 6-fold. Hence, in this range quinoline reveals its "ligand" effect, i.e., increasing the catalyst activity and selectivity by electron donation from N to Pd. Further increase of quinoline concentration leads to the further improve in selectivity, but the reactions are almost blocked at 98% and 73% conversion at

Q/Pd=120 and 480, respectively. Probably, at high concentration quinoline blocks the active sites. Thus, optimum Q/Pd ratio found for MBY hydrogenation over Pd/ZnO/SMF catalyst is 26, increasing the selectivity and yield by ~ 1 % and the catalyst activity by 6-fold in comparison with a quinoline-free reaction. Besides, quinoline addition changes the by-products distribution. MBA- to-dimers molar ratio at the point of maximum MBE yield decreases from ~2 for Q/Pd = 0í26 to 1 for Q/Pd = 120í480 (Table 1 ).

It is known that the control of selectivity can be provided not only by catalyst modification but also via variation of the process parameters. MBY hydrogenation was carried out at 3 bar H ₂ pressure, 303 K and Pd/ZnO/SMF loading corresponding to 0.6 mg Pd (against 5 bar, 308 K and 0.9 mg Pd for the previous experiments). Initial reaction rate was found to be 0.51 mol _MB γ/moWs. Selectivity at 95% conversion and maximum yield were 96.5% and 96.2%, respectively. Low hydrogen pressure is known to be beneficial for selective hydrogenation of alkynes and particularly of acetylenic alcohols carried out in the absence of external diffusion limitations. This can be explained by the enhanced formation of β-PdH phase at high hydrogen pressures, favoring direct alkyne hydrogenation to alkane.

3.3. Catalyst reuse

Pd/ZnO/SMF catalyst shows improved catalytic performance compared to the industrially used Lindlar catalyst and does not require filtration after the reaction. Moreover, the use of metallic filters as support allows design of a bubble column staged with the catalyst layers for three-phase hydrogenation. These catalysts may be recycled without significant deactivation and/or may be regenerated. Therefore, we performed reuse of the 2-month aged Pd/ZnO/SMF catalyst and the results are presented in Table 2.

Zero leaching of Pd and < 0.5 wt% leaching of Zn to the reaction mixtures were found via AAS. After storage the catalyst showed higher activity (No. 1 and 2, Table 2) but the selectivity and yield of MBE decreased by 1% due to the increased formation of both MBA and dimers. XRD

analysis of the stored catalyst (Fig. 3d) showed a decrease in the intensity of a PdZn ₂ phase peak. Pd-Zn interactions are known to take place during the storage of Pd/ZnO materials. As less Pd is involved in PdZn ₂ phase, the activity increases.

Reuse of the aged catalyst with intermediate washing with water showed progressive decrease in selectivity and MBE yield (No. 2, 3, Table 2). After 3 runs the catalyst was regenerated via the calcination-reduction procedure (see subchapter 2.5), which in some cases is known to fully restore the catalytic performance of Pd catalysts in hydrogenation reactions. However, it did not improve yield significantly (No. 4, Table 2) and decreased the activity. Calcination-reduction treatment of Pd/ZnO catalysts is known to affect the state of metallic and bimetallic phases. During calcination PdZn alloy is decomposed back to Pd crystallites and ZnO, which form PdZn alloy again during the following reduction. In order to distinguish between these morphological changes and the removal of carbonaceous deposits during regeneration, the stored catalyst (as in No. 2, Table 2) was subjected to the same calcination-reduction procedure. XRD analysis (Fig. 3e) showed again the presence of PdZn ₂ phase but selectivity, activity and yield were lower than the ones of the fresh catalyst (Nos. 1 and 5, Table 2). In both cases (Nos. 4 and 5) regeneration decreased the MBA formation attaining the value of fresh catalyst (No. 1 , Table 2). However, dimers formation increased and the catalyst activity decreased.

No. 6 of Table 2 shows the efficient catalyst regeneration via ultrasonic treatment in ethanol for 10 min (no Pd was detected afterwards in ethanol by AAS). As can be seen, selectivity and MBE yield after the ultrasonic regeneration were fully restored to the level of the fresh catalyst, however, the activity decreased twice. MBA-to-dimers molar ratio after regeneration changed from ~2 to ~1. Ultrasonic treatment is known to clean the surface of heterogeneous catalysts from carbonaceous deposits and provides morphological changes leading to the improved reusability. Therefore, ultrasonic regeneration is suitable allowing the yield of the target product for the recycled catalyst as high as the yield over the fresh one.

Quinoline addition to the reaction mixture (quinoline/Pd=26 mol/mol) increased activity ~ 6 fold with the MBE yield of 95.3%, while its higher amount decreased activity by an order of magnitude with ~2% gain in selectivity.

The Pd/ZnO/SMF catalyst possesses high mechanical stability and can be easily shaped for the structured reactors. No significant metals leaching occurred during the reactions. Catalyst storage on air and subsequent multiple reuse showed decrease of yield to MBE, but the catalyst could be regenerated fully by ultrasonic treatment.

Table 1. Influence of quinoline (Q) addition on the catalytic behavior of fresh Pd/ZnO/SMF catalyst (for the reaction conditions see Fig. 4)

at maximum MBE yield; at 98% conversion as no further hydrogenation has occurred; ^c at 73% conversion as no further hydrogenation has occurred

Table 2. Catalytic performance of the stored and regenerated Pd/ZnO/SMF catalyst in multiple runs (for the reaction conditions see Fig. 4, for the treatment conditions - subchapters 2.2 and 2.5)

¹ at maximum MBE yield