Article pubs.acs.org/IECR
Kinetic Modeling of Ethyl Benzoylformate Enantioselective Hydrogenation over Pt/Al2O3 Gerson Martin,† Paï vi Mak̈ i-Arvela,† Johan War̈ nå,† Karoliina Honkala,‡ Dmitry Yu. Murzin,*,† and Tapio Salmi† †
Åbo Akademi University, Process Chemistry Centre, Laboratory of Industrial Chemistry and Reaction Engineering, FI-20500 Turku/Åbo, Finland ‡ Department of Physics, Nanoscience Center, University of Jyväskylä, P.O. Box 35, FIN-40014 Jyväskylä, Finland ABSTRACT: A kinetic model was developed for the enantioselective hydrogenation of ethyl benzoylformate (EBF) on a modified Pt/Al2O3 catalyst. This model was based on the assumption of different numbers of sites for the adsorption of carbonyl groups from EBF as well as for the modifier ((−)-cinchonidine) adsorbed in a flat (actor) and tilted (spectator) modes. Density functional theory calculations were applied to study the energetics of EBF adsorption on Pt(1 1 1) in order to estimate the number of adsorption sites needed. The hydrogenation rate constants were determined along with the adsorption parameters by nonlinear regression analysis. A comparison between the model and the experimental data revealed very good correspondence.
1. INTRODUCTION Today, asymmetric hydrogenation is mainly carried out by employing expensive homogeneous catalysts. However, these catalysts have their limitations in large scale industrial production, and therefore, new asymmetric catalysts need to be developed. Asymmetric synthesis using modified heterogeneous catalysts has gained a lot of interest in the production of optically pure chemicals in several applications, such as pharmaceuticals, nutraceuticals, fragrances, and agrochemicals. Modified heterogeneous catalysts are industrially interesting, because some available examples have demonstrated that high enantioselectivities (>95%) close or even exceeding those obtained with existing homogeneous catalysts for the same reaction can be achieved.1,2 Enantioselective hydrogenation of α-ketoesters on (−)-cinchonidine-modified Pt catalysts has been described previously using Langmuir−Hishelwood3,4 and Eley−Rideal type5 kinetic models. The most widely studied kinetic models have been proposed for ethyl pyruvate6,7 and 1-phenyl-1,2-propanedione.8 In contrast to ethyl pyruvate and 1-phenyl-1,2-propanedione, kinetic modeling for the hydrogenation of ethyl benzoylformate (EBF) has been scarcely studied. The reaction scheme is displayed in Figure 1. Experiments performed using ethyl pyruvate have shown that the Eley−Rideal mechanism can be ruled out with considerable certainty due to the saturation effects observed at a extremely low modifier concentration as well as at increasing substrate concentration.9 The major drawback of applying these models is the lack of information about the interactions between the modifier, catalyst surface, substrate, and products. Studies on reaction kinetics can provide valuable information about reaction mechanisms and in connection with modeling offer a tool that allows a comparison of plausible reaction mechanisms and can help in ruling out completely wrong mechanisms.10 The kinetic models suggested for enantioselective hydrogenation of α-ketoesters involve different basic assumptions concerning the active sites on which the enantiodifferentiation © 2014 American Chemical Society
Figure 1. Reaction scheme for the hydrogenation of EBF (left) and the catalyst modifier (M) (−)-cinchonidine (right). Reproduced with permission from ref 10. Copyright 2010 Elsevier.
takes place. Several studies have revealed that enantioselective hydrogenation proceeds through a two-step, two-cycle mechanism.3,11,12 According to this mechanism, it is assumed that active chiral sites are formed by adsorption of the cinchona modifier on the platinum surface. The substrate from the liquid phase adsorbs reversibly on these sites in its two configurations forming intermediate complexes, which upon hydrogenation afford the formation of (R)- and (S)-ethyl mandelate, respectively. Blaser et al.3 have suggested that for hydrogenation on the unmodified catalyst, leading to racemic product mixtures, addition of the first hydrogen atom is rate-determining, whereas in hydrogenation to the major enantiomer occurring on the modified chiral sites, the rate-determining step is the addition of the second hydrogen atom. Results from previous investigations have indicated that the modifier strongly affects the enantioselectivity and the initial hydrogenation rate.13 In the absence of the modifier, a racemic mixture of (R)- and (S)-ethyl mandelates is produced. Increasing the modifier amount can lead to a high enantiomeric excess and Received: May 28, 2014 Accepted: June 25, 2014 Published: June 25, 2014 11945
dx.doi.org/10.1021/ie502170y | Ind. Eng. Chem. Res. 2014, 53, 11945−11953
Industrial & Engineering Chemistry Research
Article
L, respectively. (−)-Cinchonidine was used as a catalyst modifier with a concentration between 0 and 0.16 mmol/L. 2.3. Analytical Procedure. Samples were withdrawn from the reactor at different time intervals and analyzed with a gas chromatograph (GC; Varian 3300) equipped with a chiral column (Silica Chiralsil-DEX; length, 25 m; diameter, 0.25 mm; film thickness, 0.25 μm). Helium was used as a carrier gas with a split ratio of 33. The detector (FI) and injector temperatures were 270 and 240 °C, respectively. The temperature program of the GC was 120 °C (25 min)−20 °C/min−190 °C (6 min). The GC analysis was calibrated with ethyl benzoylformate (Aldrich, 95%, 25,891-1), (R)-ethyl mandelate (Aldrich, 99%, 30,998-2), and (S)-ethyl mandelate (Aldrich, 99%, 30,997-4). 2.4. Density Functional Theory Calculations. Density functional theory (DFT) calculations were carried out for ethyl benzoylformate adsorption on a Pt(1 1 1) surface employing the GPAW17 implementation of the projected augmented wave method18 in real space grid. The grid spacing was 0.2 Å for all directions. The exchange and correlation functional was approximated with a BEEF-vdW19 formula that includes a description for dispersion effects. The platinum surface was modeled with a four-layer-thick slab which is periodic in x and y directions and has a 5 × 3 surface unit cell. All the other atomic positions were relaxed until residual force was below 0.05 eV/Å, but the bottom metal layer was fixed to its ideal bulk positions. The Brillouin zone was sampled using a 2 × 4 grid of k-points.
hydrogenation rates up to a certain optimum value of the modifier concentration; however, a further increase of the modifier concentration results in a decreasing enantioselectivity and the reaction rate.14 The modifier can adsorb on the surface in a parallel adsorption mode (actor) contributing to enantioselective hydrogenation as well as in a tilted mode,15 which is presumed to act just a spectator on the catalyst surface. The dependence of the enantioselectivity on the modifier concentration is explained by an optimum coverage of the actor species, which starts to decrease at high modifier concentration. At high modifier concentrations, the coverage of tilted spectator species on the catalyst surface increases, thus decreasing enantioselectivity and the reaction rate by blocking the catalyst surface. The different size requirements of reacting components, i.e. the modifier in tilted and parallel adsorption modes and the reactant, should be taken into account in the kinetic model by using a different number of adsorption sites. In the present paper, a kinetic model for the three-phase hydrogenation of ethyl benzoylformate (Figure 1) in the presence of Pt/Al2O3 and a dissolved catalyst modifier [(−)-cinchonidine] is presented. The main parameters are the modifier-to-surface Pt ratio and the hydrogen pressure. The novelty in this work is related to (a) generation of a new set of data for EBF in a broad range of modifier concentrations and hydrogen pressures, (b) extension of the previous model for hydrogenation of diketone to a ketoester, (c) treatment of hydrogen pressure effects on activity and enantioselectivity which was not considered previously, and (d) a clear demonstration that the model is able to explain a rather unusual enantiomeric increase with conversion. Explanations in the literature for such behavior are related to product−modifier interactions, doubtful from a physicochemical viewpoint. We have demonstrated that this phenomenon is related to changes of coverage and showed in this work how fractional coverage varies at different amounts of modification with time.
3. KINETIC RESULTS All reaction components A, B, and C in the scheme (Figure 1) were distinguishable by chromatographic analysis. The reaction products were formed concurrently, but in the presence of the chiral modifier, (−)-cinchonidine, (R)-ethyl mandelate is formed in excess. Typical hydrogenation kinetics of ethyl benzoylformate in toluene and with (−)-cinchonidine as a catalyst modifier are displayed in Figure 2.
2. EXPERIMENTAL SECTION 2.1. Chemicals. Ethyl benzoylformate (Aldrich, 95%, 25,8911) was treated with 220 mg of activated carbon (Chemviron Carbon, IVP, FE 51012A) for 2 h prior to its use in order to remove sulfur from the raw material. Hydrogen (AGA, 99.999%), toluene (J.T. Baker, 8077, >99.5%), and (−)-cinchonidine (Fluka, 27350, 98%) were used as received. 2.2. Hydrogenation of Ethyl Benzoylformate. Hydrogenation of ethyl benzoylformate to (R) and (S)-ethyl mandelate was carried out over 5% (w/w) Pt/Al2O3 (Aldrich, Strem, 781660). Experiments were performed in a glass reactor at atmospheric pressure with a volumetric flow rate of 295 mL/ min. The condenser was set at −25 °C to avoid evaporation of the solvent. The bulk volume was measured after 24 h of experiments, and in all the experiments, a reduction in the volume less than 5% was found. A mixture of H2/Ar at different concentrations 100%/0%, 75%/25%, 50%/50%, and 25%/75% was used to study the effect of hydrogen pressure. The temperature was kept constant in all experiments (25 °C). Toluene was used as a reaction solvent. The catalyst was prereduced under flowing hydrogen at 400 °C for 2 h. To avoid interactions between the catalyst and oxygen, the reaction medium was bubbled with Ar for 10 min and then with H2 before putting it in contact with the catalyst. The optimized stirring rate was 500 rpm, and the catalyst particle size of
