Enhancing the Enantioselectivity of Novel Homogeneous Organometallic Hydrogenation Catalysts
2003; Wiley; Volume: 42; Issue: 36 Linguagem: Inglês
10.1002/anie.200250861
ISSN1521-3773
AutoresMatthew D. Jones, Robert Raja, John Meurig Thomas, Brian F. G. Johnson, Dewi W. Lewis, Jacques Rouzaud, Kenneth D. M. Harris,
Tópico(s)Catalysis for Biomass Conversion
ResumoBy designing well-defined, single-site asymmetric organometallic complexes anchored to a concave siliceous support (see picture) high ee values may be obtained in heterogeneous catalytic hydrogenations. O red, Si orange, Rh yellow, C grey, N blue, F green, B pink, H white. The exigent need to develop new asymmetric hydrogenation catalysts is universally acknowledged, and many new feasible strategies for the design and synthesis of such catalysts continue to be proposed.1–5 Herein, in addition to describing a new set of efficient, diamino-type ligands for a central rhodium or palladium ion that leads to good enantioselective (ee) performance, homogeneously, we also show that significant improvement in the stereoselectivity of the organometallic catalyst may be achieved by heterogenizing it at the inner walls of a mesoporous silica so that advantage is taken of the spatial restrictions imposed by the concave surface at which we have located the active center. Comparatively few reports have hitherto been published in which RhI or PdII complexes without phosphane ligands have been used to activate hydrogen, but a rapidly growing number employing nitrogen containing ligands has appeared of late for the purpose of asymmetric catalysis.6 One of the incentives for this change is the rather unstable nature and cost of many chiral phosphanes.6b It is not, however, the nitrogen-containing ligands that are the key feature of this work. Rather it is the fact that advantage is taken of a concave surface of a silica support, on which the homogeneous organometallic catalyst is tethered, thereby boosting the enantioselectivity to a value higher than that obtained from the same catalyst and the same tether attached to a convex silica support. The catalysts themselves are pseudo square-planar rhodium(I) or palladium(II) complexes of the bidentate amines: (S)-(−)-2-aminomethyl-1-ethyl pyrrolidine 1 a and (1R,2R)-(+)-1,2-diphenylethylenediamine 1 b, which may readily act as ligands to the metal center. The RhI catalysts are bonded either to 1,5-cyclooctadiene (COD) or norbornadiene (NBD), whereas for the palladium species we used the allyl (C3H5) group as the coordinating alkene. The BF4− salts of these organometallic cations are readily handled, air-stable, crystalline materials amenable to X-ray analysis.7 For the cations based on 1 a (paired in each case with the BF4− anion; see Scheme 1) X-ray structural analysis shows that the square-planar rhodium(I) or palladium(II) center has high steric hindrance. Access to it by a relatively bulky incoming reactant is spatially restricted (see below), and this feature is essential for good enantioselective catalytic behavior. The ligands used in this study. A fragment of the crystal structure of the RhI complex with ligand 1 a is shown in Figure 1 a, from which it is seen that the BF4− anion is hydrogen-bonded to the nitrogen of the amino groups. a) Fragment of the crystal structure of the RhI complex of 1 a illustrating the complex hydrogen-bonded network; the b axis is labeled. b) A molecular diagram of the heterogenized single site catalyst. By using described procedures8–12 whereby 3-bromopropyl trichlorosilane serves as a means of anchoring isolated organometallic species to the inner walls of mesoporous silica, we may functionalize an ion-pair of all ligands 1 a and 1 b complexed to metal so as to yield their respective single-site catalysts, as depicted in Figure 1 b. We have anchored ligand 1 b to a mesoporous silica (pore diameter 30) and to a nonporous silica (known as cabosil). This permits the comparison of catalytic performance of the organometallic species when it is either at a convex or at a concave siliceous surface (Figure 2). (It is much easier to prepare concave than convex silica surfaces. Since this work began, we have studied a series of silicas in which the pore diameters range from a low of 38 Å to a high of 250 Å; but, to date, the only satisfactory, well-defined convex silica surface that we have been able to investigate is that of the commercial material known as cabosil). Graphical model (to scale) showing the constraints of the catalyst once anchored on a) MCM-41 (concave) and b) cabosil (convex) surfaces. We chose as a test reaction the simple, well-known12 case of the hydrogenation of E-α-phenylcinnamic acid (Scheme 2) the predominant product being the 2,3-diphenyl propionic acid. We have previously examined12e this heterogeneous asymmetric hydrogenation using a Pd diphenylphosphino-ferrocenyl (dppf), tethered catalysts, in which, depending upon the manner of production of the heterogenized catalyst, variable amounts of other products (2,2- and 3,3-diphenylpropionic acids) as well as variable values of ee of the 2,3-diphenyl propionic acid were obtained. The asymmetric hydrogenation reactions under investigation. Catalytic tests were carried out (see Experimental Section) both on the new RhI homogeneous catalysts and on their anchored analogues. The homogeneous and heterogeneous palladium complex of ligand 1 a and 1 b where also tested (the crystal structure of the homogeneous catalyst is given as supplementary information). The results are summarized in Table 1. It is to be noted that, in the repeat cycle of experiments (for the case of the heterogenized organometallic catalyst), the ee values as well as the conversions and selectivities do not drop significantly in the first and second recycle compared with the respective values for the parent catalyst. This fact, as well as the observation that, compared with the corresponding homogeneous catalyst, there is a significant enhancement of the ee value upon heterogenization suggests that the original catalytically active species is the Rh-diamine-diene complex, unlike the situation that seems to prevail in the case of Rh-diphosphane analogues.13 Amine and diene Counter-ion Catalyst Reaction Metal t [h] Conv. [%] Sel [%]. ee [%] -1 BF4− homogeneous α-phenyl cinnamic acid RhI 24 74 77 93 heterogeneous 24 80 74 96 BF4− homogeneous methyl benzoyl formate RhI 24 95 58 0 heterogeneous 0.25 85 92 92 1 98 94 91 24 99 79 80 -1 BF4− homogeneous α-phenyl cinnamic acid RhI 24 88 76 64 heterogeneous 24 99 66 91 BF4− homogeneous methyl benzoyl formate RhI 0.5 85 70 18 homogeneous 20 °C 0.5 56 59 69 heterogeneous 2 100 98 99 -1 BF4− homogeneous α-phenyl cinnamic acid PdII 24 100 87 76 heterogeneous 24 93 87 93 BF4− homogeneous methyl benzoyl formate PdII 2 100 94 0 heterogeneous 1 96 97 87 2 97 97 87 -1 BF4− homogeneous α-phenyl cinnamic acid RhI 24 57 84 81 heterogeneous MCM-41 24 98 80 93 1st recycle 24 90 86 90 2nd recycle 24 89 84 90 cabosil 24 95 77 79 -1 BF4− homogeneous α-phenyl cinnamic acid PdII 24 95 82 79 heterogeneous 24 75 100 88 Whilst the asymmetric hydrogenation of E-α-phenylcinnamic acid was studied to establish proof of principle and to compare with our earlier results12e by using tethered Pd-dppf catalysts, we have also investigated another reactant, methyl benzoylformate, that is of considerable commercial value. In this case also the enantioselectivity of the heterogenized organometallic catalyst, for the production of methyl mandelate, at the concave silica surface is much enhanced, (see Table 1) compared to its homogeneous analogue. The improvements in the enantioselective performance of the catalyst clearly demonstrate the benefit of anchoring the catalysts inside the concave surfaces of mesoporous silica. Full details for the reactions are given in recently filed patents.14 It is seen that the enantioselectivity achieved with the anchored variants of the catalysts is superior to that achieved with their homogeneous counterparts, (Table 1). Moreover, further evidence to support the view that it is the concavity (and hence restricted access to the active site by the relatively bulky reactants) that is crucial as a determinant of the enhancement of the ee value is borne out by the observed diminished degree of enantioselectivity when a catalyst is anchored to a convex siliceous surface (see Table 1). In the postulated computer graphic models shown (to scale) in Figure 3 a and Figure 3 b it may be seen that access of the phenylcinnamic acid to the RhI ion is favored only when the reactant approaches the active site along the axis of the pore (Figure 3 a), whereas no such restrictions exist in the case of the same catalyst anchored at the convex surface.15 Color portrayal of the anchored RhI-organometallic -diene, as-prepared catalyst, thus illustrating the hindered access to the metal center because the incoming reactant must approach parallel to the channel (a). When rotation about the CC bond in the tether occurs so that the RhI center faces the wall of the pore, as in (b), access to the active site is no longer possible. Whilst it is feasible that some of the diene may be jettisoned during or shortly after hydrogenation, it is not likely that all of it is jettisoned in the original catalyst (as is generally the case in analogous homogeneous catalytic situations) as (I) the repeat cycles (see Table 1) show that catalytic performance is sustained in progressive experiments, and (II) there is a significant enhancement in ee upon anchoring the organometallic catalyst (see text). In previous, similar experiments, on allylic amination, involving catalysts located at the walls of mesoporous silica (e.g., Pd((S)-1[(R)-1′,2-bis(diphenylphosphino)ferrocenyl]-ethyl-N,N′-dimethylethylene-diamine)Cl2),8, 9, 12e it was the constraints imposed by the space surrounding the metal center (active site) that dominated the enantioselectivity. In the case presented herein, there is a subtle difference; it is the restricted access generated by the concavity of the pore that is the principal determinant. Even though the "capping" cyclooctadiene may be lost during or just prior to the act of catalytic turnover at the RhI center, access to the latter of the reactant is nevertheless a key factor, and this is why concavity enhances the catalytic performance observed.15 To confirm that the anchored catalysts are truly heterogeneous and that no complications arise from leaching of the organometallic moiety during use, we performed a hot filtration and analyzed the resulting product mixtures (by inductively coupled plasma atomic emission spectroscopy; ICPAES). In a typical reaction at 313 K and 20 bar of H2 pressure ( Table 1), the solid catalyst was filtered from the reaction mixture after 8 h, and the reaction continued for a further 16 h. There was no appreciable change in the conversion of the E-α-phenylcinnamic acid during the 16 h period after removal of the solid catalyst, and the product selectivities and enantiomeric excess value were also unaffected. These facts rule out the possibility that colloidal Rh(Pd) or Rh(Pd) black are formed as, if they were, no longer would we observe an enantiomeric excess or essentially unchanged product selectivities. It is not an easy task to extend the comparison between asymmetric organometallic catalysts tethered on to a range of concave/convex pairs because it is difficult to prepare nonporous convex silicas (like the commercially available cabosil). We have, however, obtained results that confirm the validity of our arguments by using three distinctly different concave silicas.16 All compounds were purchased from Aldrich, unless otherwise stated. The solvents were all predried by using standard methods. The reactions were carried out under an atmosphere of nitrogen by using standard Schlenk line techniques. All experimental procedures as well as the crystal structures are given in the supporting information, a typical procedure is detailed below. The NMR were performed on a Bruker DPX 500 MHz employing cryo-probe technology. [Rh(1,5-cyclooctadiene)(1R,2R)-(+)-1,2-diphenylethylenediamine ]BF4: [RhCl(COD)]2 (64 mg, 0.13 mmol) was dissolved in THF (10 mL) to which AgBF4 (50 mg, 0.26 mmol) was added, this solution was stirred for 1 hour at room temperature. The solution was then filtered (to remove AgCl) and (1R, 2R)-(+)-1,2-diphenylethylenediamine (50 mg, 0.26 mmol) was added to the filtrate and the solution left to stir for 1 hour. The resulting solution was slightly concentrated in vacuo and hexane (25 mL) added to precipitate the required product. The product was isolated by filtration and washed with hexane (2×20 mL) and diethyl ether (2×20 mL) and dried in vacuo giving a yellow powder (185 mg, 90 %). Elemental analysis calcd (%) for C22H28N2RhBF4: C 51.76, H 5.50, N 5.50; found C 51.44, H 5.57, H 5.29. 1H NMR (CD3OD): δ=1.95 (br m, CH2 4 H), 2.46 (br m, CH2, 4 H), 4.01 (s, NCH, 2 H), 4.24 (m, CH, 2 H), 4.35 (m, CH, 2 H), 7.1–7.3 ppm (m, Ph, 10 H). 13C NMR (CD3OD): δ=31.6 (CH2), 66.3 (NCH) 81.4 (CH), 128.5, 129.2, 129.6, 140.5 ppm (Ph). ESI (positive ion): 423 [M+]. Activation of MCM-41: To dry calcined MCM-41 (BET surface area 880 m2 g−1, determined by N2 adsorption; 2.0 g) in THF (15 mL) dichlorodiphenylsilane (0.48 g, 0.19 mmol) was added and stirred at room temperature for 1 hour (Ph2SiCl2) reacts with external Si-OH moieties[10–11] and ensures all proceeding silane species reacts at the internal surface of MCM-41). The solution was then cooled to −78 °C and 3-bromopropyltrichlorosilane (1.10 g, 4.8 mmol) was added. This mixture was allowed to slowly warm up to room temperature and stirred for a further 8 h. The resulting solution was heated at 50 °C for 1 hour. The MCM-41 was filtered and purified by soxhlet extraction with THF and finally dried in vacuo. Elemental analysis found (%): C 4.70, H 0.87, Br 2.14. 29Si CP MAS NMR indicated significant loss of SiOH moieties indicating grafting had occurred, the XRD of the resulting MCM-41 was analogous to the pure MCM-41 showing no loss of structure occurred. Tethering with (1R,2R)-(+)-1,2-diphenylethylenediamine: The activated MCM-41 (500 mg) was heated to reflux in THF (15 mL) together with the amine (150 mg) for 24 h. After this time the MCM-41 was filtered and purified by soxhlet extraction with THF and finally dried in vacuo. Elemental analysis found (%): C 9.72, H 1.92, N 0.15, Br 1.30. IR =3205 cm−1 broad (weak), NH stretch. Powder XRD indicated no loss in the structure of the mesopore during this process. Addition of rhodium: In a Schlenk [RhCl(COD)]2 (30 mg, 0.06 mmol) and AgBF4 (30 mg, 0.12 mmol) were dissolved in THF (10 mL) and stirred for 1 hour. The solution was filtered and the tethered MCM-41 was added to the filtrate and stirred for 4 h during which time the solution became pale and the solid became yellow. This pale yellow solid was filtered and washed with THF (4×20 mL), CH2Cl2 (4×20 mL) and dried in-vacuo. 3 b: C 11.23, H 2.02, N 0.12, Br 1.22. 19F MAS NMR indicated the presence of BF4− δ=−152.0 ppm (reference CFCl3), in solution δ=−154 ppm. Powder XRD only showed peaks that correspond to MCM-41 implying that the rhodium is highly dispersed inside MCM-41. Catalysis procedure: The catalytic reactions (enantioselective hydrogenations) were carried out in a high-pressure stainless-steel catalytic reactor (100 mL) lined with poly ether ether ketone (PEEK). The homogeneous catalyst (10 mg or 50 mg of the mesoporous silica anchored heterogeneous catalyst), which was previously evacuated and stored under inert conditions (nitrogen or argon), was transferred under an inert atmosphere to the catalyst delivery unit, which was subsequently sealed and introduced to the high-pressure reactor (by using dry helium). The substrate {(E)-α-phenyl cinnamic acid or methyl benzoylformate} (0.5 g), solvent (methanol, 30 g) and a suitable internal standard (cyclododecane) were then introduced into the reactor and the reactor was sealed. The reactor as well as the inlet and outlet ports were purged with dry nitrogen (three times) prior to the introduction of hydrogen (5 bar). A leak test was carried out and the reactor was then pressurized to 20 bar with hydrogen. The reactor was then heated to the desired temperature (313 K) and the contents were stirred with a mechanical stirrer at 400 rpm. (In the cases in which kinetic and rate effects were studied, a minirobot liquid-sampling valve was employed to remove small aliquots of the sample without perturbing the pressure in the reactor). At the end of the reaction, the heating was turned off the contents of the reactor were cooled (quenched). The reactor was then depressurized. A mass-balance calculation was made at this stage to check for handling and weight losses. The products were analyzed (with a suitable internal standard) by gas chromatography (GC; Varian, Model 3400 CX) employing a HP-1 capillary column (25 m×0.32 mm) and flame ionization detector with a variable ramp temperature program from 433 K to 493 K, for cinnamic acid. The identities of the products were first confirmed by using authenticated standards and their individual response factors were determined with a suitable internal standard (calibration method). The conversions and selectivities were determined as defined by the following equations and the yields were normalized with respect to the response factors obtained as above: Conversion %=[(moles of initial substrate−moles of residual substrate)/(moles of initial substrate)]×100 Selectivity %=[(moles of individual product)/(moles of total products)]×100 The products were further identified by NMR spectroscopy (1H and 13C in CD3OD) which showed the loss of the CC and the gain of a CC moiety for cinnamic acid and for methyl benzoyl formate a loss of a CO and a gain of a CO environment. The enantioselectivity values (ee) (for cinnamic acid) were first determined by GC (Hewlett Packard {HP 5890}) with a chiral column {gamma-cyclodextrin dialkyl (Chiraldex) (20 m×0.25 mm)} and the identities of the chiral products were further confirmed by LC–MS (Shimadzu LCMS-QP8000) equipped with a chiral detector (OR-990, JASCO), again, with a chiral column (CHIRALCEL OJ-R, DAICEL). Enantiomeric excess values were calculated from the peak areas of the enantiomers by using the following formula: ee=([R] − [S])×100/([R] + [S]) Finally, as a further confirmation, the products were isolated from the reaction mixture and esterified to the methyl ester by the following procedure: Boron trifluoride (14 %, 2 mL) in methanol was added to a glass vial containing the sample to be esterified (300 μL), the vial was sealed with a teflon-lined stopper and heated for 1 h at 353 K. The sample was cooled to room temperature and Milli-Q purified water (2 mL) was added with mild shaking. HPLC grade dichloromethane (2 mL) was added before the analysis, which was again performed on the LC–MS (Shimadzu LCMS-QP8000) with the same chiral column (CHIRALCEL OJ-R, DAICEL). The ee values for methyl mandelate were determined by Bayer AG Leverkusen Germany.14 Since this work was carried out we have further tested and explored17 a new method, described by Rege et al.,18 of anchoring homogeneous catalysts on to high-area supports through the use of the triflate anion. For that purpose two entirely new ligands and (for comparison) two of the ones reported here were used to produce the RhI-based asymmetric hydrogenation catalysts. This method is very convenient and it also produces superior heterogeneous asymmetric catalysts than their homogeneous analogues. Supporting information for this article is available on the WWW under http://www.wiley-vch.de/contents/jc_2002/2003/z50861_s.pdf or from the author. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Referência(s)