A series of diphosphines of the novel Walphos ligand family all based on a phenylferrocenylethyl backbone were synthesised in a four-step sequence. In the rhodium-or ruthenium-catalysed asymmetric hydrogenation of olefins and ketones enantioselectivities of up to 95% and 97%, respectively, were obtained. A 2-isopropylcinnamic acid derivative of industrial interest was hydrogenated in 95% ee and with turnover numbers of > 5000.Keywords: alkene reduction; asymmetric catalysis; asymmetric hydrogenation; ketone reduction; Pligands; rhodium; ruthenium Over a period of more than three decades homogenous enantioselective hydrogenation has been investigated extensively by numerous researchers in academia and industry, and is now being considered as a mature methodology for the production of enantiopure, bioactive ingredients and fine chemicals on an industrial scale.[1] The most efficient catalysts for the asymmetric hydrogenation of olefins, ketones or imines [2] are rhodium, ruthenium and iridium complexes of chiral diphosphine ligands. However, even though innumerable chiral diphosphines have been designed and investigated in the past, only a few out of more than thousands have been found suitable for industrial processes. Representative examples are biaryl-, phospholaneand ferrocenyl-based ligands like binap, duphos or josiphos type ferrocenes. [3] In our search for novel classes of diphosphines, we focused on the design of ligands that fulfil all relevant prerequisites particularly with regard to industrial applications: shortness and modularity of ligand synthesis that should allow an efficient fine tuning of catalysts× properties. In addition, such ligands should be readily accessible from enantiopure key intermediates. In this context, we have examined the potential of a new family with a novel phenylferrocenylethyl backbone that we named Walphos (1; Figure 1). The synthesis concept for this ligand family is straightforward: (i) the enantiomerically pure ligand framework ± an orthobromophenylferrocenylethylamine (3) ± is constructed from Ugi×s amine [4] and (ii) the final functional groups are implemented stepwise.[5] (Scheme 1) In this preliminary contribution, we report the synthesis of six representatives of the Walphos family with varying phosphino substituents R 1 and R 2 , together with first applications in the enantioselective hydrogenation.Starting from amine 2, six derivatives with electronrich and electron-withdrawing phosphino substituents R 1 and R 2 were prepared in a four-step sequence. In a Negishi coupling reaction [6] of (R c )-N,N-dimethyl-1-ferrocenylethylamine, (R c )-2, with 2-bromoiodobenzene the enantiomerically pure key intermediate (R c ,R p )-3 was built up. A subsequent lithiation of this bromide followed by trapping with the appropriate electrophile ±either chlorodiphenylphosphine or chlorobis(3,5-dimethyl-4-methoxyphenyl)phosphine ±result-ed in the formation of the corresponding aminophosphines. In order to prevent a ring closure reaction in the next step, the aminophosphin...

(Aminoferrocenyl)phosphine ligands 2-(1-(dimethylamino)ethyl)-1-(diphenylphosphino)ferrocene (PPFA) and [η5-cyclopentadienyl][η5-4-(endo-dimethylamino)-3-(diphenylphosphino)-4,5,6,7-tetrahydro-1H-indenyl]iron(II) (PTFA), were used as ligands in palladium(0) and -(II) complexes. The reaction of Pd2(dba)3·CHCl3 with PPFA or PTFA in the presence of the electron-withdrawing olefins maleic anhydride (MA) and dimethyl fumarate (DMFU) gave the complexes Pd(PTFA)(DMFU) (2), Pd(PPFA)(MA) (3), and Pd(PPFA)(DMFU) (4). Allylic complexes [Pd(η3-2-Me-C3H4)(PTFA)]Tf (5) and [Pd(η3-2-Me-C3H4)(PPFA)]Tf (6) (Tf = triflate) were obtained by reaction of PTFA or PPFA with [Pd(η3-2-Me-C3H4)Cl]2 in the presence of AgTf. In solution all these compounds exist as mixtures of two diastereomers, with either the alkene or the allyl group differently oriented with respect to the aminophosphine ligand. 1H NMR variable-temperature studies have been carried out for 2−6 and for Pd(PTFA)(MA) (1). Rotation of the alkene was observed for complexes 1−4 on the NMR time scale. ΔG ⧧ c has been calculated and values between 57.6 kJ mol -1 (298 K) and 76.6 kJ mol -1 (373 K) have been obtained. A Pd−N bond rupture which interchanges the two amino methyl groups is observed (ΔG ⧧ 328 = 63.9 kJ mol-1 to ΔG ⧧ 368 = 74.9 kJ mol-1) for derivatives of PPFA, but not for complexes containing PTFA. An EXSY experiment carried out on complex 5 has evidenced a selective η3−η1−η3 (carbon cis to phosphorus) allyl isomerization. Molecular structures of 4 and 6 were determined by X-ray structure analysis.

The ortho-deprotonation of halide-substituted ferrocenes by treatment with lithium tetramethylpiperidide (LiTMP) has been investigated. Iodo-, bromo-, and chloro-substituted ferrocenes were easily deprotonated adjacent to the halide substituents. The synthetic applicability of this reaction was, however, limited by the fact that, depending on the temperature and the degree of halide substitution, scrambling of both iodo and bromo substituents at the ferrocene core took place. Iodoferrocenes could not be transformed selectively into ortho-substituted iodoferrocenes since, in the presence of LiTMP, the iodo substituents scrambled efficiently even at −78 °C, and this process had occurred before electrophiles had been added. Bromoferrocene and certain monobromo-substituted derivatives, however, could be efficiently ortho-deprotonated at low temperature and reacted with a number of electrophiles to afford 1,2- and 1,2,3-substituted ferrocene derivatives. For example, 2-bromo-1-iodoferrocene was synthesized by ortho-deprotonation of bromoferrocene and reaction with the electrophiles diiodoethane and diiodotetrafluoroethane, respectively. In this and related cases the iodide scrambling process and further product deprotonation due to the excess LiTMP could be suppressed efficiently by running the reaction at low temperature and in inverse mode. In contrast to the low-temperature process, at room temperature bromo substituents in bromoferrocenes scrambled in the presence of LiTMP. Chloro- and 1,2-dichloroferrocene could be ortho-deprotonated selectively, but in neither case was scrambling of a chloro substituent observed. As a further application of this ortho-deprotonation reaction, a route for the synthesis of 1,3-disubstituted ferrocenes was developed. 1,3-Diiodoferrocene was accessible from bromoferrocene in four steps. On a multigram scale an overall yield of 41% was achieved. 1,3-Diiodoferrocene was further transformed into symmetrically 1,3-disubstituted ferrocenes (1,3-R2Fc; R = CHO, COOEt, CN, CH=CH2).

O-Methylephedrine was identified as a very efficient chiral auxiliary for ortho-lithiation reactions of ferrocenes. (1R,2S)-O-Methylephedrine [CH(3)NHCH(CH(3))CH(Ph)OCH(3)] was reacted with N-ferrocenylmethyl-N,N,N-trimethylammonium iodide [FcCH(2)N(CH(3))(3)I; Fc = ferrocenyl] to give (1R,2S)-N-ferrocenylmethyl-O-methylephedrine. Treatment of this compound with t-BuLi in pentane followed by quenching with the electrophiles iodine, dibromotetrafluoroethane, chlorodiphenylphosphine or benzophenone gave 2-substituted ferrocenes in 98% de and with the (R(p))-ferrocene configuration. Subsequently, the chiral auxiliary could be replaced by systems including dimethylamine, acetate, diaryl- or dialkylphosphines to give a number of enantiopure bifunctional 1,2-disubstituted ferrocene derivatives such as (R(p))-N-2-iodo- or (R(p))-N-2-bromoferrocenylmethyldimethylamine or (R(p))-2-acetoxymethyl-1-diphenylphosphinoferrocene. As an application, ferrocenyl diphosphines possessing a planar (R(p))-ferrocene configuration only [1,2-(PPh(2))FcCH(2)PR(2), R = Cy, Ph, [3,5-(CF(3))(2)Ph]] were synthesized in three steps from O-methylephedrine and N-ferrocenylmethyl-N,N,N-trimethylammonium iodide in up to 77% overall yield.

No abstract

Eleven novel aminophosphine ligands have been synthesized, all of which contain a chiral 2,2' '-bridged biferroceno unit as part of a biferrocenoazepine substructure. The efficiency of these compounds as chiral auxiliaries in palladium-mediated allylic substitution reactions has been investigated. Depending on the degree of (steric) fit between proper ligands and cyclic or noncyclic substrates, reactions with 46-87% ee were achieved. The molecular structure of a palladium dichloride complex of one of the ligands was determined by X-ray diffraction and compared to its binaphthyl analogue. In the solid state, the azepine substructure of these two complexes adopts totally different conformations with either local C(2) (binaphthyl) or local C(1) (biferrocene derivative) symmetry. These structural changes are well-reproduced by empirical force field calculations and are also reflected in significantly different behavior in asymmetric catalysis.

The (aminoferrocenyl)phosphine ligand 1-diphenylphosphino-2,1‘-(1-dimethylaminopropanediyl)ferrocene, 1, was used to synthesize new palladium(0) and -(II) complexes. The reaction of Pd2(dba)3·CHCl3 with 1 in the presence of the electron-withdrawing olefins dimethylfumarate (DMFU) and maleic anhydride (MA) gave the new complexes Pd(1)(DMFU) (2) and Pd(1)(MA) (3). The allyl complex [Pd(η3-2-Me-C3H4)(1)]Tf (4) was obtained from the reaction of 1 with [Pd(η3-2-Me-C3H4)(Cl)]2 in the presence of AgTf. In solution all these compounds exist as mixtures of two diastereomers, with either the alkene or the allyl group differently oriented with respect to the aminophosphine ligand. The orientation of these ligands in the major isomers has been determined by means of NOEs. Alkene rotation takes place in complexes 2 and 3 with free energies of activation ΔG 340 ⧧ = 73.1 kJ mol-1 (2) and ΔG 368 ⧧ = 79.9 kJ mol-1 (3), respectively. These barriers are compared with those of some analogous ferrocenyl aminophosphine ligands, PPFA (2-(1-dimethylaminoethyl)-1-diphenylphosphinoferrocene) and PTFA (1-diphenylphosphino-2,3-endo-(α-dimethylamino)tetramethyleneferrocene) complexes. For 3, the alkene rotation leads to isomer interconversion, while the observed isomerization of 2 must proceed via an olefin face exchange. Some experiments in relation to the nature of this process are discussed. Starting from PdRR‘L‘ precursors and 1 or PPFA, other Pd(II) derivatives of formulas PdRR‘(1), R = Cl, R‘ = Me, L‘ = cod, 5; R = R‘ = Me, L‘ = tmeda, 6; R = R‘ = C6F5, L‘ = cod, 7, or PdClMe(PPFA), 8, were prepared. For 5 and 8, only the isomers with the methyl group trans to nitrogen were obtained. The Pd−N bond rupture in the new complexes of 1 and in similar derivatives of PPFA and PTFA has been analyzed by variable-temperature 1H NMR studies, and in some cases a line shape analysis has been carried out. The influence of the ferrocenyl aminophosphine and ancillary ligands as well as of the oxidation state of the palladium center on this process is discussed. The molecular structures of both rotamers of 3, present in the same crystal, were determined by X-ray structure analysis.

Chiral, non-racemic 1,2,3-trisubstituted ferrocene derivatives are accessible from monosubstituted ferrocenes through two sequential ortho-deprotonation reactions; removal of the central substituent gives 1,3-disubstituted ferrocenes.