Synthesis of planar chiral ferrocenyl cyclopentadienyl chelate ligand precursors

Two families of planar chiral ferrocenyl cyclopentadienyl chelate ligands for use in ansa-half sandwich metallocene complexes of catalytically active transition metals are described. The first family was derived in 4–5 steps from an enzymatic resolution of 1-iodo-2-(methylalcohol)ferrocene and possesses a cyclopentadiene derivative [Cp(H) = 1-indenyl, 2-indenyl or Ph4Cp(H)] directly attached to the ferrocene ring with an adjacent vicinal tether CH2Z donor group (Z = OH, OMe, NHMe, NMe2 or PPh2). The second family was derived from a chiral auxiliary approach and has the donor group (Z = PPh2 or NMe2) attached directly to the ferrocene ring with an adjacent tether vicinal CH2Cp(H) group [Cp(H) = Cp(H), fluorenyl, 1-indenyl, Me4Cp(H) or Ph4Cp(H)]. 2013 Elsevier Ltd. All rights reserved.


Introduction
The development of ligand architecture has had a huge effect on the catalytic activity of some metals. Chelation of cyclopentadienyl ligands with a pendant donor atom to certain transition metals leads to ansa-half sandwich metallocene complexes (Fig. 1). This unusual ligand may confer enhanced catalytic activity, due in part to the inertness of the cyclopentadienyl ligand and the tuning available from the donor atom, the first example of which was documented for a-olefin polymerisation. 1 There are a number of reviews on the many examples of cyclopentadienyl metal complexes bearing pendant phosphorus/sulfur/arsenic, 2 oxygen 3 and amine 4 donors. A wide variety of metals have been used and the number of ansa-half sandwich metallocenes in the literature is ever increasing. However, there are relatively few examples of chiral ansa-half sandwich metallocene complexes being used in asymmetric catalysis. [5][6][7][8][9][10] Chirality is most often conferred on the complex by centres of chirality present between the chelating cyclopentadienyl and donor atom groups that help to create a chi-ral pocket around the metal centre. The success of these approaches has been limited and we postulate that a planar chiral ferrocene linker could project a more effective chiral environment around a catalytically active ansa-half sandwich metallocene.
The superiority of the planar chiral element in ferrocenyl derived ligands, above that of centres of chirality, for the transmission of asymmetry in chemical reactions has been well documented. [11][12][13][14][15] Purely planar chiral ferrocenes have been less well investigated, [16][17][18][19][20][21][22] possibly due to the challenges associated with their enantioselective syntheses compared to the multitude of diastereoselective approaches that start, for example, with the ubiquitous Ugi N,N-dimethyl-1-ferrocenylethylamine that lead to bidentate Josiphos type ligands. [23][24][25] In our own work, we have used the Snieckus approach of sparteine mediated enantioselective directed ortho metallation of N,N-diisopropyl ferrocenecarboxamide 26 to access a series of N-P, N-N, N-O and N-S planar chiral ferrocene chelate ligands (Fig. 2). 27,28 In addition we have also synthesised asymmetric chelate ligands via the sparteine mediated lithiation of the C-2 position of 1,2,3,4,5-pentamethylazaferrocene ( In extending these studies we wished to use a planar chiral ferrocene unit to control the asymmetry of a potential chelate ansahalf sandwich metallocene. We were aware of only one example of this type of asymmetric scaffold in the literature, but the two complexes were not catalytically active (Fig. 3). 30 Conceptually, in order to design catalysts for the stereocontrolled formation of new r-bonds to a pro-chiral molecule, we require a chiral metal possessing 4 quadrants of differing size (1, Scheme 1), with only one side of the metal available for binding. Upon coordination of a prochiral substrate to the available face of the metal, the least sterically congested complex 2 should result. With the prochiral face of the unsaturated system differentiated, addition of H-Nu (Nu = NR 2 , SiR 3 , BR 2 , CN), will form new stereocentres 3 in a controlled fashion. To harness the high configurational metal-ligand stability of the cyclopentadienyl ligand, and open up the metal centres coordination sphere, we set out to synthesise a new class of planar chiral ferrocenyl chelate ligands 4 containing a cyclopentadienyl group tethered to a donor ligand with the planar chiral ferrocene controlling the desired topology 5. The planar chiral ferrocene ligand links three quadrants, making the metal M chiral and differentiates which face of the metal is available for further coordination. The stabilising effect of the cyclopentadienyl (Cp) ring and donor atoms Z on the catalytically active metal centre M and any coordinated reactive intermediates or products may enhance catalytic transformations at the metal centre M. A variant of this ligand system 6 can also be envisaged.
Herein we report the syntheses of some enantiomerically pure chelate ligands 4, which rely upon an enzymatic resolution to produce either enantiomer of the desired ligands. We also report the synthesis of an example of ligand 6 which uses directed ortho-metallation methodology from a chiral acetal attached to ferrocene.

Results and discussion
Initially both the inden-1-yl and inden-2-yl groups were examined as it was thought that when the inden-1-yl group is bonded to a metal, the bulky ferrocene group should dictate that only one iso-mer is formed (isomers due to the axis of chirality in the inden-1-yl group) (Fig. 4). The atropisomerism associated with the inden-1-yl group may aid the selectivity of the metal catalyst. The symmetrical nature of the inden-2-yl functionality in metal complexes would allow this effect to be investigated.

Introduction of chirality
Our attempts to incorporate a cyclopentadienyl type ligand directly to one of the ferrocene rings using Snieckus' approach of directed ortho-metallation of ferrocenecarboxamides, 26 could be employed to synthesise the racemic ligands (using TMEDA and direct reaction with 1-indanone) but the asymmetric reaction using (À)-sparteine was unsuccessful. The synthesis of the corresponding iodide, and subsequent halogen-lithium exchange and reaction with 1-indanone allowed the incorporation of a Cp group. However, chemical manipulation of the resultant tertiary amide to useful functional groups proved problematic. This, coupled with the fact that the (À)-sparteine methodology formally delivers only one enantiomer 31 led us to consider an alternative approach that relied on halogen lithium exchange of an enatiomerically pure ferrocenyl halide. It was known that (±)-alcohol 7 can be resolved using a lipase enzyme (from Candida Antarctica) and vinyl acetate to give (pS)-acetate 8 (98% ee) and remaining (pR)-alcohol 7 (95% ee, >99% after recrystallisation from pet. ether). 32 In our hands the immobilised enzyme could be reused (at least 3 times) with no noticeable decrease in enantioselectivity and the reaction could be performed on at least a 50 g scale (Scheme 2). It was found that the quantity of enzyme and amount of DCM used could be greatly reduced from the procedure reported in the literature.  Resolution of (±)-7.

Cyclopentadiene type compounds with an oxygen donor group
The addition of both the 1-indenyl group (by halogen-lithium exchange followed by addition of 1-indanone) and the 2-indenyl groups (by Negishi cross coupling of 2-bromoindene) was found to give the highest yields when a non-coordinating substituent was present on the pseudo-benzylic position (Cp(H)CH 2 -) of the donor side chain. An array of compounds were synthesised where the hydroxyl/acetate group of 7/8 was converted into a range of donor substituents (N, O or P). The highest yields of the addition were seen when using TIPS ether 9 (Scheme 3). This may be because the less bulky groups are able to coordinate to lithium and make it less reactive and/or are too basic to react with ketones as well as inhibiting the lithium-zinc transmetallation needed for Negishi cross-coupling reactions. This was supported by the fact that no starting material remained at the end of the halogen lithium exchange reactions, with only the proton quenched material isolated in optimisation experiments. The addition of 1-indanone to (pR)-9, after halogen-lithium exchange of the iodide, gave the resultant alcohol (pR)-10 in 98% yield. Dehydration of the alcohol was achieved by conversion to the mesylate and addition of DIPEA to give (pS)-11 in 97% yield. The hindered TIPS silyl ether also proved to be the best substrate for the Negishi cross-coupling of 2-bromoindene to give (pS)-12 in 75% yield (Scheme 3). The 2indenyl substituted TIPS ether 12 proved to be much more stable than the corresponding 1-indenyl compound 11 under acidic conditions. This added stability may be due to the fact that in the 1indenyl ligands there is a doubly benzylic position between the aromatic ring of the indenyl group and the ferrocene Cp ring which may stabilise any incipient carbocation character. The tetraphenylcyclopentadiene group could be introduced by the reaction of tetraphenylcyclopentadienone with 9, after halogen-lithium exchange, to give alcohol 13 in 80% yield. Reduction using LiAlH 4 (2 equiv) gave (pS)-14, albeit in low yield (20%). A range of other reducing conditions were investigated but all led to degradation of the substrate. However, increasing the amount of LiAlH 4 used to 20 equiv surprisingly gave alcohol (pS)-15 in 60% along with (pS)-14 in 20% isolated yield.
All standard conditions for the removal of the TIPS group from the 1-indenyl compound, to give the alcohol, resulted in degradation or, under acidic conditions, substitution of the OTIPS group. 33 (pR)-10, 98%   No examples of this deprotection reaction for OTIPS ether at the aposition of a ferrocene compound exist in the literature. Both the methyl ether (pS)-16 (85% yield) using aqueous HCl and MeOH, and acetate, (pS)-17 (92% yield) using aqueous acetic acid in THF, were synthesised presumably by substitution of the -OTIPS group (Scheme 4). Subsequent basic hydrolysis of acetate 17 gave alcohol (pS)-18 (99%).

Cyclopentadiene type compounds with a nitrogen donor group
It was found that the 2-indenyl amine compounds could be directly synthesised by the addition of TMSCl to TIPS ethers and subsequent transfer of the reaction mixture onto an excess of the amine (Scheme 5). Both the dimethylamine 19 and methylamine 20 compounds were synthesised by this route. The direct addition of the amine to the reaction mixture led to the formation of dimers. The use of these conditions with the 1-indenyl TIPS ether 11 led to degradation, presumably due to the sensitivity of the 1-indenyl compounds to the formation of a doubly benzylic stabilised carbocation in the presence of an acid. Conversion of the alcohol into the mesylate and subsequent reaction with an excess of the amine gave the desired dimethylamino 21 and methylamino 22 compounds. The use of a hindered base such as DIPEA was essential as when less hindered bases such as triethylamine were used the corresponding quaternary amine salt was produced. Amine nucleophiles added in situ did not displace the triethylamine. The tetraphenylcyclopentadiene amine compounds 23 and 24 were synthesised in a similar manner from the alcohol 15, although these compounds could not be separated from small amounts of remaining starting material.

Cyclopentadiene type compounds with a phosphorus donor group
The most common phosphino ansa half sandwich metallocenes in the literature, with a similar tether length to our compounds, utilise a diphenylphosphine group to coordinate to the metal (Fig. 3). 2 The synthesis of the 1-indenyl phosphine compound was unsuccessful despite attempting a range of conditions previously employed for the synthesis of a-substituted ferrocene compounds including the conditions employed for the amino compounds; refluxing the acetate 17 in acetic acid with diphenylphosphine, 23 the use of BH 3 ÁPPh 2 H as a nucleophile, 34 via a quaternised amine 35 or via an a-carbocation. 36 Complete degradation or recovery of the starting material was observed in these cases. Similar results were seen in the attempted synthesis of the 2-indenyl phosphine compound; however, when diphenylphosphine was added to the quaternary amine salt 25, an unusual phosphine bridged ferrocene dimer 26 was observed (Scheme 6). It is unclear why the phosphonium cation forms in preference to the 3-coordinate diphenylphosphine. Only a handful of examples exist where a phosphonium cationic species is synthesised from diphenylphos-phine and 2 equiv of an electrophile but these require the presence of a base and in our case an excess of phosphine was used. 37,38 The synthesis of the tetraphenylcyclopentadiene compound proved much more straightforward and could be synthesised from the alcohol 15 using analogous conditions similar to the synthesis of the amine ligands (Scheme 7).

Donor group directly attached to the ferrocene Cp ring
In addition to the compounds above the donor group and Cp(H) group could be reversed, that is, the Cp(H) group is tethered to the ferrocene Cp ring with a CH 2 linker and the donor group is attached directly onto the ferrocene Cp ring (see 6, Scheme 1). There is only one example of a similar ligand in the literature (Fig. 3) where in addition to the planar chirality of the molecule, there is also a stereogenic centre at the a-position adjacent to the ferrocene Cp ring. 30 This ligand has been complexed to rhodium or ruthenium and used in a reconstitutive condensation reaction. The synthesis of an analogous compound without this stereogenic centre would potentially allow us to probe the effect of each stereochemical element.
Phosphinoferrocenyl aldehyde 28 is a known compound and can be accessed in either enantiopure form using Kagan's chiral auxiliary approach from ferrocene carboxaldehyde and 1,2,4butanetriol. 39,40 It was envisaged that our desired ligands, with a tethered Cp(H) group and donor group directly attached to the ferrocene ring, could be synthesised from 28. Bildstein et al. have previously shown that a Cp(H) group can be introduced at the aferrocenyl position using a condensation reaction between ferrocene carboxaldehyde and cyclopentadiene in the presence of pyrrolidine. 41 Utilisation of the Bildstein conditions to give 29 followed by a subsequent reduction of the fulvene using LiAlH 4 gave the desired diphenylphosphino ligand 30 as a mixture of Cp(H) alkene isomers (Scheme 8).
The use of this methodology for other Cp(H) derivatives needed extended reaction times and resulted in poor yields for the condensation reaction. It was thought that this may be due to the poor solubility of 28 in MeOH, the additional steric hindrance of other Cp(H) derivatives and the additional stability and hence the lower reactivity of the deprotonated Cp type anions. We developed an alternative procedure whereby the Cp(H) derivative was deprotonated using sodium hydride and the resultant anion heated at reflux with 28; reaction times varied for different analogues from 24 h to 3 d (Scheme 9). This allowed for the synthesis of fluorene 31, indene 32, tetramethylcyclopentadiene (TMCp(H)) 33 and tetraphenylcyclopentadiene (TPCp(H)) 34 type ligands.
In order to investigate the dimethylamino family of ligands, 2dimethylaminoferrocenyl aldehyde 37 was required. This compound was synthesised from Kagan's aminoacetal intermediate 35 40 using a reductive amination to give 36, prior to removal of the auxiliary (Scheme 10). As with phosphino aldehyde 28, the Cp(H) group could be installed by using a condensation between 37 and cyclopentadiene in the presence of pyrrolidine to give 38. Subsequent reduction of the fulvene using LiAlH 4 gave the desired dimethylamino ligand 39 as a mixture of Cp(H) alkene isomers (Scheme 10). Ligand 39 was synthesised in racemic form in order to demonstrate the applicability of our methodology to amino ligands in addition to the phosphino ligands. The use of enantiopure 1,2,4-butanetriol should allow for the synthesis of dimethylamino compounds in enantiopure form. 40

Conclusion
A number of potential ligands have been synthesised with a cyclopentadienyl group (1-indenyl, 2-indenyl and tetraphenylcyclopentadiene) and an additional tethered donor atom (O, N and P). The opposite ligands with a donor group (N, P) and tethered cyclopentadiene group (cyclopentadiene, 1-indenyl, TMCp(H), TPCp(H) and flourene) were also synthesised. Work is underway to complex these compounds to metals and screen them in an array of asymmetric reactions. It is hoped that the large ferrocene group and planar chirality of the complexes will provide great improvements to the current ansa-half sandwich metallocene type complexes used in catalysis.

General
Melting points were recorded on a Stuart Scientific SMP3 apparatus and are uncorrected. Optical rotations were recorded on a Jasco DIP370 Digital Polorimeter. Infrared spectra were recorded on a Perkin-Elmer 1600 FTIR instrument as solutions in chloroform unless otherwise stated. The NMR spectra were recorded on Bruker AM/Bruker AMX/Bruker AVANCE III spectrometers at 600, 500, 400 and 300 MHz in a solution in CDCl 3 unless otherwise stated. Chemical shifts are reported in ppm relative to CHCl 3 ( 1 H: 7.27), ( 13 C: 77.2). Coupling constants are reported in Hz and rounded to the nearest 0.1 Hz. Signal assignments are as follows: b (broad), s (singlet), d (doublet, t (triplet), q (quartet), m (multiplet), dd (doublet of doublets), etc. Mass spectra were recorded using Thermo Finnigan Mat900xp (EI/CI) VG-70se (FAB) and Waters LCT Premier XE (ES) instruments. Elemental analysis was performed on an Exeter Analytical Inc. EA440 horizontal load analyser or Hewlett Packard 1100 series. HPLC analysis was acquired on a Hewlett Packard Capillary HP4890A GC analyser.