Iron cyclopentadienone complexes derived from C 2 -symmetric bis-propargylic alcohols; preparation and applications to catalysis †

A series of complexes containing the iron-cyclopentadienone structure were prepared by cyclising bis-propargylic alcohols and their derivatives with iron pentacarbonyl. The resulting complexes were characterised and tested in the catalysis of ketone reduction and alcohol oxidation. The complexes are competent catalysts for ketone reduction and alcohol oxidations.


Introduction
There is a growing interest in the application of iron-based complexes for the catalysis of asymmetric transformations, primarily due to its relative low cost and toxicity relative to more commonly used catalysts based on precious metals. 1 Iron cyclopentadienone complexes (general structure 1) have recently emerged as promising reagents for hydrogenation reactions and for hydrogen transfer processes. [2][3][4][5][6][7][8][9][10][11][12][13] Key to their application in this capacity is the formation of the derived hydrides of general structure 2, which can be achieved in situ using a number of activating agents, or through formation and isolation of the hydride prior to use. Complex 2 can transfer two atoms of hydrogen to an acceptor such as a ketone or imine and in doing so is converted to the unsaturated form 3. Complex 2 can be regenerated from 3 using a reducing agent such as formic acid or an alcohol (in the case of asymmetric transfer hydrogenation -ATH) or hydrogen gas (as in the case of asymmetric pressure hydrogenation -APH). The hydrogen transfers are believed to take place through a cyclic transition state as depicted in Fig. 1. 13 The complexes have also been used in alkyne and alkene reductions, using a derivative in which the OH bond on the Cp ring is modified. 14 Iron cyclopentadienone complexes were reported in detail by Schrauzer in 1959, through the cyclisation of two alkynes with an iron carbonyl complex, 2a and this approach remains the most common one. 2-13 Intramolecular cyclisation of a 1,7-dialkyne can be used to form complexes such as 4a and 4b. The formation and isolation of the iron hydride 5 was reported by Knölker in 1999, 3b and this hydride was demonstrated in 2007 by Casey and Guan, to be an effective catalyst for the hydrogenation of a series of ketones. 4 The iron hydride species can also be generated in situ from an iron tricarbonyl complex using an activating agent such as Me 3 NO (TMAO) or a close derivative, 15 or K 2 CO 3 (in an aqueous environment) 7b or through the use of photoactivation, in the presence of a hydrogen source. 5 Apart from reductions, the hydrogen-transfer properties of the catalyst have allowed it to be applied to what can be regarded as formal oxidations of alcohols (using acetone as a hydrogen acceptor), 9 reductive amination, 10 'hydrogen borrowing' reactions in which a C-N bond is formed between an alcohol and an amine 12a-c and to conversion of an amide to a nitrile. 12d Table 1 summarises recent applications of the catalysts, together with the catalyst derivatives commonly employed.
Very few examples of asymmetric modifications of iron (cyclopentadienone) complexes have been published to date. Berkessel et al. reported the use of complex 6, in which one CO ligand was replaced by the chiral monodonor ligand MONOPHOS, thus rendering the iron hydride chiral and offering the potential for this asymmetry to be relayed in the hydrogen-transfer step. 5 Using this catalyst, products with up to 32% ee were obtained for acetophenone reduction. We reported the application of the asymmetric complex 7 and close derivatives to ketone reduction by transfer hydrogenation, with a maximum ee of 25%. 6 In a recent example Gennari et al. created a cleverly designed catalyst 8 derived from a binapthyl-scaffold, and this gave the highest asymmetric inductions so far recorded for ketone reduction with this class of complex; up to 77% in one case. 8 In this paper we describe the synthesis and applications of a series of asym-metric iron(cyclopentadienone) complexes derived from C 2 -symmetric diols, which represent a new derivative class of these complexes.

Results and discussion
In earlier unrelated work, we demonstrated an efficient route to the synthesis of diol 12 in high enantioselectivity through the asymmetric reduction of the precursor diketone. 16 Treatment of 12 with Fe(CO) 5 (130°C, 20 h, sealed tube) resulted in the formation of complex 13 in 76% yield. The breaking of the C 2 -symmetry of the substrate was clear in the product, with distinct signals in the 1 H-NMR spectrum observed for each of the methine protons adjacent to the hydroxyl groups. In addition a racemic/meso mixture of diols was prepared by reduction of the precursor diketone with NaBH 4 . When this was cyclised in the same manner, the peaks of the racemic product could be observed but in addition the presence of the meso product was confirmed by the observation of extra methine resonances (ESI †). A complex with a hydroxyl group in the fused ring has been reported by Pearson et al. 2c In addition, it was possible to prepare each of the O-protected ethers 14-17 and to convert these to the corresponding complexes 18-21 in the same manner (Scheme 1). Each of the catalysts were characterised by NMR, MS and IR analysis, following purification by chromatography on silica gel, to which the complexes are stable. As expected, characteristic peaks were observed in the 1 H-NMR spectrum for the diastereoisomeric protecting groups. In addition, taking the OBn and OTBDMS derivatives as substrates, it was possible to substitute one CO for MONOPHOS using the TMAO-activated process described by Berkessel. 5 Both diastereomeric combinations of the resulting complexes 22-25 were formed, i.e. derived from both enan-tiomers of MONOPHOS, in order to examine the possibility of matched and mismatched stereocontrol in subsequent applications.
Although complexes 22 and 23 appeared to be single isomers, each diastereoisomer of complexes 24 and 25 bearing OTBDMS groups were found to contain a ca. 10% of the other diastereoisomer. This suggested that either racemisation of the BINOL component during the preparation of MONOPHOS or during its subsequent complexation had taken place. Racemisation of the diether ligand prior to or during complexation seems unlikely as this would have been expected to lead to formation of some of the meso isomer, but this was not observed. The lack of diastereoisomeric impurities in 22 and 23 may be Scheme 1 Synthesis of iron(cyclopentadienone) catalysts. Reagents and conditions: (i) NaH, THF, BnBr, tBu 4 NI, rt, 24 h (for 14); TBDMSCl, Imidazole, DMF, o/n, rt (for 15); TIPSCl, imidazole, DMF; o/n, rt (for 16); TBDPSCl, imidazole, DMF, o/n, rt (for 17). (ii) 3.0 eq. Fe(CO) 5 , toluene, 130°C, 20 h. (iii) MONOPHOS (2.0 eq.), TMAO (2.0 eq.), toluene, 60°C, o/n. the result of purification from the minor diastereoisomers during the isolation procedure. The complexes 13, 18-21 were first applied to the asymmetric reduction of the representative substrate acetophenone, using both asymmetric transfer hydrogenation (ATH) and pressure hydrogenation (APH) conditions. The complexes were also used in the oxidation of racemic 1-phenylethanol using acetone as a hydrogen acceptor. In all cases the active catalyst was generated in situ using methods previously reported, and in some cases comparisons with the unfunctionalised catalyst 4b 2b were also made. [2][3][4][5][6][7][8][9][10][11][12]15 In the ATH reactions (Table 2 shows selected results, further results are given in the ESI, Table S1 †), at 60°C using a 5 : 2 (molar) formic acid : triethylamine azeotrope (FA/TEA) and 10 mol% catalyst, full conversion was observed in several cases however the asymmetric inductions were extremely low and a significant amount of formate co-product was also formed, presumably through formylation of the initial alcohol product (confirmed to be of the same absolute configuration as the alcohol). 6a TMAO was added to ensure full activation of the catalysts efficiently although, as demonstrated using unsubstituted 4b, it could be omitted from the reaction 6a at the cost of a slower activation; a reaction complete in 24 h using TMAO reached just 52% conversion in the same time without TMAO (entries 1-3). In this respect, however the diol complex 13 appeared to be less sensitive to the additive (entries 4-7). Although the enantioselectivities were low, the OH and OBn complexes gave products of opposite configurations to those observed with the O-silylated complexes (entries [8][9][10][11][12][13][14][15]. The use of lower catalyst loadings (5%, 1%) gave much lower conversions, as did lowering the temperature to 40°C (ESI †). Another clear trend was the observation of improved ees when using the more hindered silylsubstituted complexes, for both the alcohol and formate products.
A similar pattern emerged for the pressure hydrogenation reactions (APH ; Table 3), although some unusual observations were made with respect to the method of activation. Control reactions run under nitrogen indicated that significant background transfer hydrogenation was also operating (entries 4-6, 12-14), however the conversions were lower. The similar ees (for 13) observed in the absence of hydrogen gas indicated that the catalyst was still operating in the reaction. Initially, unsubstituted complex 4b was tested and this gave 100% conversion under 30 bar hydrogen pressure at 80°C in 24 h in an iPrOH/water solvent mixture, provided that an activator; either K 2 CO 3 7a,b or TMAO, 15 was added (entries 1-3). Given this precedent, K 2 CO 3 was used in all the subsequent tests. Unfortunately, the substituted catalysts were not as active as 4b, giving much lower conversions under the same conditions (entries 8,17,24,30). No advantage was gained from running the reactions at the higher temperature of 100°C (entries 9, 18,21,25,30); catalyst decomposition was suspected) or for a longer time (72 h) at a lower temperature of 60°C (entries 16,20,23,28). Whilst most of the other complexes behaved in a similar manner to 4b, requiring some form of activation for best results, for diol-containing complex 13, omission of the K 2 CO 3 (and no other activator) resulted in full acetophenone reduction (entry 7), as did the use of 1 mol% of TMAO (entry 10). The use of 1.5 mol% of TMAO, however, gave a lower conversion, possibly due to partial catalyst decomposition (entry 11). The use of 1 mol% of TMAO also proved to be the most effective way to activate the other catalysts where tested (entries 19, 26, 31). The asymmetric induction was not improved in either case however and remained modest, not exceeding 20% ee in any case, although again the more hindered silylated complexes gave the best enantioselectivities. This may reflect the distant separation of the chiral centres from the likely reduction centre, which requires some further optimisation to extend its influence to the transition state of the reduction; the trend in the results suggests that a further increase in the steric hindrance of the groups on the 'bridging C-4' unit could provide a route to such improvements. Given the unexpected result obtained when K 2 CO 3 was omitted from the reaction with diol catalyst 13, we questioned whether the addition of water to a dry solvent could reproduce

Entry Catalyst Activator
Time this effect. This proved to be the case; in dry THF, 31% reduction of acetophenone was observed after 18 h at 80°C, possibly due to trace amounts of adventitious water, however as increasing amounts of water were added, the conversion increased, reaching 99% when 100 mol% (1 eq.) relative to substrate, and 100% when a 5/2 THF/water solvent mixture was usedalthough no change to the ee was observed (ESI ,  Table S2 †). A 1 H-NMR study of the reaction revealed the formation of an iron hydride complex, suggesting that water was initiating the formation of the active species. A similar effect, although to a lesser extent, created by addition of water was also observed when toluene was used as solvent (ESI , Table S3 †).
The complexes proved to be efficient at the catalysis of alcohol oxidation using acetone as hydrogen acceptor, a process which has been reported for a number of iron(cyclopentadienyl) catalysts. 6b,9 However both enantiomers of substrate were oxidised with little selectivity (Table 4), although with excellent conversions, particularly when 10 mol% of catalyst was used. The catalyst loading could be reduced to 5 mol% in some cases although at 1 mol% loading, incomplete conversion was observed (ESI, Table S4 †). The observed enantiomeric excesses indicated that no significant level of kinetic resolution was taking place in the oxidations.
The design of the MONOPHOS-containing catalysts 22-25 was anticipated to be capable of delivering improved results since; (i) MONOPHOS is known to be compatible with complexes of this type and (ii) the additional element of chirality in the ligand could be matched or mismatched to that on the cyclohexyl ring. 5 In the event, 22-25 exhibited very low activity in the applications in which they were tested, despite the use of several methods to activate them. Results for the ATH of acetophenone and 3,3-dimethyl-2-butanone with the MONO-PHOS complexes are given in the ESI (Table S8 †). Using 10 mol% catalyst, for 24 h at 25-60°C, and activation with TMAO, conversions were below ca. 7%. Performing the reaction using TMAO (in the dark or in the presence of light), or blue or UV light (365 nm) to activate the precatalyst, no improvement was observed. In one case, the use of the OTBDMS 24 catalyst (60°C, 24 h) in the reduction of acetophenone gave 12.2% alcohol (18.0% ee) and 8.9% formate (12.4% ee), both of R configuration. The same outcome was observed in pressure hydrogenation tests. Using 1 mol% catalyst, 60°C, 72 h, gave <5% conversion and ees of less than 10% (ESI, Table S9 †).
To examine the effect of a phosphine we added increasing amounts of triphenylphosphine to 1 mol% of 13 under APH conditions (ESI, Table S10 †). In this case, as the amount of PPh 3 increased, the conversion (80°C, 18 h) decreased and with 2% PPh 3 relative to 1 mol% catalyst 13, no reduction was achieved. This may indicate the formation of a less reactive phosphine-containing species similar to 22-25. This study was also carried out using MONOPHOS as an additive and a similar reduction in activity was observed although the enantioselectivity was not significantly changed (ESI, Table S11 †). Funk and Moyer have described closely related complexes which demonstrate lower activity towards hydrogen transfer when a CO ligand was replaced by a phosphine, and our results mirror these. 9a The X-ray crystallographic structures of complexes 4b 2b (Fig. 2), 22 (Fig. 3) and 23 (Fig. 4) were obtained (see ESI † for full details). In the case of 23, two independent molecules of similar conformation were observed (see ESI † for full detail of each structure). Both complexes 22 and 23 are quite congested and hindered, particularly in comparison with 4b, which benefits from a more 'open' structure. This may account for the observed low reactivity of the derived hydride from the MONOPHOS-containing complexes, not only in comparison with 4b, but also compared to complex 6 reported by Berkessel, which does not contain substituents on the carbon backbone.
In order to investigate in more detail the reasons for the slow reactivity of 22-25 an 1 H-NMR experiment was carried out using catalyst 24 under hydrogenation conditions in a sealed NMR tube (Scheme 2). Upon irradiation over a period of 4 h, a major signal for an iron hydride 26 was observed at −12.11 ppm ( J = 88.6 Hz) and a minor doublet at −12.18 ppm ( J = 80.1 Hz) in ratio of ca. 12 : 1 which are tentatively assigned as diastereoisomers at the Fe atom based on analogy with Berkessel's observations. 5 There was also a further doublet at −11.40 ppm tentatively assigned to the hydride from complex 25 present in 24. This was subsequently confirmed by independent formation of the hydride from complex 25 (see ESI †).
In our case the high ratio of Fe isomers did not translate into a high enantioselectivity in the reduction. However a sample of added acetophenone was not significantly reduced in the reaction after heating the reaction to 80°C for 3 days, whilst the peak at −12.18 ppm became the major species, together with some additional peaks which corresponded to 24 although this was not reisolated (see ESI †). This strongly indicates that whilst an iron hydride does form in the reaction, the transfer of hydrogen from the complex to the substrate is very slow. Likewise the high level of steric hindrance created in the complex through the introduction of a bulky phosphine may also explain the observed dramatic reduction in reactivity upon addition of triphenylphosphine in the APH tests described above. However, electronic effects could also be important; a computational study on this class of complex 13c Fig. 2 The X-ray crystallographic structure of 4b (CCDC 1431241).  has revealed that increasing the acidity of the CpOH group can contribute to greater activity. Hence replacement of the electron-withdrawing CO with either PPh 3 or MONOPHOS could reduce the acidity of the CpOH and hence reduce catalytic activity. The authors of the computational study indicated that phosphines containing electron-donating groups, of moderate steric size, could potentially increase the CpOH acidity and hence the activity, and this remains the subject of future studies. Similar observations have been made in experimental studies on the ruthenium analogues of the iron catalysts used in this study; in particular the rapid and reversible formation of a bond from the substrate carbonyl to the OH of the CpOH has been shown to be the first step of the catalytic mechanism. 17 In conclusion, we have prepared a series of enantiomerically-pure (cyclopentadienone)iron complexes through the cyclisation of a C 2 -symmetric diol and its derivatives. Replacement of a CO with a phosphorus-donor ligand has also been achieved. The complexes are competent catalysts for the reduction of ketones under a range of conditions and for alcohol oxidation however no significant enantiomeric inductions were achieved in these transformations. An ideal catalyst might benefit from a balance between the size of the phosphorus-donor and any groups on the other part of the complex.

Experimental section
General Solvents and reagents for the synthesis of complexes and catalytic reactions were degassed prior to use and all reactions were carried out under either a nitrogen or argon atmosphere. All heated experiments were conducted using thermostatically controlled oil baths. Reactions were monitored by TLC using aluminum backed silica gel 60 (F254) plates, visualized using UV 254 nm and phosphomolybdic acid (PMA), potassium permanganate or vanillin dips as appropriate. Flash column chromatography was carried out routinely using 60 micrometer silica gel. Reagents were used as received from commercial sources unless otherwise stated. 1 H NMR spectra were recorded on a Bruker DPX (300, 400 or 500 MHz) spectrometer. Chemical shifts are reported in δ units, parts per million relative to the singlet at 7.26 ppm for chloroform and 0.00 ppm for TMS. Coupling constants ( J) are measured in Hertz. IR spectra were recorded on a Perkin-Elmer Spectrum One FT-IR Golden Gate. Mass spectra were recorded on a Bruker Esquire2000 or a Bruker MicroTOF mass spectrometer. Melting points were recorded on a Stuart Scientific SMP 1 instrument and are uncorrected. GC analysis was performed using a Hewlett Packard 5890. Dry solvents were purchased and used as received.
Procedure for oxidation of alcohols with acetone as the hydrogen acceptor Acetophenone (100 mg, 0.83 mmol, 1.0 eq.), Catalyst (0.10 eq./ 10 mol%) and acetone (4.3 mL) were added to a pressure tube, degassed and stirred for 10 min. TMAO (0.1 eq./10 mol%) was added, the reaction tube was sealed and the mixture was heated at the temperature indicated for the time stated. At the end of this time the reaction was allowed to cool to rt and the sample was worked up and analyzed as previously described.