Javier
Francos
,
Silvia
Zaragoza-Calero
and
Charles T.
O'Hara
*
WestCHEM, University of Strathclyde, Department of Pure and Applied Chemistry, 295 Cathedral Street, Glasgow, G1 1XL, UK. E-mail: charlie.ohara@strath.ac.uk; Tel: +44 (0)141 548 2667
First published on 1st November 2013
A study has been conducted to determine whether lithium magnesiates are feasible candidates for the enantioselective deprotonation of 4-alkylcyclohexanones. The commercially available chiral amine (+)-bis[(R)-1-phenylethyl]amine (2-H) was utilised to induce enantioselection. When transformed to its lithium salt and combined with nBu2Mg, improved enantioselective deprotonation of 4-tert-butylcyclohexanone (with respect to the monometallic lithium amide) at 20 °C was observed. In an attempt to optimise the reaction further, different additives were added to the lithium amide. The best performing deprotonations at 0 °C were those in which (Me3SiCH2)2Mg (er pro-S 74:26) and (Me3SiCH2)2Mn (er pro-S 72:28) were added, hence the lithium magnesiate “LiMg(2)(CH2SiMe3)2” was used in the remainder of the study. The optimum solvent for the reaction was found to be THF. NMR spectroscopic studies of a D8-THF solution of “LiMg(2)(CH2SiMe3)2” appear to show that this mono-amide bis-alkyl species is in equilibrium with a bis-amide mono-alkyl compound (and a tris-alkyl lithium magnesiate). When a genuine bis-amide lithium magnesiate solution is used, the deprotonation results were essentially identical to those obtained for “LiMg(2)(CH2SiMe3)2”. By adding LiCl to “LiMg(2)(CH2SiMe3)2” the er at 0 °C improved to 81:19. At −78 °C good yields and an er of 93:7 were obtained. This LiCl-containing base was used to successfully deprotonate other 4-alkylcyclohexanones.
Scheme 1 Reaction of lithium chiral amides and amido (bis)alkyl lithium magnesiates with 4-tert-butylcyclohexanone. |
At 0 °C, the additive-free reaction produced an enantiomeric ratio (S:R) of 60:40 in a yield of 76%. Inorganic salts are known to have an effect – advantageous or detrimental – on the course of reactions.10 When simple inorganic salts such as MgBr2 and CoBr2 were utilised (entries 2–4, Table 2), essentially none of the desired product was formed, although addition of ZnCl2 produced a yield of 25% with a S:R ratio of 65:35. If an organometallic reagent was added instead of the salt, the main trend observed was that the yields improved (in excess of 78%) although enantiomeric ratios were variable.
Entry | Metal reagent added to 2-Li | Yieldb (%) | er (S:R)b |
---|---|---|---|
a CoBr2 and ZnCl2; reaction time was 16 h. b Determined by GC analysis. | |||
1 | — | 76 | 60:40 |
2 | MgBr2 | 0 | — |
3a | CoBr2 | 4 | — |
4a | ZnCl2 | 25 | 65:35 |
5 | n Bu2Mg | 83 | 70:30 |
6 | (Me3SiCH2)2Mg | 96 | 74:26 |
7 | Me2Zn | 99 | 58:42 |
8 | t Bu2Zn | 98 | 60:40 |
9 | Me3Al | 89 | 64:36 |
10 | iBu3Al | 78 | 60:40 |
11 | (Me3SiCH2)2Mn | 99 | 72:28 |
As the n-butyl ligand contains β-hydrogen atoms, and is therefore prone to decomposition via a β-hydride elimination pathway we switched our attention to the trimethylsilylmethyl anion (Me3SiCH2−). The best S:R enantiomeric ratio (74:26) was obtained when the dialkylmagnesium (Me3SiCH2)2Mg was used (entry 6, Table 2). When Zn or Al organometallics were used (entries 7–10, Table 2) the er observed were essentially identical to those obtained when 2-Li was utilised without an additional metal (entry 1, Table 2). Interestingly, on adding the transition metal alkyl (Me3SiCH2)2Mn to the lithium reagent, essentially quantitative conversion to the silyl enol ether was observed with a high degree of enantioselection (S:R, 72:28; entry 11, Table 3). Thus far our best enantioselection was achieved using 2-Li and (Me3SiCH2)2Mg (entry 6, Table 3). Therefore we focused our attention on trying to optimise this reaction (and that involving commercially available nBu2Mg) further. By adding a deficit (0.5 equivalents) or an excess (two equivalents) of nBu2Mg to 2-Li we discovered that the yield of product increased for the former, but decreased for the latter, and the enantiomeric ratio in both cases dropped to 62:38 (entries 2–4, Table 3). When 1:1, 1:2 and 2:1 ratios of 2-Li and (Me3SiCH2)2Mg are utilised, the product yields are excellent (>89%) and all the er are improved from the optimum nBu2Mg case, the best obtained was for the magnesium-rich reaction (pro the S enantiomer, 78:22) although it was modestly better than the 1:1 case (74:26) (entries 5–7, Table 3).
Entry | Metal reagent added to 2-Li | Yielda (%) | er (S:R)a |
---|---|---|---|
a Determined by GC analysis. | |||
1 | — | 76 | 60:40 |
2 | 1 equivalent nBu2Mg | 83 | 70:30 |
3 | 2 equivalents nBu2Mg | 52 | 62:38 |
4 | 0.5 equivalent nBu2Mg | 99 | 62:38 |
5 | 1 equivalent (Me3SiCH2)2Mg | 96 | 74:26 |
6 | 2 equivalents (Me3SiCH2)2Mg | 89 | 78:22 |
7 | 0.5 equivalent (Me3SiCH2)2Mg | 99 | 72:28 |
The next variable to be considered was ascertaining what effect changing the solvent medium from THF would have on the results. In carrying out the reaction of 2·Li and (Me3SiCH2)2Mg in neat toluene, only a trace of the desired product formed. In diethyl ether, the GC-yield was 40% and the er was only 67:33 in favour of the S enantiomer. To aid solubility in hexane, one molar equivalent of THF was added; however, the product yield was low (23%) and the er was 65:35; hence it appears that for the systems tested THF remains the optimum solvent choice.
At this juncture it is appropriate to detail some of the NMR spectroscopic work that was performed in an attempt to shed light on the active species that is likely to carry out the deprotonation. Full details and spectroscopic data can be found in the ESI.† The 1H (and where appropriate 7Li) NMR spectra of: (a) 2-H; (b) 2-Li; (c) (2-)2Mg; (d) 1:1 2-Li and (Me3SiCH2)2Mg; and, (e) Me3SiCH2Li in D8-THF were compared. The NCH methylene H-atom proved an important handle for comparing the spectra of (a)–(d). For (a)–(c) the respective 1H NMR shifts were 3.50, 3.64 and 3.84 ppm. For (d) – the active base system – two distinct resonances in approximately equal proportions were observed (3.53 and 3.69 ppm). It was initially envisaged that these two resonances belonged to the diastereotopic H atoms present in the organometallic complex. However, a DOSY NMR experiment on a D8-THF solution of (d) revealed that the solution contained two distinct amido-containing species.† Thus we believe that the expected mono(amido)-bis(alkyl) species “LiMg(2)(CH2SiMe3)2” undergoes ligand reorganisation and is in equilibrium with a bis(amido)-mono(alkyl) relative “LiMg(2)2(CH2SiMe3)” and the tris(alkyl) “LiMg(CH2SiMe3)3” (Scheme 2 and Fig. 1). To provide evidence of this equilibrium, we prepared genuine D8-THF solutions of “LiMg(2)2(CH2SiMe3)” and “LiMg(CH2SiMe3)3” and obtained their respective 1H NMR spectra corroborating that these species were present in the initial 1:1 2-Li and (Me3SiCH2)2Mg reaction (Fig. 2).
Scheme 2 Proposed equilibrium which occurs when a 1:1 mixture of 2-Li and (Me3SiCH2)2Mg is combined in D8-THF. |
Interestingly, when the bis(amido) magnesiate “LiMg(2)2(CH2SiMe3)” was utilised in the enantioselective deprotonation reaction at 0 °C, an identical enantiomeric ratio (74:26 pro-S) was obtained when compared to the 1:1 2-Li and (Me3SiCH2)2Mg reaction (Table 3, entry 5).† The key finding from this experiment is that the addition of a second equivalent of the chiral amine does not improve enantioselection. This begged the question could a simple homometallic alkylmagnesium chiral amide function well in this enantioselective deprotonation? By reacting a 1:1 nBu2Mg–2-H mixture in THF with the ketone for one hour, followed by work-up, only a 3% yield of the desired silylenol ether was obtained. By increasing the reaction time to 16 hours, the yield increased to 72% but the er was 53:47. Hence, for a homometallic magnesium reagent to function it appears that two amides (i.e., the absence of alkyl groups) are required (vide infra).2
Returning to the 1:1 mixture of 2-Li and (Me3SiCH2)2Mg, in an attempt to improve the enantioselectivity of the reaction further, we decided to introduce some common additives (Table 4). Focusing on the reactions involving 2-Li and (Me3SiCH2)2Mg, it was discovered that high yields were maintained using additives; however, TMEDA (N,N,N′,N′-tetramethylethylenediamine) addition caused a slight drop-off in er (from 74:26 to 71:29), but DMPU [1,3-dimethyl-3,4,5,6-tetrahydro-2(1H)-pyrimidinone] and LiCl both enhanced selectivity (to 78:22 and 81:19 respectively). The same 81:19 ratio was obtained when the LiCl was generated in situ by reaction of 2·HCl with two equivalents of nBuLi.
Although the primary purpose of this work was to find an optimum system to operate at close to ambient temperature, for completeness we also investigated the reaction at various temperatures down to −78 °C (Table 5). As expected the optimum er was obtained at the lowest temperature (93:7) with an isolated yield of 87%. When compared with literature work2 which utilised the chiral magnesium bis(amide) (2)2·Mg, it was noted that an identical er was obtained; however, using the magnesiate system gave an increased isolated yield (87% versus 66%). As alluded to earlier, in the monometallic magnesium system two equivalents of chiral amide are required, as one equivalent would produce an alkylmagnesium amide that has an inherently different reactivity and is also subject to Schlenk-type equilibria and perhaps oligomerisation in solution. In our magnesiate system, only one equivalent of chiral amine is required and our sole additive is LiCl i.e., there is no need for DMPU. A final advantage is that the magnesiate reaction time is 3 hours; whereas for the magnesium bis(amide) it is 16 hours.
To summarise, the optimal results that we have obtained thus far are when we utilise a 1:1:1 mixture of 2-Li, (Me3SiCH2)2Mg and LiCl. As such we tested this reaction mixture with other 4-alkylcyclohexanones (Table 6). In all cases attempted, moderate to good yields were obtained (52–89%). The best er obtained was when the alkyl group's steric bulk was reduced to isopropyl, producing an er of 82:18. Good enantioselection was also obtained for nPr, Ph and Me which were 78:22, 77:23 and 75:25 respectively (Table S1†). Finally, to ascertain whether the opposite enantioselection is possible we employed (−)-bis[(S)-1-phenylethyl]amine in our optimum reaction instead of its enantiomer 2. We obtained 75% isolated yield of the compound with an S:R er of 18:82 at 0 °C (Table 6).
Achiral G.C. analysis: (i) CP Chirasil-DEX CB column; (ii) carrier gas, H2 (45 cm s−1): (i) injector/detector temperature, 250 °C; (ii) initial oven temperature, 90 °C; (iii) temperature gradient, 45 °C min−1; (iv) final oven temperature, 220 °C; and (v) detection method, FID.
Chiral G.C. analysis: (i) CP Chirasil-DEX CB column; (ii) carrier gas, H2 (45 cm s−1): (i) injector/detector temperature, 250 °C; (ii) initial oven temperature, 70 °C; (iii) temperature gradient, 1.7 °C min−1; (iv) final oven temperature, 120 °C; and (v) detection method, FID.
We gratefully acknowledge the support of the EPSRC (J001872/1 and L001497/1) for the award of a Career Acceleration Fellowship to CTOH.
Footnote |
† Electronic supplementary information (ESI) available: Full experimental procedures, and NMR data. See DOI: 10.1039/c3dt52577e |
This journal is © The Royal Society of Chemistry 2014 |