E.
Louise Hazeland
,
Andy M.
Chapman
,
Paul G.
Pringle
* and
Hazel A.
Sparkes
School of Chemistry, University of Bristol, Cantock's Close, Bristol, BS8 1TS, UK. E-mail: paul.pringle@bristol.ac.uk
First published on 18th May 2015
A chlorosilane elimination reaction has been developed that allows the efficient synthesis of optically pure C1-symmetric, C1-backboned diphosphines with a wide variety of stereoelectronic characteristics.
Fig. 1 Phosphorus ligands for asymmetric hydrogenation. A = diop;3B = dipamp;4C = binap;5D = DuPhos;6E = miniphos;7F = trichickenfootphos;8G = monophos.9 |
The high enantioselectivity obtained with catalysts based on C2- or C1-symmetric diphosphines has been rationalized in terms of the degree of control of the metal binding site offered by the chelates involved.13 For example the rigid 4-membered rings formed by the C1-backboned E and F (Fig. 1) have been spectacularly effective for asymmetric hydrogenation,7,8,14 and it is the rigidity of the metal chelates that appears to be a critical feature of these catalysts. Despite the multitude of diphos ligands that have been prepared, there continues to be a need for new ones because, as several authors have noted, ligand discovery remains largely an empirical rather than a rational endeavour.15 A disadvantage of diphos ligands is that their synthesis is often multistep and/or requires an optical resolution step, making systematic refinement of their structures time-consuming and laborious.16 A major reason why monophos ligands such as G have attracted attention17 is that their synthesis is simple, modular and so reliable that they have been employed in high-throughput experimentation (HTE). Here we report a simple, one-step route to C1-linked diphos ligands that has the capacity to create a library of optically-active diphos ligands rapidly.
The construction of achiral C1-linked diphosphines by an Si–P exchange reaction such as that shown in eqn (1) has been previously reported.18,19 The attraction of this route is that the volatile by-product is readily removed and therefore we have investigated its potential as the basis for a general route to optically active, C1-linked diphos ligands.
(1) |
The reaction of chlorophosphite 1a with the trimethylsilylmethyl-phosphine 2a gave the diphos ligand La quantitatively (eqn (2)).
(2) |
The reactants in eqn (2) are readily varied and easily prepared.19–22 Thus Lb–f are produced in high yields from the reactions of trimethylsilylmethylphosphines 2b–d with the corresponding optically-pure halophosphites 1a–c (Scheme 1). A significant extension of this process was achieved by employing the optically pure chlorophosphacycle 1d (Scheme 1) to produce Lg–j. The crude products La–j were sufficiently pure to be used in catalysis without further purification.
Scheme 1 Synthesis of the diphos ligands La–j. X = Cl in all cases except for 1b and 1c where X = Br. |
The complexes [Rh(diene)L] (3) where diene = 1,5-cyclooctadiene or norbornadiene were generated by the addition of La–j to [Rh(diene)2][BF4] in CH2Cl2 and in each case the product was identified from the characteristic AMX pattern in its 31P NMR spectrum (see ESI† for details). Representative examples of 3, where L = Ld or Lg–j, have been isolated and fully characterised. The ligands were screened for the asymmetric hydrogenation of the three benchmark substrates DMI, MAA and MAC (structures shown below) and the results are given in Table 1 and depicted graphically in Fig. 2 from which it is clear that significant variation in selectivity occurs for ostensibly small changes in ligand structure.
Entry | Ligand | DMI | MAA | MAC | Conv. (%) |
---|---|---|---|---|---|
a Reaction conditions: S/C = 100:1, 5 bar H2, 20 °C, 1 h, CH2Cl2. Enantioselectivities were determined by chiral GC analysis (see ESI for details). | |||||
1 | La | 68 (S) | 100 | ||
2 | La | 46 (R) | 100 | ||
3 | La | 35 (R) | 100 | ||
4 | Lb | 90 (S) | 100 | ||
5 | Lb | 72 (R) | 100 | ||
6 | Lb | 56 (R) | 100 | ||
7 | Lc | 95 (S) | 100 | ||
8 | Lc | 76 (R) | 100 | ||
9 | Lc | 61 (R) | 100 | ||
10 | Ld | 91 (S) | 100 | ||
11 | Ld | 65 (R) | 100 | ||
12 | Ld | 64 (R) | 100 | ||
13 | Le | 94 (S) | 56 | ||
14 | Le | 98 (R) | 100 | ||
15 | Le | 94 (R) | 49 | ||
16 | Lf | 91 (S) | 47 | ||
17 | Lf | 72 (R) | 35 | ||
18 | Lf | 93 (R) | 20 | ||
19 | Lg | 62 (R) | 100 | ||
20 | Lg | 57 (S) | 100 | ||
21 | Lg | 46 (S) | 99 | ||
22 | Lh | 97 (R) | 100 | ||
23 | Lh | 90 (S) | 100 | ||
24 | Lh | 85 (S) | 39 | ||
25 | Li | 92 (R) | 100 | ||
26 | Li | 96 (S) | 100 | ||
27 | Li | 85 (S) | 22 | ||
28 | Lj | 99 (R) | 100 | ||
29 | Lj | 98 (S) | 100 | ||
30 | Lj | 99 (S) | 26 |
For the complexes of the binol-derived La–d with DMI and MAA, the highest ee was obtained with the PCy2 derivative: La < Lb < Lc > Ld (entries 1–12 in Table 1). For the complexes of the 3,3′-substituted ligands Ld–f, the highest ee was obtained when the 3,3′-substituents were Ph: Ld < Le > Lf (entries 10–18 in Table 1). With complexes of the phospholane-derived ligand Lg–j, the enantioselectivity was greatest for the PtBu2 ligand: Lg < Lh < Li < Lj (entries 19–30). It is apparent from Fig. 2 that the performance of any particular ligand can be highly substrate-dependent.
The absolute configuration of the asymmetric hydrogenation products obtained with Rh-diphos complexes generally obey the quadrant-blocking rule; that is, blocked upper left quadrant leads to R-configuration for MAC and MAA and S-configuration for DMI.13 The nature of the quadrant blocking is best discerned from crystal structures and so crystals of the [Rh(cod)(Lj)]BF4 (3j) were grown and its crystal structure determined, which has two molecules in the asymmetric unit (Fig. 3). Attempts to grow crystals suitable for X-ray crystallography of Rh-complexes of the binol-derived ligands (La–d) have so far been unsuccessful, although crystals of the chelate [PtCl2(Ld)] (4d) have been obtained and its structure is shown in Fig. 4. In both structures (3j and 4d), the acute P–M–P angles of 72.84(3)° in 3j and 73.74(3)° in 4d indicate the degree of strain present in the 4-membered chelates; these values are very similar to the 72.55(6)° that was determined in an analogous complex of Trichickenfootphos (F in Fig. 1).8 The mean planes through M–P–P–C have rms deviations of 0.035/0.049 Å in 3j and 0.003 Å in 4d showing that the chelates are almost planar (see Fig. 3 and 4). It is evident from Fig. 4 that the upper left quadrant is blocked in the Ld complex (and presumably the same would be the case for all the ligands La–f) while Fig. 3 shows lower left quadrant is blocked in the in Lj complex (and presumably the same would be the case for all the ligands Lg–j). Therefore, the absolute configurations of the products of asymmetric hydrogenation (Table 1) conform to the quadrant rule.
The remarkable efficiency of the ligand synthesis (Scheme 1) coupled with the ready removal of the volatile chlorosilane by-product suggested that a one-pot procedure may be feasible. This was carried out according to Scheme 2 for Lj and the product tested for asymmetric hydrogenation of MAA. The 97% ee that was obtained compares favourably with the 98% ee recorded with the isolated complex (Table 1).
The simplicity and generality of the chlorosilane elimination route shown in Scheme 1 to C1-symmetric, C1-backboned, optically pure diphos ligands has been demonstrated by varying the nature of the two P-reagents. The success of the one-pot procedure (Scheme 2), coupled with the fact that the number of potential ligands increases geometrically with each new chlorophos or silylmethylphosphine component, opens up the possibility of applying HTE methods to diphosphine synthesis and catalyst screening in a way that previously, have only been applied to monophos ligands. This is currently under investigation as is the mechanism of the ligand formation reaction.
We thank EPSRC for supporting this work with a studentships to ELH.
Footnote |
† Electronic supplementary information (ESI) available: Experimental details, additional spectra and crystallographic data. CCDC 1059329 and 1059330. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5cc03517a |
This journal is © The Royal Society of Chemistry 2015 |