Allylic substitution 1 (Pd)
Allylic substitution 2 (Pd)
Allylic substitution 3 (Ir)
Iridium Catalysed Allylic substitution
A key breakthrough in the use of iridium in enantioselective allylic substitution reactions was the discovery that Feringa-type ligands such as 1 work very well [JACS02-124-15164, JACS03-125-14272]. The initially formed complex A, resulting from the addition of 1 to [Ir(COD)Cl]2, is not the catalyst. Instead formation of an iridacycle by methyl C-H group activation results in B, dissociation of L resulting in the catalyst C. As a result, these reactions need to be run under basic conditions.
The allyliridium intermediate D may be synthesised from the ligand, [Ir(COD)Cl]2 and an allylic carbonate (with the counter ion X- introduced from AgX – CEJ09-15-11087). The complexes are air-stable and may be purified by column chromatography. Just one, the thermodynamically most stable, of the 16 possible stereoisomers is formed. Replacement of COD with dbcot gives a catalyst that operates under aerobic conditions, and results in an improvement in enantioselectivity and regioselectivity [Angew08-120-7652].
Use of a related achiral catalyst with an enantiomerically pure substrate revealed a conservation of enantiomeric purity [CC99-741]. This stereospecificity reveals that the rate of nucleophilic attack is faster than the rate of isomerisation of the intermediate allyl complex. Oxidative addition and attack of the nucleophile both proceed with inversion of configuration, resulting in overall retention of configuration. By extension, with a prochiral substrate the stereo-determining step is oxidative addition, and use of racemic branched allylic substrates results in low enantioselectivity.
Use of ligand (Sa,S,S)-1 results in a much faster reaction compared to the use of its (Ra,S,S) diastereoisomer [JACS02-124-15164]. This is accounted for, at least in part, by the much faster rate of cyclometallation of (Sa,S,S)-1 compared its (Ra,S,S) diastereoisomer [JACS05-127-15506]. In addition, (Sa,S,S)-1 is the matched ligand, and the (Ra,S,S) form the mismatched ligand, with respect to product enantioselectivity [see also EJOC03-1097].
Replacement of the BINOL derived moiety by 2,2’-biphenol to give ligand (S,S)-3 resulted in a modest reduction in enantioselectivity [Angew04-43-2426] (95% ee for 1 vs. 87% ee for 3 in a representative allylic amination). A reduction in the number of chirality elements to just one (i.e. this being the stereogenic centre within the resulting iridacycle ring) was achieved with a cyclododecane containing ligand, this resulting in ee values of >90% ee for allylic amination reactions [JACS05-127-15506].
Review: CR19-119-1855. The following illustrates a number of representative reactions (and is very far from exhaustive).
Phosphoramidite 1. Excellent regioselectivity (branched : linear typically about 50 : 1, lowest = 11 : 1). With R2 = H typically small quantity of double alkylation (<5%). Lower regioselectivity (6 : 1) with R1 = p-NO2C6H4 (86% ee) and ee = 76% with R1 = o-MeOC6H4.
Phosphoramidite 2. Significant acceleration in the reaction rate noted with phosphoramidite 2 compared to phosphoramidite 1.
Phosphoramidite 2. Most R1 substituents gave ≥99 : 1 branched/linear. Lower ration with R1 = Cy (13 : 1) and R1 = n-Pr (4 :1). With R1 = o-MeOC6H4, ee = 79%.
Solvent concentration not given. Somewhat lower branched/linear ratio (ca. 6 : 1) where R1 = i-Pr, n-Pr. Both CsF and ZnF2 needed – no reaction with ZnF2 only. Substoichiometric CsF aids avoidance of diaallylation.
Phosphoramidite 1. Excellent regioselectivity (branched : linear typically about 25 : 1, lowest = 7 : 1). With R1 = 2-MeOC6H4 (R2 = Me, Ar = Ph) ee = 75%. Lithium phenolates prepared from the reaction of the corresponding phenol with n-BuLi.
This chemistry was exemplified using the derivative of phosphoramidite 1 containing two 1-naphthyl groups rather than two phenyl groups [Ra,R,R] [Sa,S,S] – 2 mol% ligand, 94% ee, 91% yield, 99:1 branched/linear. Used double the catalyst loading with sec and tert-alkoxides (and longer reaction times). Copper (and zinc) alkoxides required (no reaction with lithium alkoxide). Tert-butyl ester to avoid transestrification. Use of both enantiomers of chiral sec-alkoxides resulted in essentially one or the other product diastereoisomers (i.e. no matched/mismatched influences).
The iridacycle catalyst is derived from either Phosphoramidite 1 or Phosphoramidite 2 and dbcot as described in the supporting information. Higher e.e. values obtained with Ar = o-MeC6H4 (approx 5% higher). These iridacycles catalysts are stable to air and water – by extension, no useful results were obtained with a corresponding COD-containing catalyst. The bicarbonate anion is thought to be the nucleophile, followed by decarboxylation.
The iridacycle catalyst (not commercially available) contains a labile ethene ligand, loss of which gives a 16-electron species with which the catalytic cycle is initiated by oxidative addition (see JACS09-131-8971). Generally excellent branched to linear selectivity (≥99 : 1).
No reaction times given.