PLANTEAMIENTO DEL PROBLEMA 2.1 Identificación del problema

As reported by Noyori and co-workers the rhodium catalysed isomérisation of A/,A/-diethylgeranylamine (64) proceeds to give the corresponding enamine in 96% yield, and greater than 95% enantiomeric excess (Figure 65).^^° This is far higher than that which has been achieved fo r allylic alcohols, even with the same chiral catalyst (Figure 64)V^ It therefore seemed appropriate to use this substrate with the Ni(COD)2/pyridine cataiyst system. A/,A/-Diethylgeranyiamine (64) was prepared in one step by the addition of diethylam ine to myrcene (223) in the presence of n-BuLi (Figure 179).^^^ A/,A/-Diethylneranylamine (224) can be prepared in sim ilar fashion from isoprene (225) 270

N E t, (223) NEt (64) (225) Figure 179 N E U (224)

Isom érisations were conducted using a variety of ligands, namely with pyridine, DIPHOS and bipyridyl, however in each case, analysis of the isomérisation reaction by GC indicated only starting m aterial, despite prolonged reaction times. It is possible that the allylic am ine is coordinating to the nickel catalyst, displacing the original ligand to form a hexaamine complex, not too dissim ilar to the [Ni(NH3)6f '" com plex form ed by the addition of am m oniacal solution to aqueous nickel(ll)chloride.

Aithough this substrate has been shown not to a be suitable substrate with the Ni(C0 D)2/iigand catalyst systems, nevertheless it was curious as to why rhodium could isomerise this substrate, whereas the nickel could not.

138 Results and Discussion

Studies On Reqiocontrol

Earlier studies on regiocontrolled enolate form ation found that the (Cy3P)2NiCl2/n-BuLi catalyst could isomerise the alkoxide of 1-phenyl-4-penten-3-ol (102) regioselectively in 92% yield. Problems were encountered however using the [Rh(DIPHOS)r^^^ and the NiCl2(DIPHOS)/LiBEt3H catalyst systems in which a mixture of regioisomers were obtained

(Figure 94).

It was envisaged that using the highly stereoselective Ni(COD)2/pyridine catalyst system that sim ilar degrees of regiocontrol would also be exerted. The alkoxide of 1-phenyl-4-penten-3-ol (102) was therefore treated with the catalyst system in THF heated to reflux for 2 hours, then quenched with acetic anhydride. Purification of the reaction mixture afforded the ketone (226) in 17% yield, the enol acetate products as a mixture of regioisomers in 26% yield (Figure 180) and the starting material as In 21% yield

O A c O A c (!), (ii). (iii) 8 .2 ; 1 (226) O A c

+

Regiochemical ratio: i.e. {(227 )+{228)}:{(229)+{2 30)} 1 :1 .1 6 (!) n-BuLI, TH F ; (ii) N i(C0 D )2/pyrldine, then reflux for 2 hours; (iii) AC2O , -7 8 ”C

3-phenyl-2-cyclohexen-1-ol (161) had previously been shown to isomerise well using the Ni(COD)2/b/s(oxazoline) catalyst, affording the corresponding 3-phenyl-1 -cyclohexanone in 83% yield, one of the highest yields obtained so far. It was necessary to probe this reaction further by trapping the enolate out with acetic anhydride and investigating the regiochemical outcom e of the reaction. Pyridine and triphenylphosphine were used as ligands (Figure 181).

139 Results and Discussion

(i) n-BuLi

(ii) N i(C0 D )2/Ligand, reflux, 2 hours

(iii) ACgO. -7 8 °C (161) Figure 181 O A c O Ac

+

Ligand Yield (168) Yield (231) + (232) (231) : (232) Ratio

Pyridine 4 8 % 2 9 % 2.0 : 1

PhaP 4 4 % 2 2% 2.1 : 1

Using either pyridine or triphenylphosphine as ligand offered the same relative quantities of ketone and enolate product, with a combined yield of 77% and 6 6% respectively. The form ation of significant quantities of ketone despite the addition of 10 equivalents of acetic anhydride was surprising. Although the problem of ketone formation has previously been encountered, the possibility of incomplete acétylation of the enolate cannot be ruled out.

The lack of regio control exhibited by the Ni(COD)2/pyridine catalyst, but not with the NiCl2(Cy3P)2/n-BuLi catalyst, clearly indicates that these two catalysts operate via different mechanisms. Deuterium labelling techniques have been em ployed as m echanistic probes to determ ine whether or not the isomérisation mechanism is operating via m etal-hydride addition and elim ination (See Mechanistic Studies).

Mechanistic

Studies

141 Results and Discussion

Introduction

A ttem pts to introduce asym m etry into the product of an isomérisation by the use of chiral transition metal câtalysis had failed, despite the range of chiral ligands employed. Following the result whereby optically pure lithium enolate had racemised under the conditions of the isomérisation, it was clear that enantioselective isomérisation would not be possible, under the existing conditions.

It is notable that the isomérisation of A/,A/-diethyigeranyiamine using the chiral rhodium cataiyst, [Rh('S>BINAP)]'^ can proceed over a range of temperatures, from 0 to 80°C, and still achieve high levels of enantioselectivity.^^° It was of some concern that the observed racém isation with the Ni(COD)2/iigand catalysts was due to the relatively high reaction tem perature (ca. 65°C). However, attem pts to isomerise the aikoxide of geraniol (61) at room tem perature met with little or no success, neither using benzene as solvent at room tem perature which gave the enolate product with high stereocontrol, but with no enantiomeric excess. Repeating the same reaction at 60°C produced sim ilar results, but the greater yield of enolate implies a need fo r activation energy for the isomérisation process.

Although the racémisation experim ent offered no explanation for the absence of any enantiocontrol the reason why racemisaton should occur at all, regardless of the chiral environm ent imposed by the ligands on the nickel metal centre, was not very clear.

A possible clue to the problem could rest with elucidating the mechanism through which the isom érisation process is thought to proceed (Figure 68, 69). If the isomérisation was operating

via m etal-hydride addition and elim ination, then the enantioselective step would be the first step when the m etal-hydride adds across the prochiral olefin in the allylic moiety. The second step would be elim ination to reform the m etai-hydride and to generate the enoiate product. The reversible nature of this m echanistic pathway is such that the metal-hydride may add and then elim inate several tim es on the same substrate molecule during the course of the reaction. The result is a decrease in the enantioselective capability of the chiral nickel cataiyst. 7i-A ilyl and enone m echanism s (See Figures 142 and 143) are more likely to ,offer asym m etric induction due to the greater binding and rigidity of the substrate-catalyst interm ediates during these two processes.

Deuterium labelling experim ents have been shown in the past as m echanistic probes, specifically to differentiate between the metal-hydride addition and elim ination mechanism, and the n-ailyl and enone m echanisms. Replacement of the carbinol hydrogen with deuterium allows the fate of the abstracted deuteride to be determined. A mechanism operating via m etai-hydride addition and elim ination is likely to distribute the deuterium within the vicinity of the ailyl moiety, whereas

via a 7i-aliyl or enone m echanism the deuterium will be delivered only to the (3-position in the enolate product (Figure 182).

142 Results and Discussion

X

Via Metal-Hydride

Addition and Elimination

Via Tc-allyl or

enone mechanism

Figure 182

This technique of deuteration and analysis of the deuterium distribution in the isom erised product has been em ployed by several groups. Both Fe(CO)s^^'^ and [Rh(DIPHOS)]^ on alcohols (70) and (233) respectively, were shown to operate via 1,3 deuterium migration, and hence via a n-

allyl or enone mechanism, which is in contrast to the hydridoplatinum catalyst, H Pt(Ph3P)2(acetone), which isomerised the deuterated allyl methyl ether (234) with significant distribution of the deuterium, indicative of metal-hydride addition and elim ination (Figure 183).^^^

^ O H D D (70) O H Fe(GO)5 [Rh(DIPHO S)]^ D (235) Po-85 O M e (234) HPt(Ph3P)2(actetone) Figure 183