Cálculo de la utilidad de los pronósticos

Geraniol (61) possesses the asym m etrical terminal disubstitution required fo r the creation of a new chiral centre upon isomérisation (Figure 99), as well as a second double bond in addition to the allylic m oiety. A successful isomérisation of the trisubstituted allylic alcohol m oiety on geraniol (61) would therefore be a step closer to achieving asym m etric catalysis.

Chiral Catalyst

(61)

Figure 99 : Asymmetric isomérisation of geraniol (61), creating a new chiral centre (*).

The isom érisation of the alkoxide was conducted using a variety of different solvents, i.e. refluxing THF, benzene and toluene heated to 60°C. In each case analysis by GC of aliquots taken from the reaction mixture during the course of the isom érisation indicated the form ation of the isomerised product citronellal (62). However as the reaction progressed, TLC and GCMS indicated the form ation of the transfer hydrogenation product, citronellol, in addition to several other unidentifiable products. Although isolation of the products was not attempted, GC analysis of the crude reaction m ixture indicated a trace amount of citronellal (62) using 0.05 m olar equivalent of the NiCLlDIPHG Sj/LiBEtsH catalyst, and approxim ately 22% of citronellal (62)

76 Results and Discussion (i) n-BuLi (ii) N iC y O IP H O S ) / LiBEtgH

₊

+

(112)

Catalyst Solvent (Reflux) Reflux Tim e (62) (61) (112)

5 mol % THF 20 h Trace 56% 28%

20 mol % THF 3 h 22% 38% 13%

5 mol % Benzene 2 h 13% 18% 18%

5 mol % Toluene (60°C) 1 7 h 7% 10% 21%

The results were slightly disappointing as geraniol had previously been isomerised to the corresponding enol acetates in 69% yield using 10 mol% of the {Cy3P)2NiCl2/n-BuLi catalytic s y s t e m . O u r attention then turned to a different prochiral allylic alcohol, namely ("E)-3-phenyl-2- buten-1-ol (63). This alcohol may be prepared in two steps following literature procedures and involve an initial a Reformatskii reaction between acetophenone and ethyl bromoacetate,^^®’^^® follow ed by the reduction of the ester (113) to the allylic alcohol (63) using diisobutylalum inium hydride (Figure 101).

pA

+

Figure 101 : Preparation of CE)-3-phenyl-2-buten-1-ol (6 3) (I) Zinc, PhH, reflux; (ii) DIBAL, toluene, -7 8 °C

Ph O H

(63)

Prior to using this prochiral alcohol in our catalytic system we attem pted the isomérisation on the structurally analogous compound, (E)-3-phenyl-2-propen-1 -ol, or cinnamyl alcohol (58)

(Figure 102, Equation 1). However, despite numerous attem pts on this substrate, no reaction took place. It was therefore no surprise to find that the isomérisation of the (EJ-3-phenyl-2-buten-

77 Results and Discussion

No Reaction

(1)

^

No Reaction

(2)

F ig u re 102 : (i) n-BuU; (ii) NICl2(DIPHGS)/UBEt3H, reflux 10-18 h; (iii) NH4CI(aq).

From a therm odynam ic standpoint, the form ation of the enolate does not apparently com pensate for deconjugation of the styrene chromophore.

During the isomérisations of secondary allylic alkoxides, a dark brown-black solution or suspension was observed, which was m aintained throughout the period of reflux. This was in contrast to observations made during the isomérisation of prim ary allylic alcohols in which a gradual loss of colour finally gave rise to a yellow solution containing a white sediment. It is evident that during the isomérisation of these substrates, deactivation of the catalyst is occurring. This could be achieved via decarbonylation of the alkoxide-catalyst com plex giving rise to an inactive Ni-CO s p e c i e s . ^ S u c h a process has been reported by Y-Lin and X-Lu,^^^ who were able to isolate the compound c/s-Mo(CO)2(DIPHOS)2 following the isomérisation of prim ary alcohols with the catalyst Mo(N2)2(DIPHOS)2. Such a deactivation is not in fact uncommon.^^'^^^’^^®

The problems encountered with the NiCl2(DIPHO S)/LiBEt3H catalyst, and in particular its decom position with primary allylic alkoxides, meant that its potential fo r evolution as an asym m etric catalyst was in some doubt. The use of reducing agents, such as />BuLi or LiBEtgH also m ade it difficult to predict what the actual active species was, be it a nickel(O) or a nickel(l) species, or perhaps a combination of the two. Attem pts to rationalise the events occurring during this ‘activation’ are made more form idable as n-BuLi can activate NiCl2(Cy3P)2, but not NiCl2(DIPH0S), whereas 'PrMgBr and LiBEt3H can.

It was considered that treatm ent of NiCl2(Cy3P)2 (81) with 1 equivalent of n-BuLi m ight have given a chlorohydridonickel(ll)b/s(tricyclohexylphosphine) species (82) via p-hydride elim ination of the form ed o-alkylnickel com plex (114) (Figure 103).

78 Results and Discussion C l \ / P 0 V 3 " x / ' ' % Ni NI NI CVaP 1 C ) / / Cy^P^ (81) / (114) (82) LICl Figure 103

Such a process has been shown by W ilke to occur upon treatm ent of 7c-allylnickel bromide with ethyllithium /^^ However, the presence of a p-hydrogen on the alkyllithium reagent is not a prerequisite for the form ation of an active isomérisation catalysts, as demonstrated by the addition of MeLi, which does not posses a p-hydrogen, to NiCl2(Cy3P)2 which furnished an active catalyst able to isomerise 1-phenyl-4-penten-3-ol (102) in 63% yield. The picture is further com plicated by Otsuka’s observation that the alkyl metal com plex, NiBr(n-C6H i3)(Ph3P)2, decom poses to the nickel(l) complex NiBr(Ph3P)3 upon heating.

It is known that the addition of two equivalents of methyllithium to dichloronickel(ll)b/s(phosphine) com plexes in the presence of an arene leads to a (7c-arene)nickel(0)b/s(phosphine) species, follow ing the reductive elimination of ethane from the b/s(alkyl)Ni(ll) intermediate (Figure 104).

civ .PCVa NI / C y ,P/ 'c i \ y (i). (ii) CvjP

Figure 104 : (l) M e U (2 equlv.), T H F , -78°C ; (II) hv, ca. -5 0 °C or A, 2 5 °C , 1-3 hours

Therefore the alkyllithium reagents may generate an active isomérisation species in two ways, either by the form ation of a chlorohydridonickel(ll) intermediate as depicted in Figure 103, or by reductive elim ination of an alkane from b/s(alkyl)Ni(ll) intermediate to form a N i(0 ) species, as depicted in Figure 104.

The procedure to prepare this Ni(0) species using MeLi was only applicable to monodentate phosphine ligands, and not to bidentate ligands, such as 1,2-/?/s(diethylphosphino)ethane. The lack of reactivity exhibited by these nickel-b/s(phosphine) com plexes to alkyllithium reagents was also observed following the addition of n-BuLi to the NiCl2(DIPH0S) com plex, which failed to isomerise the alkoxide of 1-phenyl-2-propen-1-ol (8 8), compared with near quantitative conversions using the (Cy3P)2NiCl2/n-BuLi, [Rh(DIPHOS)]^ and NiCl2(DIPHO S)/LiBEt3H catalyst systems.

79 Results and Discussion

It is not clear either whether the Grignard reagent 'PrMgBr is behaving in the same way as n-BuLi to generate an intermediary N i(ii)H C i species, or form s a diaikyinickei(ii) species, which then reductiveiy elim inates to form a Ni(0) complex.

W e therefore decided to attempt to resolve the question of whether or not a Ni(0) species was functioning as the active isomérisation catalyst by selecting the com plex Ni(C0D)2.^^^ The loosely bound COD ligands have been shown to undergo ready ligand exchange with phosphines to generate Ni(0)phosphine complexes. Therefore, Ni(C0D)2 presented itself as a suitable entry into Ni(0) chem istry whereby any ligand might be employed, particularly ligands possessing C2-sym m etry, with the intention of employing these chiral Ni(0) catalysts in our asym m etric isomérisation studies.

Isomérisations with