6758
the cycloaddition between 2,5-norbornadiene (1) and phenylacetylene (2) (Scheme 1). Reducing
agents such as manganese, Devarda reagent, lithium naphthalenide, electrochemical reduction
(carbon electrode), borohydrides, and aluminium hydrides were used, but to our surprise, even
for prolonged reaction times, no or little product could be detected and only in the cases where
the original elemental zinc or alanes (DIBAL) were used could the desired product be isolated in
reasonable yields.
Scheme 1.
We therefore assumed that the addition of a Lewis acid as a cocatalyst to the reaction mixture
might be useful. When zinc iodide was used as an additive, the reaction of 1 and 2 proceeded
smoothly with lithium triethoxyaluminiumhydride (LiAlH(OEt)3) or sodium borohydride
(NaBH4) as reducing agent. In addition, even cobalt bromide and cobalt chloride triphenylphosphine
complexes, unreactive under the original conditions, catalysed the cycloaddition between 1 and 2
after the addition of several equivalents of zinc iodide.
In a series of reactions with dierent ratios of zinc iodide and the cobalt(II) complex, it was
found that the best results were obtained in the presence of 3.0 mol equivalents of ZnI2, while at
higher catalysts to the zinc iodide ratios no further increase in reactivity was observed. The
reaction also showed an induction phase (colour changed from green towards a brown solution)
whose length was dependent on the reducing agent used. While zinc had a somewhat long
induction period, the colour change was observed within minutes for lithium triethoxy
aluminiumhydride and immediately for tetrabutylammonium borohydride (Bu4NBH4), which is
the best soluble reducing agent in dichloromethane used in this series of experiments. We
therefore chose a system consisting of CoBr2(dppe), ZnI2, and Bu4NBH4 as a catalyst mixture for
homo Diels±Alder reactions. As expected, the reaction of 1 with various alkynes was successfully
catalysed by our catalyst system and the products could be obtained in comparable good yields
and purities (Scheme 2).4
Scheme 2.
Conversely, the homo Diels±Alder reaction failed for other cyclic 1,3- and 1,4-dienes (such as
1,3-and 1,4-cyclohexadiene, 1,3- and 1,5-cyclooctadiene) with 2a as dienophile. Such reactions
are to our knowledge also not described for the original catalyst systems (CoI2/PR3/Zn) and
(Co(acac)3/PR3/Et2AlCl). However, the reaction of acyclic 1,3-dienes like 2,3-dimethyl-1,3-butadiene,