Organic Letters
Letter
propanols because of the fast scission process. To overcome
this problem, we switched THF for MeCN, which seemed
inert to oxidation and allowed the transformation to occur,
providing dioxane 7a in 61% yield after 48 h at room
temperature (Table 1, entry 5). Increasing the temperature to
40 °C could decrease significantly the reaction time to 8 h,
improving the overall process (Table 1, entry 6). Using a
stoichiometric amount of cobalt(II) salt did not improve
significantly the kinetics of the reaction but lowered the yield
due to the appearance of some degradations (Table 1, entry 7).
Co(II) salt was essential to this reaction since starting directly
from a source of Co(III) proved to be unproductive (Table 1,
entry 8).
Scheme 3. Diverse Attempts Toward the Synthesis of 1,2-
Dioxan-6-ols 7b−g and 8a−e
Oxidative ring expansion was then applied to other
cyclobutanols 5b−f and 6a−e. Secondary alcohols 5b−f are,
in general, the best substrates for such a transformation.
Indeed, the reaction works well, with 2,2-disubstituted
cyclobutanols 5b−d and 5f, affording isolated yields above
60%. Monosubstituted cyclobutanol 7e was also able to
undergo an oxidative ring extension but required a longer
reaction time (up to 48 h). We also observed no reaction at
room temperature with this last substrate, so heating was
necessary in this case. Tertiary alcohols 6a−e were also
examined, and the first trials did not lead to good results.
Monosubstituted cyclobutanols 6a and 6e gave unsatisfactory
results: no conversion into dioxane 8e was observed from 6e.
Isolation of diketone 9a and hydroxyketone 10a in 36% and
24% yield, respectively, from cyclobutanol 6a was experienced
in place of dioxanol 8a. Isolation of these two unexpected
products means the reaction was probably working, but
compound 8a might undergo a Kornblum−DeLamare
rearrangement to form 9a, on one hand because the benzylic
position facilitates an elimination process. On the other hand,
hydroxyketone 10a might have been produced from 1,2-
dioxane 8a by homolytic cleavage of the peroxide bond, but
the exact pathway is currently unclear. In contrast, it was found
that 2,2-disubstituted cyclobutanols 5b−d and 5f were better
substrates for an oxidative ring expansion. However, the rate of
the reaction was slower compared to secondary alcohol
derivatives 6b−d, affording the desired products in lower
yields, sometimes with recovery of the unreacted starting
materials. This observation was unexpected and might be
attributed to steric hindrance for the formation of the
cyclobutyloxy radical rather than a decrease of the kinetics of
the radical scission (Scheme 3).
Although similar oxidation was reported in the literature, the
exact mechanism of how the peroxy radical species are formed
has never been determined.27 It seems plausible that
Co(II)(acac)2 is reacting with oxygen to form a
superoxocobalt(III) radical and a μ-peroxocobalt(III) spe-
cies.30 Indeed, the reaction mixture, in the presence of oxygen,
turns rapidly from a light brown color to intense green, which
is characteristic of Co(III) ions. Also, the inability of
Co(III)(acac)3 to promote the reaction (Table 1, entry 8)
indicates the crucial role of these oxocobalt(III) species.
Alkoxy radicals are less stable than peroxy radicals by about 4
kJ/mol31 and by consequence make hydrogen abstraction of a
hydroxyl group from A an endothermic process. However, the
strain ring scission which follows, with stabilization of the
radical at position 4, is highly exothermic, making the overall
process highly favorable [about −71 kJ/mol for scission of a
trans-2-methylcyclobutyloxyl radical (determined by calcula-
tion at the B3LYP/6-31+G(d,p)32 level of theory)]. Coordi-
nation of the hydroxyl group to the Co(III) species is also very
plausible during the hydrogen abstraction step, facilitated by a
directed process such as represented in B. Moreover, the
observed difficulty applying the reaction to tertiary cyclo-
butanols is consistent with higher steric hindrance of these
substrates, which strengthens the hypothesis of a coordination.
Regioselectivity in the carbon−carbon bond cleavage to C is
dependent on the ability of the new site to stabilize the new
radical species. Therefore, this driving force explains some
differences of reactivity between mono- and disubstituted
cyclobutanols. A molecule of 3O2 is then trapped by the
radical, making a new peroxyradical D, which can evolve
between the opened form E and the closed one F. One or the
other form might abstract a hydrogen atom from a new
cyclobutanol to propagate the reaction such as in path a, or a
cobalt(III) hydroperoxide species could reinitialize the process
from the beginning such as in path b (Scheme 4).
In order to prepare different 1,2-dioxanes, the so obtained
1,2-dioxanols 7a−g were then acylated using Yb(OTf)3 as a
Lewis acid catalyst to produce different acetate derivatives
11a−g.22,23 Further functionalization of these compounds was
initiated with different silylated nucleophiles. The catalytic
23
version of the nucleophilic substitution using Sc(OTf)3 was
chosen for this screening for better convenience. The reaction
of acetate 11e with allylTMS produced compound 12 as a sole
cis-product in 65% yield. In contrast, reaction of acetate 11e
with TMS cyanide was less selective and gave product 13 with
only a 65:35 cis:trans ratio,33 while Mukayama aldol reaction
with a ketene TMS-thioethylacetal furnished 1,2-dioxane 14
with a quite good diastereoselectivity. The synthesis of allyl
1,2-dioxane 15 from 11a was achieved in poor yield due to an
elimination process, which might take place more easily at the
benzylic position and produced many degradation products.22
Sakurai reactions on acetate 11b produced compounds 16 and
17 with pretty good yields. Mukaiyama reaction on acetate 11b
with silyl enol ether of acetophenone gave dioxane 18 in good
yield, and a vinylogous process was also possible to produce 19
from TMS-siloxyfuran as a 2:1 mixture of diastereomers.34
Reduction of acetal 11d with triethylsilane furnished
compound 20 in moderate yield. With unsymmetrical
dialkylated dioxanes 11c and 11d, the diastereoselectivity is
not as good as with monosubstituted compounds, but
generally a major isomer is observed such as in the synthesis
1628
Org. Lett. 2021, 23, 1626−1631