COMMUNICATION
1
rotation would favorably result in anti-1 (relative configura-
tion of amidine against R5 group) to avoid steric repulsion
between the R5 group and the resulting produced amidine
moiety. By contrast, when the R4 group is sufficiently large,
a steric repulsion between the R4 and the R5 groups would
result in inward selective cycloreversion to afford syn-1. The
R5 group directed toward the amidine moiety would signifi-
The H NMR spectrum of 1ca in [D6]DMSO at 308C re-
vealed that protons of two methyl substituents on the benzo-
sultam are observed as two independent singlet signals.
Thus, the two methyl substituents are diastereotopic and a
significant chirality on the NMR time scale exists at this
temperature (Figure 2). The observed chirality would, as ex-
À
pected, be derived from the rotation barrier around the C2
À
cantly increase the rotation barrier around the C2 C3 axis.
C3 bond, since the crystal structure[11] of 1cb indicates a
highly hindered environment around this axis (Figure 3) and
the dihedral angle of two double bonds is 90.28. Dynamic
NMR study revealed the coalescence temperature (Tc) of
two methyl peaks to be 748C, and DG°
was calculated to
748C
be 17.6 kcalmolÀ1 (entry 3). In contrast, no atropisomers of
anti-1aa and anti-1ba were detected even at À908C.
Next, we investigated the effects of the Ar1 and the Ar2
groups, which are substituents on amidine nitrogen and b-
carbon, respectively, on the rotation barrier (Table 1, en-
tries 3–12). Steric bulkiness of the Ar1 group affects the
atropisomerism. Namely, the coalescence temperature in-
creased 78C when the para-CF3 group was moved to the
meta-position (entries 4 and 5). Furthermore, introduction
of a substituent on the ortho-position of Ar1 significantly in-
fluences the rotational barrier of a,b-unsaturated amidine
(entries 6 and 7). In contrast, the substituent effect of Ar2 is
not so simple. Introduction of a bulky substituent on the
ortho-position apparently decreases the rotational barrier;
replacement of the phenyl group in 1cb with 1-naphthyl
group lowers Tc as much as 208C (entry 8). Comparison of
the X-ray structures of 1cb and 1cf revealed that, while the
phenyl ring seems to conjugate olefin and both p-systems
appear planar in 1cb, the naphthyl ring is skewed from the
olefin plane in 1cf (Figure 3). Interestingly, the electron den-
sity of the Ar2 group is rather important. Introduction of an
electron-donating substituent on the Ar2 group afforded
higher Tc than that of an electron withdrawing CF3 or NO2
substituent (entries 3 vs 4, 9 vs 10). These observations indi-
cate that conjugation of olefin with the Ar2 group would be
important. Finally, we found that the five- membered heter-
oaromatic ring as a R2 group significantly hindered the rota-
tion of the axis (entries 11 and 12).
Scheme 1. Our synthetic strategy toward a,b-unsaturated amidine 1: A
proposed torquoselectivity in thermal cycloreversion of 2.
Our initial exploration of the formal aza–enyne metathe-
sis of nitrogen-substituted alkynes with imines indicated that
ynamides 3 bearing a sulfonyl moiety and N,C-diarylaldi-
mines 4 are suitable substrates, owing to the chemical yields
of the products and easy handling of both substrates.[9]
Moreover, the reaction was effectively catalyzed by triflic
imide (Tf2NH).[10] The results of the reaction are summar-
ized in Table 1. The reaction proceeded smoothly at ambient
temperature to 608C to give a,b-unsaturated amidines in
moderate to excellent yield. In all cases, no azetine 2, a reac-
tion intermediate, was detected. This result indicates that
the electrocyclic ring-opening step is much faster than the
initial [2+2] cycloaddition.
Although a,b-unsaturated amidines 1ca–cj (Table 1) dis-
play axial chirality within the NMR time scale, it was impos-
sible to separate their enantiomers by chiral HPLC under
any conditions tested. As far as molecular modeling based
on the obtained conformation from the X-ray crystallogra-
Compounds 1aa and 1ba, which were synthesized from
terminal alkyne 3a and phenylalkyne 3b, respectively, were
obtained as a single geometrical isomer in which the p-tolyl
substituent on the C(3) position and the amidine moiety are
on opposite sides of the double bond (entries 1 and 2 in
À
phy, the rotation of the C2 C3 axis might be completely
1
Table 1). Their configuration was assigned by H NMR and
suppressed owing to the bulky substituents. We considered
that rapid racemization of 1 would be caused by the E/Z iso-
merization about the amidine C=N double bond. In the
ground state E configuration of the C=N double bond, the
C2-C3 axis seems to be “locked” by the E-oriented N2-aryl
group. However, the N2-aryl group can be partially “un-
locked” by isomerization of the amidine moiety (Scheme 2).
We considered that large ortho-substituents on the N2-
aryl group would prevent this “unlocking” by steric repul-
sion between the N1-benzosultam group. To synthesize
more atropisomerically stable a,b-unsaturated amidines,
X-ray crystallography (see Supporting Information). On the
other hand, to our delight, when the terminal ynamide was
substituted with the bulky triisopropylsilyl (TIPS) group, we
found that the geometrical selectivity was completely re-
versed. Thus, the reaction of ynamide 3c with 4a gave exclu-
sively syn-1ca; its configuration was confirmed by X-ray
analysis (entry 3). Only the E configuration of the amidine
C=N double bond was observed in all cases. The 1,3-allylic
À
strain of the amidine moiety (N=C N) caused the stereose-
lection.
Chem. Eur. J. 2009, 15, 7026 – 7030
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7027
Shindoh, Naoya
