Journal of the American Chemical Society
Communication
-neutral (3c, 3d), and -deficient (3e, 3f) substituents on the
aryl group had a minimal effect on the enantioselectivity.
Arenes with strong electron-withdrawing groups (3e, 3f) did
suffer a decrease in regioselectivity. This is expected given the
similar electronic natures of the two groups (e.g., −CN vs −4-
C6H4CF3, 3e). Both ortho (3g) and meta (3h) substituents
were tolerated. Both electron-rich (3i) and electron-deficient
(3j) heterocycles could be incorporated. Beyond alkynonitriles,
alkynoesters (3k−m), a trifluoromethyl alkyne (3n), and a
thioalkyne (3o) also provided the corresponding E-NiAAC
products with good selectivity.
The scope of allylic azides was explored (Table 3). The
pendent aryl tolerated both electron-rich and electron-deficient
a
Table 3. Azide Scope of E-NiAAC
Figure 1. Initial E-NiAAC mechanistic investigation. (A) Reaction
progress plots showing the er of 1a and 3a vs time (linear fit), the
conversion of 1a vs time (logarithmic fit), and the yield of 3a vs time
(logarithmic fit). (B) Plot of ligand L5 ee vs product 3a ee (linear fit).
(C) Plot of the relative catalyst loading vs the relative initial rate,
normalized to a catalyst loading of 5 mol % (linear fit).
a
Yields are reported for isolated and purified products. Regioisomeric
ratios (rr) were determined by SFC analysis. Enantiomeric ratios (er)
were determined by SFC with a chiral stationary phase. Yields, rr
values, and er values are averages of duplicate trials. See the
nonlinear effect (Figure 1b), indicating that only 1 equiv of
ligand L5 is involved in the enantiodetermining step. Lastly,
the reaction was found to be first-order in catalyst by initial
rates (Figure 1c).74 This result is consistent with the DFT
calculations on NiAAC18 and contrasts with the higher order
with respect to the catalyst for the CuAAC reaction.48
Together, the data in Figure 1b,c indicate that NiAAC likely
proceeds through a mononuclear L5−Ni complex. This is
distinct from CuAAC, which is thought to proceed through
[L2Cu2X2] dimers or higher-order oligomers.
In conclusion, ligand L5 can promote an E-NiAAC reaction
by DKR. This is the first enantioselective alkyne−azide
cycloaddition that is reported to be catalyzed by a metal
other than copper. The E-NiAAC reaction provides synthetic
access to 1,4,5-trisubstituted α-chiral triazoles that are not
directly accessible by E-CuAAC. Furthermore, preliminary
mechanistic experiments indicate that the nature of the NiAAC
catalyst is a monometallic L1 species, which is distinct from
CuAAC.
groups (3p−r). The cyclohexyl core was modified (3s−v). In
the acyclic allylic azide 1w, both the (E)- and (Z)-alkene
isomers were observed (7:1 E/Z). Interestingly, triazole 3w
was isolated as a single isomer.
A series of preliminary mechanistic experiments were
conducted to contrast E-CuAAC and E-NiAAC (Figure 1).
First, the er of azide 1a and triazole 3a were determined as a
function of reaction time course (Figure 1a). At very early time
points, a slight enrichment of the azide er was measurable (up
to 55:45), indicating that this DKR67−70 likely occurs in a
nonideal kinetic regime.71 The er of triazole 3a was constant as
a function of reaction progress, and the smooth time-course
data indicated well-behaved kinetics. Next, a nonlinear
experiment was conducted.72 E-CuAAC reactions are known
to proceed with positive55,73 or negative52,60 nonlinear effects.
In contrast, the E-NiAAC reaction did not demonstrate a
5310
J. Am. Chem. Soc. 2021, 143, 5308−5313