.
Angewandte
Communications
reactions, Montgomeryꢀs triethylborane/protic solvent proto-
col[11] allowed homogenous reaction mixtures at lower
temperatures. However, in our case, the advantage of using
chiral carbene ligands came along with two issues: 1) the
competing transesterification of the reactive phenol ester with
the added methanol gave unreactive methyl cinnamate 4, and
2) the direct 1,4 reduction gave saturated esters 5.
In order to get a first overview over the potential of the
ligand, we evaluated a set of chiral N-heterocyclic carbenes
using phenyl cinnamate and 3-hexyne as model substrates
(Table 1, entries 1–4; for the full optimization study, see the
trimethyl phenyl group (entry 8) provided the optimal
balance, giving 3aa in 64% yield and 97.5:2.5 e.r. Further
increase in the steric bulk with an even larger 2,6-diisopropyl
group shut down the formation of 3aa completely and led
almost exclusively to the formation of the 1,4-reduced
product (entry 9). Restriction of the amounts of MeOH and
BEt3 had a positive influence and the optimal ratio consisted
of five equivalents of alcohol and two equivalents of borane.
Switching to CPME (cyclopentyl methyl ether) as the solvent
allowed to reduce the reaction temperature, and the bulkier
tBuOH as proton source almost completely suppressed
transesterification and reduction pathways, thus increasing
the yield of cyclopentenone to 92%, but slightly reducing the
enantioselectivity (entry 11). Given the favorable character-
istics of the bulky IPr ligand for the cyclopentenone
formation, we explored L5, a ligand with a mixed design
between L2 and IPr. Notably, L5 gave rise to a substantially
superior reactivity and selectivity, resulting in 3aa in 92%
yield and 3:97 e.r. (entry 12).[14] Moreover, under these
conditions, the catalyst loading could be reduced to
1.25 mol% without affecting the reaction outcome signifi-
cantly (entry 13). Such loadings are still rather rare for
complex Ni0-catalyzed transformations. L5 was previously
used only once in a single trial in asymmetric catalysis,
resulting in a very poor selectivity for Pd-catalyzed oxindole
formation.[15] This outcome is in stark contrast to the excellent
results of the C2-symmetric carbene L2 for the same trans-
formation as reported by Kꢁndig. This striking performance
discrepancy for two different reactions is clearly another
testimony that one cannot conclude on the power of a chiral
ligand for a specific transformation from a few established
benchmark reactions.
Table 1: Optimization of the cyclopentenone formation.[a]
Entry L*
R
Solvent R’OH
Yield e.r.[c]
[%][b]
1[d]
2
L1 Ph
L2 Ph
L3 Ph
L4 Ph
L2 4-MeO-C6H4
L2 4-CF3-C6H4
L2 3,5-Me2-C6H3
L2 2,4,6-Me3-C6H2 THF
L2 2,6-iPr2-C6H3
L2 CH(Ph)2
THF
THF
THF
THF
THF
THF
THF
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
32
52
1
76:24
92:8
74:26
–
3[d]
4[d]
5
0
53
27
82
64
1
92:8
6
7
8
9
10
11[e]
12[e]
13[f]
14
81:19
91.5:8.5
97.5:2.5
97:3
–
94:6
3:97
2.5:97.5
2:98
THF
THF
0
L2 2,4,6-Me3-C6H2 CPME
L5 2,4,6-Me3-C6H2 CPME
L5 2,4,6-Me3-C6H2 CPME
tBuOH 92
tBuOH 92
tBuOH 71
tBuOH 60
L5 3,5-Me-C6H3
CPME
To explore the generality of the optimized process, we
evaluated different substitution pattern on the acrylic ester. A
wide range of cinnamic esters were tolerated (Scheme 2). The
aromatic portion accommodated the most common electron-
donating and electron-withdrawing groups. Irrespective of the
position of the substituents (ortho, meta, or para), the yields
and selectivities were consistently high. Notably, substrates
with condensed arenes, heterocyclic substituents such as 3-
furyl, 3-thienyl reliably provide cyclopentenones 3ba–3qa.
However, a 2-furyl group reduced both yield and selectivity,
presumably because of a chelation of the nickel center with
the oxygen atom of the furan. The absolute configuration of
the cyclopentenones was unambiguously established by X-ray
crystallographic analysis of the ferrocenyl-containing deriva-
tive 3la.[16] Alkyl-substituted acrylates, such as 1p and 1q,
undergo the annulation reaction, however with lower selec-
tivity. Next, different internal alkynes were evaluated. A
variety of dialkyl alkynes with functional groups were well
tolerated and provided the cyclopentenones with excellent
enantioselectivities. However, diaryl alkynes did not provide
the desired cyclopentenones under these conditions.
[a] Reaction conditions: enoate (0.10 mmol), 3-hexyne (0.15 mmol),
[Ni(cod)2] (10.0 mmol), L* (11.0 mmol), MeOH (0.8 mmol), BEt3
(0.5 mmol), at 708C for 17 h. [b] Yields of isolated products. [c] Deter-
mined by HPLC on a chiral stationary phase. [d] At 508C. [e] With
[Ni(cod)2] (5.0 mmol), L* (7.0 mmol), tBuOH (0.5 mmol), BEt3
(0.2 mmol), at 508C. [f] 1.25 mol% catalyst, [Ni(cod)2] (1.25 mmol), L5
(1.30 mmol), tBuOH (0.5 mmol), BEt3 (0.2 mmol), at 508C for 42 h,
79% conversion. CPME=cyclopentyl methyl ether, Np=naphthyl.
Supporting Information). The ratio 3aa/4/5 varies signifi-
cantly, depending on the chiral ligand NHC*. From the tested
carbene scaffolds, only the Kꢁndig-type ligands[13] L1 and L2
gave the desired enone product 3aa in reasonable yields and
selectivity, while many other carbenes or phosphine ligands
failed. With this initial hit, the role of the aromatic ester was
investigated next. Electron-poor aryl groups (entry 6) led to
a reduced enantioselectivity and a largely increased propen-
sity for transesterification with the methanol additive. Elec-
tron-rich aryl groups (entry 5) reacted more sluggishly and
alkyl esters (entry 10) were completely unreactive. We found
that the steric bulk of the phenol group is highly important for
the reactivity and selectivity. In this respect, a larger 2,4,6-
We next evaluated nonsymmetrical internal alkynes.
Especially for these challenging substrates, the advantage of
the “mixed-design” carbene L5 over the C2-symmetrical
carbene L2 was apparent (Scheme 3). L5 resulted in dramat-
ically improved regioselectivities, while concomitantly
increasing the enantioselectivity. Although aryl alkyl alkynes
2
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1 – 6
These are not the final page numbers!