.
Angewandte
Communications
efficient mode of activation, a-ketoesters could be a suitable
class of allenolate acceptors. Hence, the combination of
alkynal 1a with a-ketoester 2a in the presence of lithium tert-
butoxide and IMes·Cl as precatalyst was explored (A, Table 1,
activating methyl benzoylformate toward addition. The
introduction of chiral Brønsted acid F (20 mol%) with achiral
NHC from B produced a significant increase in yield (54%),
but the annulation product was effectively racemic (entry 4).
We then turned our attention toward exploring the use of
a chiral NHC in combination with a chiral Brønsted acid
(CBA). While established chiral triazolium-based NHC
catalysts are ineffective for this transformation (see the
Supporting Information), it became apparent that C1-sym-
metric biaryl saturated-imidazolium-derived NHC catalyst C,
originally developed by the Hoveyda group,[13,14] provided the
desired product in excellent yield (entry 6). Indeed, the use of
these 2,3-dihydroimidazole-2-ylinene structures in carbene
catalysis is far less developed than their imidazolium- and
triazolium- derived counterparts. Interestingly, conducting
the reaction under more dilute conditions in the presence of
molecular sieves resulted in a further enhancement in
enantioselectivity (from 79:21 to 90:10 e.r., entry 7). With
these intriguing results in hand, we investigated various C1-
symmetric biaryl saturated-imidazolium precatalysts, with the
highest yield and enantioselectivity observed with catalyst D
(entry 9). The introduction of more sterically demanding aryl
or naphthyl groups on the chiral phosphoric acid, such as G
(i.e., (S)-TRIP) and H, or on the imidazolium salt, such as E,
resulted in decreased reactivity and enantioselectivity
(entries 10–12). Therefore, precatalyst D and chiral phospho-
ric acid F were chosen as our optimized catalysts for further
study (entry 9). The use of the (R)-phosphoric acid instead of
its (S)-enantiomer did not diminish catalyst reactivity and
enantiomeric ratio, thereby indicating a lack of an expected
match/mismatch relationship between the phosphoric acid
chirality and the NHC (entry 13). However, racemic phos-
phoric acid F provides the desired product with slightly
decreased yield and enantioselectivity (entry 14). At our
current level of understanding of this complex reaction, we
assume that unexpected inactive species might be formed
between the lithium ion and the (R)- and (S)-phosphate that
generates the product in lower yield and a slightly decreased
enantiomeric ratio. Clearly, this is a multivariable system and
further investigations to delineate more fully the specific roles
of the lithium cation, NHC, and Brønsted acid/chiral phos-
phate are ongoing. In addition, achiral phosphoric acid I can
also serve as a co-catalyst, albeit with diminished yield and
selectivity (entry 15). However, other Brønsted acids, such as
4-nitrobenzoic acid, did not provide any desired product
(entry 19).
Table 1: Optimization of reaction conditions.[a]
Entry
Azolium
Co-catalyst
Yield [%][b]
e.r.[c]
1
2
3
4
A
B
B
B
C
C
C
D
D
D
D
E
D
D
D
C
C
C
C
–
–
no reaction
–
–
–
trace
24
54
trace
90
85
34
85
85
trace
52
84
74
LiCl (1 equiv)
F
–
F
F
52:48
n.d.
79:21
90:10
89:11
93:7
91:9
n.d.
92:8
93:7
91:9
89:11
n.d.
–
5
6
7[d]
8[d]
LiCl (1 equiv)
9[d]
F
G
H
F
ent-F
rac-F
I
F
F
F
J
10[d]
11[d]
12[d]
13[d]
14[d]
15[d]
16[d,e]
17[d,f]
18[d,g]
19[d]
77
trace
no reaction
no reaction
no reaction
–
–
[a] Conditions: 1a (0.05 mmol, 1 equiv), 2a (1.5 equiv), azolium
(0.2 equiv), phosphoric acid (0.2 equiv), LiOtBu (0.4 equiv) in THF
(0.15m) at 238C for 48 h. [b] Determined by GC-MS with n-dodecane as
internal standard. [c] Determined by HPLC analysis. [d] 4 ꢀ molecular
sieves (M.S.) were used at 0.014m concentration in THF. [e] 20 mol%
LiOtBu was used. [f] NaOtBu was used as the base instead of LiOtBu.
[g] Mg(OtBu)2 was used as the base instead of LiOtBu. Entry in bold
marks optimized conditions.
To gain further information about the role of the CBA, the
reaction was performed using a decreased amount of base
(20 mol%; Table 1, entry 16). This modification provided
only trace amounts of the product. Considering the pKa values
of the azolium and Brønsted acids,[15,16] it is likely that under
these conditions phosphoric acid F would be deprotonated
initially, and the NHC might not be generated in high enough
concentrations from the azolium salt precatalyst. A second
aspect is whether the lithium cation is involved in organizing
the transition state[7f,17] or if lithium tert-butoxide was simply
acting as a base. To probe this possibility, sodium tert-butoxide
or magnesium di-tert-butoxide were employed as the base, but
no product was obtained, suggesting the involvement of the
entry 1). These conditions did not produce any of the desired
product. However, the use of saturated imidazolium
(SImes·Cl) B afforded trace amounts of butenolide 3a
(entry 2). Guided by our previous NHC/Lewis acid studies,
a slightly improved yield was observed with one equivalent of
lithium chloride (entry 3).[7f] Even though the reaction yield
was poor, this result indicated that lithium cations may be
2
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1 – 6
These are not the final page numbers!