Angewandte
Chemie
versions are often impeded by high reaction temperatures and
the use of a monodentate ligand imposed by the concerted
metalation–deprotonation (CMD) mechanism.[15] The pro-
cess we envisioned would require the enantioselective
formation of a rare seven-membered palladacycle[16] of an
unbranched substrate (2), thus significantly expanding the
boundaries of the current methodology. Among the very few
reported catalytic asymmetric processes, all examples are
limited to six-membered palladacyclic intermediates.[17]
however, the enantioselectivity is modest (Table 1, entry 2).
In contrast, SagePhos (L2), a ligand we previously introduced
À
for related C H functionalizations of aryl triflates leading to
indolines,[17c] showed a very low reactivity with aryl bromides
(Table 1, entry 3). We then turned our attention toward less-
electron-rich phosphoramidites. While binol-based ligand L3
did not promote the reaction at all (Table 1, entry 4), taddol-
derived ligand L4 gave excellent reactivity and selectivity,
providing 3a in 88% yield and e.r. = 94.5:5.5 (Table 1,
entry 5). A careful choice of aromatic substituents on the
taddol backbone is critical for reactivity and selectivity.[19]
Whereas the reactivity was very low with simple phenyl
groups (L5; Table 1, entry 6), bulky 3,5-di-tert-butylphenyl
groups (L6) are detrimental to the enantioselectivity
(entry 7). In conjunction with a dimethylamino group, the
3,5-xylyl backbone (L8) gave 3a almost quantitatively with
e.r. = 96:4 (Table 1, entry 9). As we have previously demon-
strated,[17c] the carboxylic acid co-catalyst plays a critical role
on the enantioselectivity of the activation by relaying the
chiral information of the phosphine ligand during the
selectivity-determining CMD step. Without acid additive,
the reaction proceeds sluggishly and with reduced selectivity
(Table 1, entry 11). Contrasting our previous findings with
aryl triflate substrates, bulky acids that bear aromatic groups
are not as efficient, leading to incomplete conversions with
2 mol% of palladium catalyst (Table 1, entries 12–15),
whereas aliphatic acids give rise to excellent yields (entries 16
and 17). Among them, pivalic acid displays the best enantio-
selectivity. Without affecting the reaction outcome negatively,
the catalyst loading could be reduced to 1 mol% (Table 1,
entry 18). Notably, such a low catalyst loading is quite
The general feasibility of our hypothesis for the cyclo-
3
À
propane C(sp ) H bond functionalization was proven using
aryl bromide 1a and tricyclohexylphosphine as achiral ligand
(Table 1, entry 1). The desired tetrahydroquinoline 3a was
obtained in excellent yield at 1308C. For the development of
an efficient asymmetric version, we first investigated related
chiral and electron-rich monodentate phosphines. P-Alkyl
phospholane L1[18] worked well with 2 mol% catalyst loading,
[a]
À
Table 1: Optimization of the enantioselective C H arylation of 3a.
À
uncommon in the field of C H functionalization and dem-
onstrates the efficiency of our catalytic system.
To explore the generality of the optimized process, we
evaluated different substitution patterns on the cyclopropane
ring. Several aliphatic and functional groups are tolerated
(Table 2, entries 1–3). The yields and selectivities are consis-
Entry Carboxylic acid
L
Yield [%][b]
e.r.[c]
[d,e]
1
2
3
4
5
6
7
8
pivalic acid
pivalic acid
pivalic acid
pivalic acid
pivalic acid
pivalic acid
pivalic acid
pivalic acid
pivalic acid
pivalic acid
no acid
PCy3
L1[e]
L2
L3
L4
L5
L6
L7
L8
95
95
–
tently excellent. A phenyl and a benzyl group do not react in
35.5:64.5
–
–
94.5:5.5
–
62:38
94.5:4.5
96:4
2
À
a competing C(sp ) H activation (Table 2, entries 4 and 5).
<5[f]
<5[f]
88
With the unsubstituted cyclopropane 1g, exclusively methine
À
C H activation occurs and gives rise to spirocycle 4 (Table 2,
<5[f]
91
entry 6). The better accessibility of the six-membered palla-
dacycle clearly overrides the preference for the less-substi-
97
98
À
tuted C Pd bond. In contrast, silyl-substituted cyclopropane
9
1h cleanly gives 3h, offering further opportunities for
functionalization (Table 2, entry 7). The aromatic portion of
the substrates tolerates the most common electron-donating
and electron-withdrawing groups (Table 2, entry 7–14). Both
meta and para substitution with respect to the bromine atom
have little influence on the reaction efficiency. Notably,
tetrasubstituted substrate 1p with a fluorine atom in ortho
position to the nitrogen atom that disturbs the proper
orientation of the cyclopropyl moiety most, is also tolerated
(Table 2, entry 15). However, a somewhat lower enantiose-
lectivity is observed for 3p.
10
11
12
13
14
15
16
17
18
L9
L8
86[f]
58[f]
59[f]
75[f]
62[f]
38[f]
92
95:5
82:18
84.5:15.5
84:16
88.5:11.5
90:10
91.5:8.5
95:5
9H-xanthene-9-carboxylic acid L8
triphenylacetic acid
L8
L8
L8
L8
2,2-diphenylpropanoic acid
anthracene-9-carboxylic acid
cyclohexylcarboxylic acid
adamantylcarboxylic acid
pivalic acid
L8
94
95
L8[g]
96:4
[a] Reaction conditions: 1a (0.10 mmol), carboxylic acid (30.0 mmol),
[Pd(dba)2] (2.00 mmol), L (3.00 mmol), Cs2CO3 (1.5 equiv), 0.30m in p-
xylene at 1308C for 12 h. [b] Yields of isolated products. [c] Determined
by GC on a chiral stationary phase. [d] With Pd(OAc)2 (5 mol%) and PCy3
(10 mol%). [e] Used as HBF4 salt. [f] Incomplete conversion. [g] With
[Pd(dba)2] (1 mol%). Cy=cyclohexyl, dba=trans,trans-dibenzylidenea-
cetone.
To showcase the practicality of the process, we carried out
a gram-scale reaction using substrate 1e and a catalyst loading
of 1 mol% palladium (Scheme 2). The desired product 3e was
Angew. Chem. Int. Ed. 2012, 51, 12842 –12845
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim