Angewandte
Chemie
Scheme 2. Decarboxylative cross-coupling of 2-naphthyl mesylate.
In the Stille reaction, where the transmetalation step is
often limiting, it has been found that the use of tri-2-
furylphosphine rather than triphenylphosphine enhances
reaction rates by two orders of magnitude.[24,25] This ligand
features a potentially coordinating heteroatom in close
proximity to the phosphorus, which enhances p-back-dona-
tion. It has also been reported that only one phosphine should
be present in transmetalations to palladium.[26]
Scheme 3. Study of ligand performance in the decarboxylative coupling
of aryl mesylates. Reaction conditions: 1a (0.3 mmol), 2a (0.6 mmol),
[Pd(acac)2] (5.0 mol%), L (7.5 mol%), Cu2O (2.5 mol%), 1,10-phenan-
throline (5.0 mol%), NMP (2.0 mL), 1708C, 16 h. Yields determined by
GC analysis using n-tetradecane as the internal standard. Cy=cyclo-
hexyl, iPr=isopropyl, tBu=tert-butyl. [a] 50% yield of protodecarbox-
ylation product was observed.
As a lead structure for ligand design, we thus chose
a benzimidazolyl skeleton rather than the indolyl system of
CM-Phos. This ligand class, which was first reported by
Altenbach[27] and has recently been used by Kwong in Suzuki
couplings,[28] should have reduced s-donor strength, offer
a weakly coordinating group that may aid an associative
transmetalation step, and provide enhanced p-back-donation.
We synthesized various imidazolyl phosphines by a sequence
of cyclization, alkylation, and phosphonation reactions,[29] and
systematically evaluated their activities in decarboxylative
cross-couplings of aryl mesylates (Scheme 3).
In the decarboxylative cross-coupling of 1a with 2a, the
new ligands L1–L15 directly proved to be effective. A clear
trend was evident for the influence of the phosphorus
substituents (L1–L3). The moderately bulky, electron-rich
dicyclohexyl-substituted phosphine showed the highest activ-
ity (L3), whereas tert-butyl and phenyl groups were less
effective. As a substituent on the imidazole nitrogen (R3),
more sterically demanding groups such as isopropyl (L4),
phenyl (L5), and methoxymethyl (L6) had a beneficial effect.
When the substitution on the phosphorus-bound aryl ring was
increased using alkoxy groups in the manner of a Brett-Phos
ligand (R2 = OMe, L7), the yields dropped. Increasing the
electron density of the benzimidazole nitrogens by methyl
substituents (R4) on the ring improved the yields (L10). A
tetrahydrobenzimidazole fragment (L11), which would also
have a high electron density on the nitrogens, was nearly as
effective as L10. Further optimization of the N-substituents
(R3) on the final ligand core led to another step-up in the
yields (L12–L15), with n-octyl being the most effective
residue.
À
(2.042 ꢁ) than the Pd Cs bond in the Pd–CM-Phos complex
(1.986 ꢁ),[16] confirming the targeted labile chelating coordi-
nation mode.
The reaction conditions were systematically optimized
using the solid ligand L10 (Table S1 in the Supporting
Information). A step-up in yields to 68% was achieved by
changing the solvent to a 1:3 mixture of NMP and mesitylene
and changing to [Pd(dba)2] as the Pd precursor. Further
experiments confirmed that [Pd(dba)2] and Cu2O are the
most effective precatalysts. Silver decarboxylation co-cata-
lysts can also be used but are less effective. Among the
N ligands, the best results were obtained with 3,4,7,8-tetra-
methyl-1,10-phenanthroline (Me4-phen). Control experi-
ments revealed that the decarboxylative coupling reaction
requires both palladium and copper to proceed. Under these
optimized conditions, the use of ligand L14, a viscous oil, still
made a positive difference. Up to 79% yield was obtained
with the catalyst generated in the reaction solution and almost
quantitative yields were obtained using the preformed PdCl2–
L14 complex. The reaction is easy to perform also in
a laboratory microwave reactor.
We next investigated the scope of the new transformation.
Various aromatic carboxylate salts were successfully coupled
with 2-naphthyl mesylate using a catalyst system generated in
situ from [Pd(dba)2] (5 mol%), L14 (12 mol%), Cu2O
(2.5 mol%), and 3,4,7,8-tetramethyl-1,10-phenanthroline
(5 mol%) within 30 min at 1808C in a laboratory microwave
reactor. As illustrated in Table 1, various ortho-substituted
arenecarboxylate salts were converted into the corresponding
biaryls in high yields (3aa–3ea). Even the highly sterically
demanding substrate 1g gave satisfactory yields. Potassium
ortho-methoxybenzoate was converted only in low yield,
while non-ortho-substituted benzoates did not react (3ha).
In the crystal structure of the optimal catalyst system
(Figure S1 in the Supporting Information),[30] the palladium
center is coordinated by both amino and phosphino groups in
À
a distorted square–planar geometry. The Pd N bond is longer
Angew. Chem. Int. Ed. 2013, 52, 2954 –2958
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2955