of great importance to discover a catalyst system such that
aryl arenesulfonates can be used as synthetically useful
inexpensive starting materials to make aryl carboxylic acid
derivatives. Herein, we report such a general and efficient
catalyst system that can satisfactorily effect the carbonylation
of diverse aryl arenesulfonates.
Aryl arenesulfonates have been a difficult class of
substrates toward transition-metal-catalyzed C-C bond
formation. Successful examples of such reactions have not
been reported until recently. Buchwald’s group has reported
Suzuki-Miyaura and carbonyl enolate coupling of aryl
arenesulfonates catalyzed by Pd(OAc)2 and 2-dicyclohexyl-
phosphino-2′,4′,6′-triisopropyl-1,1′-biphenyl.5 Hu et al. also
reported Suzuki-Miyaura coupling of aryl arenesulfonates
by employing a Ni(COD)2/tricyclohexylphosphine catalyst.6
These precedents indicate a strong ligand dependency for
the activation of a C(sp2)-O bond and thus have prompted
us to conduct broad ligand screens for the carbonylation of
aryl arenesulfonates. It has been reported that bidentate
phosphine ligands were superior to monodentate phosphine
ligands in the Pd-catalyzed carbonylation of aryl or pyridinyl
chlorides due to their chelating effects on Pd(0) and therefore
stabilize the active catalyst species.3a,b Hence, our initial
screens have been focused on bidentate ligands, including
ferrocenyl phosphines, biaryl diphosphines, and phos-
pholanes, together with those well-established monodentate
ligands, such as N-heterocyclic carbenes (NHC) and Buch-
wald’s monodentate phosphines, which have proved to be
good supporting ligands for Pd-catalyzed cross-coupling
chemistry.
Figure 1. Ligand screening for Pd-catalyzed carbonylation of
p-tolyl p-fluorobenzenesulfonate. Either 4.4 mol % of bidentate
ligands or 8 mol % of monodentate ligands were used. HPLC yield.
As the model substrate for our study, we chose p-tolyl
p-fluorobenzenesulfonate, which bears a fluoro group and
therefore is more reactive than p-tolyl tosylate. Some selected
screening results are summarized in Figure 1 with the
corresponding ligand structures depicted at the bottom. Of
approximately 100 ligands screened, only a handful of
ligands showed some carbonylation activity (Figure 1). The
Josiphos ligand 1 was discovered to be the most active one
that afforded the carbonyl product in 74% yield after 12 h
at 135 °C. The next best ligand is binapine, 7, which gave
28% yield under the same conditions. The Josiphos ligand
2, which was reported by Beller et al. to be an excellent
supporting ligand for Pd-catalyzed carbonylation of aryl
chlorides,3b only afforded 13% yield. Ligand 14 (dppp),
which is effective for the carbonylation of 4-acetylphenyl
tosylate,4 showed no reaction for our model substrate. A few
monodentate ligands also showed some levels of carbony-
lation activity. Among them, the phosphoramidite ligand 8
is the most effective, affording the product in 23% yield,
whereas the Buchwald ligand 9 and NHC 11 gave ∼6%
yields. It is generally believed that the catalytic cycle for
carbonylation of an aryl halide first involves oxidative
addition of the aryl halide, followed by CO insertion and
alcoholysis of the acyl palladium complex.1 The character-
istics of bulky and electron-rich structures of the best ligands
in Figure 1 indicate that electron richness and bulkiness are
two favorable properties to facilitate the carbonylation
process. However, ligands having these two properties are
not all effective, as demonstrated by the Josiphos ligand
family; analogues 2-5 did not perform as well as ligand 1.
In our initial screens, we tested a variety of solvents and
bases with n-heptanol as the nucleophile for a broad range
of ligands. For ligand 1, the solvent and base effects are
summarized in Table 1. Polyethylene glycol (PEG) 400,
diphenyl ether, and o-xylene were all effective for the
reaction, but the use of PEG 400 resulted in the highest yield
(entries 1-5). In contrast, aprotic solvents such as DMAc
and NMP were less effective (entries 2 and 3). A number of
organic or inorganic bases, for example, DABCO, EtN(iPr)2,
NaOAc‚3H2O, NaOAc, KF, and KOtBu, were effective for
the reaction. With PEG 400 as solvent, NaOAc‚3H2O was
found to be the most effective one (entry 13). After
identifying these improved conditions, we switched the
solvent to ethanol, which also acted as the nucleophile and
is more commonly used to make carboxylic esters. Ethanol
proved very effective; with NaOAc‚3H2O or NaOAc as the
base, yields of >90% were obtained (entries 14 and 15).
Next, more reaction parameters, such as the ligand-
palladium (L/Pd) ratio, temperature, and CO pressure, were
further investigated (Table 2). Although the reaction has not
been completely optimized, it can be accomplished with
(5) Nguyen, H. N.; Huang, X.; Buchwald, S. L. J. Am. Chem. Soc. 2003,
125, 11818-11819.
(6) Tang, Z.-Y.; Hu, Q.-S. J. Am. Chem. Soc. 2004, 126, 3058-3059.
5162
Org. Lett., Vol. 8, No. 22, 2006