Angewandte
Communications
Chemie
by the use of enolates as nucleophiles[9] or organozinc
compounds obtained by tert-butoxycarbonyl (Boc)-directed
lithiation and Li–Zn transmetalation.[10] Herein, we show that
terminal (linear) cross-coupling products 3 can be obtained in
a regioconvergent manner in just one step from bromoalkanes
4 under operationally simple Barbier (in situ) conditions.
Furthermore, in most cases a similar selectivity pattern was
observed regardless of the position of the bromine atom in the
alkyl chain. This observation led us to employ mixtures of
bromides 4’ obtained by the radical bromination of linear
alkanes 1. Indeed, this process is known to give rise to
mixtures of secondary bromides,[11] and hence has been of
little synthetic use so far. However, we reasoned that such
a mixture could be productively utilized by applying the same
in situ metal-insertion/migrative Negishi coupling sequence:
A mixture of branched organopalladium intermediates 6’
would converge to the same linear complex 7, thereby
furnishing the same linear product 3.
At the outset of our studies, we sought suitable and
practical conditions for the generation (4!5) and cross-
coupling (5!3) of alkyl zinc reagents in situ to maximize the
overall step economy. Such direct Barbier–Negishi coupling
reactions have been reported by Lipshutz and co-workers,
who employed water as the solvent and a diamine as an
additive, either in the presence of a surfactant[12] or under “on-
water” conditions.[13] An improved catalytic system that
enables a range of cyclic bromoalkanes to be coupled with
aryl electrophiles was recently disclosed by Buchwald and co-
workers.[14] However, to the best of our knowledge, alkyl–aryl
coupling under non-aqueous Barbier conditions has not been
reported.[15,16] We reasoned that aryl triflates (Y= OTf) would
be suitable electrophiles, since they should not undergo zinc
insertion, and therefore should show orthogonal reactivity to
alkyl halides 4. Hence, we started to investigate the coupling
of aryl triflate 8a with an excess amount of the organozinc
reagent generated from 2-bromopropane (4a), zinc dust, and
LiCl in THF,[6a] in the presence of a Pd catalyst formed in situ
from [Pd2dba3] and a suitable phosphine ligand (Table 1). The
choice of the latter was guided by our previous studies on
migrative Negishi coupling reactions, wherein flexible phe-
nylazole-based ligands were found to induce the highest
selectivity in favor of the linear product.[10]
We tested a library of homemade ligands from this family
(see Table S1 in the Supporting Information) and found that
the phenyl–pyrrole-based phosphine L1 containing n-butyl
P substituents provided the highest linear selectivity, presum-
ably owing to its high degree of flexibility at both the phenyl–
pyrrole and the P–alkyl bonds. Gratifyingly, the use of Pd
(5 mol%) and L1 (7.5 mol%), which is readily synthesized in
one step from N-phenylpyrrole, led to a mixture of the linear
product 3a and branched product 9a with good selectivity
(9:1), albeit in very low yield, presumably as a result of slow
zinc insertion (Table 1, entry 1).
We thus looked for conditions that would allow the
organozinc intermediate to be generated more rapidly.
Gratifyingly, when we carried out the reaction under con-
ditions described by Knochel and co-workers with a mixture
of Mg, LiCl, and ZnCl2,[6b] the yield was greatly improved, and
the good linear/branched (l/b) selectivity was maintained
Table 1: Linear-selective Barbier–Negishi coupling: Optimization of the
reaction conditions.
Entry Organometallic reagent
(equiv)
R
Ligand l/b[a]
Yield of 3
[%][b]
1
2
3
4
5
6
7
8
9
4a/Zn/LiCl (2)
4a/Mg/LiCl/ZnCl2 (4)
4a/Mg/LiCl (4)
4b/Mg/LiCl (4)
4c/LiCl (1.3)
4b/Mg/LiCl/ZnCl2 (4)
4a/Mg/LiCl/ZnCl2 (4)
4a/Mg/LiCl (4)
4a/Mg/LiCl/ZnCl2 (4)
4a/Mg/LiCl/ZnCl2 (2)
OMe L1
OMe L1
OMe L1
OMe L1
OMe L1
OMe L1
OMe L2
CO2Et L1
CO2Et L1
OMe L1
OMe L1
OMe L1
90:10
91:9
87:13
89:11
88:12
92:8
<2:98
63:37 <10
87:13
92:8
92:8
92:8
18
73 (91)[c]
52
62
72
45
82[d]
76[c]
10
82 (83)[c]
76 (80)[c]
66 (64)[c]
11[e] 4a/Mg/LiCl/ZnCl2 (2)
12[f] 4a/Mg/LiCl/ZnCl2 (2)
[a] The linear/branched ratio was measured by GCMS. [b] The yield was
determined by GCMS with tetradecane as an internal standard. [c] Yield
of the isolated mixture of linear/branched products. [d] Yield of the
isolated branched product 9a. [e] Catalyst loading: 1.25 mol% [Pd2dba3]/
2.5 mol% L1. [f] Catalyst loading: 0.625 mol% [Pd2dba3]/1.25 mol% L1.
dba=dibenzylideneacetone, Tf=trifluoromethanesulfonyl.
(entry 2). Interestingly, similar results were observed with 2-
bromopropane (4a) or 2-chloropropane (4b) when zinc
chloride was omitted (entries 3 and 4). A control experiment
with the commercially available, preformed Grignard reagent
4c also led to a similar outcome (entry 5). Notably, 2-
chloropropane also proved a competent reactant in the
presence of ZnCl2, although the yield of the reaction was
lower (entry 6). Importantly, when the more rigid and bulky
phosphine L2 developed by Buchwald and co-workers was
employed,[17] the selectivity was completely reversed, thus
leading exclusively to the branched product 9a in 82% yield
(Table 1, entry 7). These results highlight the role of the
ligand in the linear/branched selectivity control. Moreover,
they show that both linear and branched products can be
synthesized in a divergent manner from the same reactants by
choosing a different ligand.
The reaction of aryl triflate 8b containing the more
electrophilic ester substituent was next studied to further
probe the utility of the Mg-to-Zn transmetalation (Table 1,
entries 8 and 9). As suspected, the addition of ZnCl2 proved
to be crucial for the desired coupling (entry 9), since only
a small amount of the coupling product was observed in its
absence (entry 8). These results show that ZnCl2 can be
omitted for less sensitive substrates, such as 8a (entries 3–5),
but also that the Mg/LiCl/ZnCl2 system provides better
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
These are not the final page numbers!