Journal of the American Chemical Society
Communication
conditions.13 However, to date no examples of asymmetric
dimerization with substrates bearing phosphine oxides or
phosphonates have been reported.
We next explored the scope of the dimerization (Table 2).
Product 2 was isolated in 87% yield and 96% ee on 1 mmol scale.
Similar results were observed on a multi-gram scale. Variation of
the substitution around the aryl ring was tolerated. Notably, high
ee’s were observed with various oxygen substituents ortho to the
iodide, including both simple ethers and other oxygen
heterocycles. Other substitution was also tolerated, including
aromatic, bulky, electron-rich, and electron-poor groups.
Phosphonate substitution could also be varied, and larger
aromatic cores also underwent asymmetric dimerization. In
nearly all cases, both yields and ee’s were high. One exception
was for binaphthyl 17, which required modified conditions
[ArBr, Ni(COD)2 with increased loading, and Mn], and even
then modest ee and low yield were observed. In this case, as well
as others with limited yield, proto-dehalogenation of the starting
material was the major byproduct.15 In contrast, we also
examined ortho-(iodo)phosphine oxide substrates (Table 2,
bottom) and found that they also underwent effective
asymmetric dimerization with good generality.
Given the need for methods to prepare axially chiral
bisphosphine ligands, we undertook study of this reaction and
now report its successful development. We describe the nickel-
catalyzed asymmetric Ullmann-type homocoupling of ortho-
(iodo)phosphine oxides and ortho-(iodo)phosphonates to
prepare chiral biaryl bisphosphine oxides and bisphosphonates.
Good yields and high ee’s are observed across a range of
substrates. Both classes of products can be converted to highly
enantioenriched chiral biaryl bisphosphines. In the case of the
bisphosphonates, families of different chiral biaryl bisphosphines
can be obtained from a common precursor. We also show that
this process can be extended to other classes of asymmetric
biaryl homocouplings.14
We began by examining the dimerization of ortho-(bromo)-
arylphosphonate 1a (Table 1). Using catalytic Ni(COD)2 and
stoichiometric Mn as the terminal reductant, a variety of
potential ligands were examined. With most, little to none of the
desired biaryl product was observed (e.g., entries 1−3). An
exception was with the use of 2-(pyridyl)oxazolines, as
exemplified by ligand 7 (entry 4). Using this ligand, a good
yield of the desired product was observed; however, the product
proved to be essentially racemic. Fortunately, examination of
additives revealed that the addition of co-catalytic amounts of
cobalt phthalocyanine (CoPc)11a,h provided a dramatically
altered result, and the product was observed in similar yield
with 82% ee (entry 5). Based upon the literature precedent, we
assume that the CoPc assists in electron transfer from the
stoichiometric reductant (see below).11a,h
Both classes of products were converted to enantioenriched
biaryl bisphosphines. First, bisphosphine oxide 18 [98% ee from
the (S)-enantiomer of 7] was reduced to SEGPHOS in 72%
yield with complete retention of absolute configuration.4e
Analysis by HPLC and optical rotation established the major
product as the (S)-antipode (eq 1), establishing the absolute
selectivity of the dimerization reaction.18
Other 2-(pyridyl)oxazoline ligands were examined, but none
proved more effective than 7.15 Other classes of stoichiometric
reductants were also examined. Interestingly, significant differ-
ences in yield and ee of the desired product 2, as well as in the
production of byproduct 3, were observed. Other metallic
reductants (Zn and Mg) negatively impacted the reaction
(entries 6 and 7). Initially, these results led us to suspect that the
divalent cations that result from the metallic reductants were
competing with the Ni catalyst for the ligand. We then examined
the use of tetrakis(dimethylamino)ethylene (TDAE) as the
reductant,16 as the resulting organic cation would not compete
for ligand. With this reductant, significant improvement in the ee
of the desired product was observed, along with modest
improvements in chemical efficiency of dimerization (entry 8).
Notably, subsequent control experiments conducted with
TDAE as the reductant revealed that added MnI2 reduces the
yield of the dimerization. However, the ee of the product formed
under these conditions is essentially unchanged.15 This suggests
that the ligand binding by Mn2+ is important with regard to the
amount of active Ni catalyst in solution, but that Mn2+ does not
impact the enantiodetermining step. No reaction is observed in
the absence of nickel. Thus, the importance and role of TDAE
are more complicated than initially suspected (see below).
Further improvements to the reaction conditions were
observed by increasing the reaction temperature and decreasing
the catalyst loading (entry 9).17 Switching from aryl bromide to
aryl iodide with 1 mol % CoPc further improved the reaction
(entry 10). Finally, deuterium-labeling studies showed that the
reduction product 3 results from proton quenching,15 and
addition of 4 Å molecular sieves (4 Å MS) decreased the amount
of 3. Inexpensive NiBr2·DME could be used instead of
Ni(COD)2 (entry 11).
A distinct advantage of the bisphosphonates is that they allow
preparation of a family of chiral bisphosphine ligands from a
single enantioenriched staring material. As a demonstration, a
series of biaryl bisphosphines were prepared from 2 (≥96% ee,
Table 3). 2 was converted to the bis(dichlorophosphate)19,20
Table 3. Bisphosphonates to Biaryl Bisphosphines
a
b
TMSCl, NaI, CH3CN, 50 °C. (COCl)2, DMF, DCM, 40 °C.
c
d
PMHS, Ti(OiPr)4, THF, 70 °C. HSiCl3, PhNMe2, PhMe, 110 °C.
e
Ee determined after re-oxidation of product.
1330
J. Am. Chem. Soc. 2021, 143, 1328−1333