.
Angewandte
Communications
DOI: 10.1002/anie.201108270
Synthetic Methods
Rhodium Dinaphthocyclooctatetraene Complexes: Synthesis,
Characterization and Catalytic Activity in [5+2] Cycloadditions**
Paul A. Wender,* Adam B. Lesser, and Lauren E. Sirois
New reactions provide new ways to think about bond
construction and thus more, and often greener, options for
achieving greater step,[1] atom[2] and time economical,[3] if not
ideal, syntheses.[4] Guided by these considerations, we pre-
viously introduced a reaction for seven-membered ring
synthesis involving metal-catalyzed [5+2] cycloadditions of
vinylcyclopropanes (VCPs) and p systems.[5] While rhodium
complexes have shown the greatest generality in catalyzing
this process, working both intra- and intermolecularly and
with absolute stereocontrol[6] and even in water,[7] ruthe-
nium,[8] nickel,[9] and iron[10] catalysts have also been effective
in many cases. We report herein the first studies of a new
family of catalysts for [5+2] cycloadditions based on relatively
little studied rhodium cyclooctatetraene (COT) complexes.
We describe the synthesis and metal complexation of our
dinaphtho[a,e]cyclooctatetraene (dnCOT) ligand 5 derived
from a recently introduced [2+2+2+2] cycloaddition of
diynes.[11] The resulting Rh–dnCOT catalyst provides [5+2]
cycloadducts in high yields, often in minutes at room temper-
ature, is compatible with a variety of functionalities, and
exhibits enhanced or even reversed regiocontrol in selected
cases, relative to known catalysts.
The [2+2+2+2] cycloaddition of diynes has proven to be
an excellent reaction for the synthesis of highly substituted
COTs.[11] In addition to the value of such COTs as synthetic
building blocks and components of novel materials and
devices,[12–16] a further motivation for our interest in this
process was the potential use of COTs as ligands for catalysis.
Due to their tub-shaped conformation, certain COTs can
coordinate transition metals in a 1,2,5,6-h4 manner analogous
to dienes such as cyclooctadienes (CODs).[17] Interestingly,
the distance and bite angle between the binding alkene
moieties in metal complexes of both COD and COT are the
same across a variety of crystal structures, with a distance of
2.8 ꢀ and a bite angle of 868.[18] While it is tempting to
associate CODꢁs generally superior binding ability to tran-
sition metals with pre-organization of the alkenes, the
distance between alkenes in unbound COD (3.20 ꢀ) is
greater than that for unbound COT (3.09 ꢀ).[18] Furthermore,
CODꢁs inherently greater flexibility (and thus entropic bind-
ing penalty) should bias metals in favor of COT entropically.
Despite this, COD has been more commonly used than COT
as a ligand in metal complexes.[17] The similarities between
CODs and COTs and the ease of synthesis of substituted
COTs using the recently introduced [2+2+2+2] cycloaddition
methodology prompted our interest in determining whether
judiciously modified COTs could be effective ligands for
metal catalysis.
While many COTs are relatively labile metal ligands
(1,2,5,6-h4 coordination),[17,19] structural modifications to the
COT scaffold, such as benzannulation, often enhance their
ability to bind to transition metals.[20,21] Members of the
dibenzo[a,e]cyclooctatetraene (dbCOT) subfamily have been
complexed with a variety of transition metals (e.g., Pd, Pt, Rh,
Ir, Co, Mo, Cr, Cu).[22] However, the catalytic activity of these
complexes remains relatively underexplored. In fact, dbCOT
has found use as an effective poison in tests for homogeneous
catalysis.[23] Reports of catalytic activity of dbCOT–metal
complexes are relatively rare,[24,25] and, significantly, no
dbCOT complexes have been evaluated as catalysts for
cycloadditions.
Based on anticipated beneficial properties of its com-
plexes (e.g., crystallinity) and ease of synthesis, dnCOT 5 was
targeted for this study. Treatment of commercially available
1,2-bis(bromomethyl)benzene (1) with copper trimethylsilyl
acetylide followed by deprotection affords diyne
3
(Scheme 1) which upon Ni-catalyzed [2+2+2+2] cycloaddi-
tion provides cycloadduct 4 in excellent yield (up to 87% over
3 steps). Oxidation with 2,3-dichloro-5,6-dicyanobenzoqui-
none (DDQ) at room temperature furnishes dnCOT 5 as
a crystalline solid (see X-ray, Figure 1a). This step economical
(4 steps) sequence is highly efficient (up to 70% overall yield)
and has been carried out successfully on a multi-gram scale,
offering advantages in yield, flexibility, and/or brevity relative
to a previous synthesis of dnCOT (6 steps, 4.1% overall
yield)[26] and notable syntheses of the related dbCOT (4 steps,
47% overall yield; 3 steps, 38% overall yield).[27]
Complexation of dnCOT 5 with rhodium was accom-
plished by treatment of the former with [{RhCl(CO)2}2] which
proceeds with evolution of CO and the formation of a poorly
soluble intermediate, putatively the [{RhCl(dnCOT)}2] dimer
6 (Scheme 1). Treatment of this intermediate with silver
hexafluoroantimonate in DCM/MeCN gives rise to RhI–
[*] Prof. P. A. Wender, A. B. Lesser, L. E. Sirois
Department of Chemistry, Department of Chemical and Systems
Biology, Stanford University
Stanford, CA 94305-5080 (USA)
E-mail: wenderp@stanford.edu
[**] We thank the National Science Foundation (CHE-0450638) for an
NSF Graduate Research Fellowship (L.E.S.) and Amgen and Eli Lilly
(A.B.L.) for financial support. We also thank Allen Oliver at the
University of Notre Dame for X-ray crystal structures. Justin P.
Christy is acknowledged for his supporting work on the syntheses of
COTs and for helpful discussions.
Supporting information for this article (full experimental details,
characterization of new compounds, preparation and character-
ization of dnCOTand its rhodium and iridium metal complexes, and
procedures for [5+2] cycloadditions) is available on the WWW
2736
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 2736 –2740
Wender, Paul A.
