Mechanism-based design of a ROMP catalyst for sequence-selective copolymerization

Article abstract of DOI:10.1002/anie.200502606

(Chemical Equation Presented) Back and forth like a metronome is how the carbene moiety swings in a dual-site catalyst which produces a predominantly alternating copolymer of norbornene and cyclooctene from a one-pot reaction of appropriate mixtures of th

Full text of DOI:10.1002/anie.200502606

Angewandte
Chemie
or catalytic functions in polypeptides, information storage and
transmission in polynucleotides, and molecular recognition in
polysaccharides, derive ultimately from sequence-selective
copolymerization of simple monomer units, one can presume
that synthetic polyolefins, as high-performance materials,
would be substantially enhanced if even simple sequence-
selective copolymerizations could be achieved routinely.
Treating regular copolymerization more generally as a prob-
lem in metal-catalyzed organic synthesis, one sees that the
issue is not stereoselectivity, or even regioselectivity. We seek
to introduce into polymerization catalysts the element of
chemoselectivity, which is the most basic kind of selectivity in
organic synthesis. We encode the sequence information in the
catalyst itself, and in this present study succeed in the most
primitive, two-component, alternating sequence of ring-open-
ing metathesis copolymerization.
Phosphine ligand 1 was prepared by sequential alkylation
and arylation of phenyldichlorophosphine with tBuMgCl and
ortho-methoxyphenyllithium, isolation by careful distillation,
cleavage of the methyl ether by BBr3, and then deprotonation
with NaH. Reaction of one equivalent of 1 and [(Cy3P)2RuCl2
=
( CHPh)] (2, Scheme 1) resulted in phosphine exchange,
Copolymerization
Scheme 1. Synthesis of the chemoselective ROMP catalyst 3.
Cy =cyclohexyl.
DOI: 10.1002/anie.200502606
Mechanism-Based Design of a ROMP Catalyst
for Sequence-Selective Copolymerization
followed by elimination of NaCl. A similar approach has been
reported by Hoveyda and co-workers for the preparation of a
tethered, second-generation metathesis catalysts.[4] The
resulting carbene complex 3 was purified by column chroma-
tography under rigorous exclusion of oxygen, and checked by
1H and 31P NMR spectroscopy, as well as ESI-MS, to confirm
that a single compound was prepared. Importantly, no other
carbene species was present (analytical details and further
information on the synthesis of 1 and 3 can be found in the
Supporting Information). Polymerization of norbornene,
cyclooctene, and mixtures thereof, together with 0.05 mol%
catalyst, were conducted under dry N2 at room temperature in
either dichloromethane or cyclooctene as solvent. In experi-
ments with cyclooctene as solvent, the norbornene is con-
sumed quantitatively after 17 h. If a less-coordinating solvent,
such as CH2Cl2, is used, then norbornene is consumed much
more rapidly, with phenomenological rates comparable to
those observed with catalyst 2 under the same conditions. The
resulting polymer was isolated by removal of solvent at
0.01 mbar pressure until no further weight loss was seen. In
copolymerizations with 3, the weight of the dry polymer
corresponded to the weight of norbornene, plus up to one
equivalent of cyclooctene. The polymer samples were char-
Marc Bornand and Peter Chen*
We report here the mechanism-based design of a ring-opening
metathesis polymerization (ROMP)[1] catalyst which prefer-
entially assembles a mixture of cyclooctene and norbornene
into an alternating copolymer. While there have been reports
of regular copolymerizations, highly selective examples are
uncommon outside of free-radical polymerization, with the
1:1 alternating copolymers of ethylene and carbon monox-
ide,[2] or of epoxides and carbon dioxide,[3] being perhaps the
most prominent examples. Given that the exemplary func-
tional characteristics of biopolymers, for example, structural
[*] Dr. M. Bornand, Prof. Dr. P. Chen
Laboratorium für Organische Chemie
Eidgenössische Technische Hochschule (ETH)
Zürich (Switzerland)
Fax : (+41)44-632-1280
E-mail: chen@org.chem.ethz.ch
1
acterized by H and 13C NMR spectroscopy (Varian Gemini
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
300 MHz for 1H, and 75.4 MHz for 13C; CDCl3) and gel
Angew. Chem. Int. Ed. 2005, 44, 7909 –7911
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7909

Communications
permeation chromatography (PL-GPC-220, 1,3,5-trichloro-
benzene, 1808C, PLgel 10 mm column) with refractive index
and viscometry detection calibrated against polyethylene
standards. Reference polymer samples were prepared anal-
=
ogously with [(Cy3P)2RuCl2( CHPh)] (2) for comparison.
The 13C resonances were assigned with the help of group
equivalents, which gave good predictions for the homopol-
ymers. The relaxation time was set to t2 = 5 s so that reliable
integrations could be taken from the 13C NMR spectra.
While the ordinary ruthenium carbene complex 2 poly-
merizes both norbornene and cyclooctene efficiently, com-
plex 3 shows almost no reactivity with cyclooctene alone
under the conditions of the experiment, but is nevertheless as
active (qualitatively) as 2 in ROMP of norbornene. More
interesting is the behavior with respect to copolymerization.
Polymerization catalyzed by 3 in cyclooctene with various
amounts of norbornene proceeds until the norbornene is
exhausted and then stops, but the resultant polymer (Mw
ꢀ 106, Mw/Mn ꢀ 5) contains increasingly long stretches of
alternating copolymer as the ratio of norbornene to cyclo-
octene decreases. GPC analysis indicates that the polymer is
not a physical mixture of the homopolymers of norbornene
and cyclooctene. Both the viscosity and refractive indextraces
show a single, smooth, monomodal peak at the same retention
time, which is significant because norbornene homopolymer
would not appear in the latter trace, its refractive indexbeing
coincidentally the same as that of the 1,3,5-trichlorobenzene
solvent. Moreover, the 13C NMR spectra (Figure 1) of the
homopolymers of norbornene and cyclooctene, as well as the
copolymers with various ratios of the two monomers,
demonstrate that alternate incorporation of cyclooctene and
norbornene is preferred for complex 3 but not for ROMP
catalyst 2.
Figure 1. Olefinic regions of the 13C NMR spectra of the polymer
samples prepared from norbornene/cyclooctene mixtures at different
mole ratios. The top spectra show results for catalyst 2; spectra for
catalyst 3 are at the bottom. The olefinic carbon resonances for the
ring-opening metathesis polymer of norbornene are marked with red
dotted lines, the resonances for the polymer from cyclooctene are
marked in green. The black dotted lines mark the resonances for the
alternating copolymer. There are at least two partially resolved signals
in the high-field resonance, assigned to the end of the double bond
derived from cyclooctene (cis and trans), whereas there are four signals
in the low-field resonance, assigned to the end derived from norbor-
nene. One presumes that, not only the cis and trans configuration of
the double bond, but also the cis and trans configuration of the next
double bond, causes the splitting.
Catalyst 2 in norbornene/cyclooctene mixtures forms
predominantly norbornene homopolymer until the norbor-
nene is exhausted, and then proceeds to homopolymerize
cyclooctene. As the norbornene is increasingly dilute, the 13
C
NMR signals arising from its homopolymer disappear
because of the increasing preponderance of cyclooctene
polymer. Catalyst 3, on the other hand, displays an increasing
preference for the production of the alternating copolymer as
the amount of norbornene, relative to cyclooctene, is
decreased, with about two-thirds of the olefinic linkages in
the copolymer comprising the alternating stretches in the best
cases seen so far. Preliminary results for the negative control
with a symmetrical analogue of 3 having two tert-butyl groups
instead of one phenyl and one tert-butyl show that the
symmetrical catalyst 4 (see Supporting Information for
synthesis and characterization) homopolymerizes norbornene
well and cyclooctene poorly. No alternating copolymer is
formed in the negative control for mixtures of norbornene
and cyclooctene.
The design concept for 3 derives from our earlier
mechanistic work on olefin metathesis. Secondary deuterium
isotope effects, measured in the gas phase, indicated that the
structure of the rate-determining transition state for the
metathesis reaction catalyzed by the first-generation ruthe-
nium catalyst was a metallacyclobutane.[5] Computational
work from our own research group,[6,7] as well as others,[8,9]
however, found minima at the metallacyclobutane structure.
Experimental observation of the olefin p complexby Snapper
and co-workers[10] and a ruthenacyclobutane intermediate by
Romero and Piers[11] supported the DFT calculations. In our
computations, we did find, however, that the rate-determining
transition step for olefin metathesis by 2 was a rotation of the
tricyclohexylphosphine ligand in a metallacyclobutane struc-
ture,[6] which is in fact consistent with the isotope effect
studies, the other computational work, and the solution-phase
observations. For a degenerate metathesis reaction, the
rotation is required by microscopic reversibility. If the
rotation were, however, to be prevented, and, furthermore,
the phosphine were to bear two different substituents, then
the mechanism would require that the active site would swing
back and forth with each productive metathesis step between
two distinct states in a motion reminiscent of a windshield
wiper. The two sites can be made sterically or electronically
different enough so that chemoselectivity in an appropriate
mixture of two substrates can be achieved. Scheme 2 depicts
the catalytic cycle for ROMP of cycloalkenes by 3. The
carbene moiety in the propagating species can lie either on
7910
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 7909 –7911

Angewandte
Chemie
a cyclooctene solution yields a largely 1:1 alternating
copolymer.
Complex 3 is the simplest implementation of the
general concept. It is a dual-site catalyst in which one
site shows chemoselectivity and the other not, which,
together with a concentration difference between the
two monomers, constitutes the bare minimum case for
an alternating copolymerization. Selectivity at both sites
would eliminate the need for widely divergent concen-
trations of the two monomers, but even such a restric-
tion can be easily accommodated by slow addition of
one monomer into a neat solution of the other. One
might expect similar behavior from an unsymmetrically
substituted N-heterocyclic carbene complexreported by
Mol and co-workers,[12] but the complexdid not even
polymerize norbornene well, presumably because of the
much-too-large adamantyl substituent. One should note
that a similar example of a dual-site catalyst can be
found in the C1- or Cs-symmetric metallocene catalysts
for syndiotactic polypropylene.[13] In these systems, the
activated catalyst alternately selects for the re and si face
of the single monomer, propylene, but there is no
documented case where selective copolymerization has
been reported with these catalysts. The alternating
selectivity has other potential uses as well. Selective
oligomerization and selective ring-closing metathesis
should be achievable with the new, designer catalyst
architecture. Furthermore, while the designed structure
demonstrates the target selectivity, no optimization of
the catalyst has been attempted yet. Further work in this
direction is underway.
Received: July 25, 2005
Keywords: alkenes · copolymerization ·
homogeneous catalysis · metathesis · ruthenium
.
Scheme 2. Catalytic cycle for ROMP of cycloalkenes with catalyst 3.
the right, as in species A, or on the left, as in species D.
Molecular modeling studies indicate that A is much less
sterically crowded than D, which breaks what would have
been a degeneracy in the more usual case of a rotatable
phosphine or a phosphine with at least two identical
substituents.
[1] K. J. Ivin, J. C. Mol, Olefin Metathesis and Metathesis Polymer-
ization, Academic Press, New York, 1997.
[2] E. Drent, J. A. M. van Broekhoven, M. J. Doyle, J. Organomet.
Chem. 1991, 417, 235.
[3] D. J. Darensbourg, M. W. Holtcamp, Coord. Chem. Rev. 1996,
153, 155.
[4] J. J. Van Veldhuizen, D. G. Gillingham, S. B. Garber, O.
Kataoka, A. H. Hoveyda, J. Am. Chem. Soc. 2003, 125, 12502.
[5] C. Adlhart, C. Hinderling, H. Baumann, P. Chen, J. Am. Chem.
Soc. 2000, 122, 8204.
[6] C. Adlhart, P. Chen, Angew. Chem. 2002, 114, 4668; Angew.
Chem. Int. Ed. 2002, 41, 4484.
[7] C. Adlhart, P. Chen, J. Am. Chem. Soc. 2004, 126, 3496.
[8] L. Cavallo, J. Am. Chem. Soc. 2002, 124, 8965.
[9] S. F. Vyboishchikov, M. Bꢀhl, W. Thiel, Chem. Eur. J. 2002, 8,
3962.
[10] J. A. Tallarico, P. J. Bonitatebus, M. L. Snapper, J. Am. Chem.
Soc. 1997, 119, 7157.
[11] P. E. Romero, W. E. Piers, J. Am. Chem. Soc. 2005, 127, 5032.
[12] M. B. Dinger, P. Nieczypor, J. C. Mol, Organometallics 2003, 22,
5291.
If we consider the productive direction around the
catalytic cycle for ROMP (A!B!C!D!E!F!A) with
each step reversible, it would be expected that the forward
reaction A!B!C would only occur if there is a large strain
release in the B!C step so that the intermediate metal-
lacyclobutane B would preferentially partition forward to C
rather than return to A. On the other hand, the reaction D!
E!F should proceed in the forward direction for any
cycloalkene, strained or not, because the metallacyclobutane
E should partition forward to the sterically less hindered
carbene F. Consequently, the intermediate A can only
incorporate norbornene into the growing chain, but inter-
mediate D can take either norbornene or cyclooctene, the
probability being largely determined by the relative concen-
trations. Accordingly, dilute concentrations of norbornene in
[13] Y. van der Leek, K. Angermund, M. Reffke, R. Kleinschmidt, R.
Goretzki, G. Fink, Chem. Eur. J. 1997, 3, 585.
Angew. Chem. Int. Ed. 2005, 44, 7909 –7911
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7911