-
Design of stereoselective Ziegler–Natta propenepolymerization
catalystsVincenzo Busico*, Roberta Cipullo, Roberta Pellecchia,
Sara Ronca, Giuseppina Roviello, and Giovanni Talarico
Dipartimento di Chimica, Università di Napoli Federico II, Via
Cintia, 80126 Naples, Italy
Edited by Tobin J. Marks, Northwestern University, Evanston, IL,
and approved July 25, 2006 (received for review April 14, 2006)
After five decades of largely serendipitous (albeit
formidable)progress, catalyst design in Ziegler–Natta olefin
polymerization,i.e., the rational implementation of new active
species to targetpredetermined polyolefin architectures, has
ultimately become arealistic ambition, thanks to a much deeper
fundamental under-standing and major advances in the tools of
computational chem-istry. In this article, we discuss, as a case
history, a unique class ofstereorigid C2-symmetric
bis(phenoxy-amine)Zr(IV) catalysts withcontrolled kinetic behavior.
A large variety of polypropylene mi-crostructures have been
obtained with these catalysts by modu-lating the steric demand of
one key substituent, without alteringthe nature and symmetry of the
ancillary ligand framework, underthe guidance of computer modeling.
This unusual achievement isrelevant per se and for the perspective
implications in catalystdiscovery.
enantioselectivity � olefin polymerization � polypropylene
Z iegler–Natta olefin polymerization is probably the
mosteffective and atom-economical large-volume industrialchemical
process (1–3). The catalytic cycle (Fig. 1 for ethene)
isdisarmingly simple and entails merely the insertion of a mono-mer
molecule into a M-C � bond (where M is a coordinativelyunsaturated
transition metal center that can be isolated or resideat a suitable
surface). What makes Ziegler–Natta catalysts‘‘unique and
marvelous’’ (4) is that several thousand cycles mayclose in �1 s
under mild conditions before chain transfer(usually, a �-H
elimination event; Fig. 1), an individual M canproduce several
thousand chains before it ultimately deactivates,and, last but not
least, the process is amenable to thoroughchemocontrol,
regiocontrol, and stereocontrol (1–3).
The enantioselectivity in the insertion of propene, in
partic-ular, is truly amazing. Although neither isotactic nor
syndiotacticpolypropylene are chiral molecules (Fig. 2) (5), each
individualmonomer insertion implies a chiral recognition, the
catalyticspecies selecting one of the two enantiofaces of propene
in thefavored regiochemistry with an enantiomeric excess that,
inseveral reported cases (6, 7), is �99.8% (an unrivaled
perfor-mance for a substrate with no functional groups other than
anolefinic double bond).
More than in these impressive figures, though, the real
fascina-tion and challenge of modern Ziegler–Natta chemistry are in
thepossibility of implementing catalysts for specific homopolymer
orcopolymer architectures. In particular, the ability to enchain
pro-pene, alone or in combination with other monomers (e.g.,
ethene),with more or less precisely (pre)determined
chemoselectivity,regioselectivity, and enantioselectivity, and the
dramatic effect ofchain microstructure on the physical properties
of the materials(ranging from ultra-rigid thermosets to elastomers
via all conceiv-able grades of thermoplastic elastomers), is a key
to understandingwhy polypropylene production continues to grow
exponentially,faster than for any other large volume resins (e.g.,
other polyolefins,polyesters, polyamides, polycarbonates,
polyvinylchloride, etc.), andto some extent at their expense
(2).
Like most scientific breakthroughs, the discovery of
Ziegler–Natta catalysts has been largely serendipitous since the
verybeginning. Without the accidental contamination of an auto-
clave with colloidal nickel in Karl Ziegler’s laboratory in 1953
(4,8), the fortuitous chlorination of MgO to MgCl2 by TiCl4
(late1960s) (1), or the partial hydrolysis of AlMe3 in a broken
NMRtube to yield methylalumoxane (mid-1970s) (9), each
triggeringformidable chains of deductions and consequent actions,
catalystevolution and diversification would have occurred at a
muchslower pace. The sound argument that the chance of a lucky
findis roughly proportional to the number of attempts has now
beenset at the foundation of catalyst discovery program by
high-throughput screening technologies (10), which have already
ledto industrially relevant achievements (11, 12).
In this scenario, catalyst design (meaning the rational
imple-mentation of active organometallic species for tailored
applica-tions) has been, euphemistically, marginal until now. On
theother hand, the hundreds of metallocene (7, 13) and
nonmetal-locene (14) catalyst structures that have been disclosed
andtested in the last two decades, and the thorough
theoreticalstudies on realistic models thereof (6, 7), highlighted
a numberof basic principles relating catalyst structure and
performance.These can now be used successfully, if not (yet) for
attempts ofde novo design, certainly for catalyst fine-tuning. In
this article,we illustrate, with a case history based on the
current researchof our own laboratory, how wide the margins of
useful improve-ment can be. Before that, though, it may be useful
to recall a fewimportant notions of Ziegler–Natta
stereochemistry.
Stereoselectivity of Propene Polymerization
Catalysts:Preliminary RemarksMost Ziegler–Natta propene
polymerization catalysts are highlyregioselective in favor of 1,2
(primary) insertion [M-R �
Author contributions: V.B. designed research; R.C., R.P., S.R.,
G.R., and G.T. performedresearch; S.R. contributed new
reagents�analytic tools; V.B., R.C., and G.T. analyzed data;and
V.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS direct submission.
Abbreviation: QM, quantum mechanics.
Data deposition: The atomic coordinates have been deposited in
the Cambridge StructuralDatabase (CSD), Cambridge Crystallographic
Data Centre, Cambridge CB2 1EZ, UnitedKingdom (CSD reference nos.
603920 and 603921).
*To whom correspondence should be addressed. E-mail:
[email protected].
© 2006 by The National Academy of Sciences of the USA
Fig. 1. The catalytic cycle of Ziegler–Natta olefin
polymerization.
www.pnas.org�cgi�doi�10.1073�pnas.0602856103 PNAS � October 17,
2006 � vol. 103 � no. 42 � 15321–15326
CHEM
ISTR
YSP
ECIA
LFE
ATU
RE
Dow
nloa
ded
by g
uest
on
July
3, 2
021
-
CH2�CH(Me)�M-CH2-CH(Me)(R)], because of electronicand�or steric
effects (6). For the insertion to also be enantio-selective, at
least one element of chirality needs to combine withre or si
propene coordination (Fig. 3) and give rise to appreciablefree
energy differences between the competing diastereomerictransition
states. This element can be the configuration of thegrowing
polypropylene chain (and in particular of the asymmet-ric C in the
last-inserted monomeric unit) and�or the possiblechirotopicity of
the active site(s) (6, 7).
The two cases are commonly referred to as ‘‘chain-end con-trol’’
and ‘‘site control,’’ respectively (6, 7). In the latter,
theenantiodiscrimination is traceable to direct nonbonded
contactsbetween the incoming monomer and the ancillary ligand(s),
or,much more frequently, it is mediated by the growing
chain,sterically constrained in a chiral conformation favoring
propeneinsertion with the enantioface that directs the methyl
substituentanti to the first chain C-C bond (Fig. 4; P � polymer
chain) (6,7, 15).
Although, in principle, all said elements can have
nonnegli-gible effects, in general one is largely predominant, and
theconfiguration of all known stereoregular polypropylenes
ob-tained so far with single-center catalytic species, determined
bymeans of high-resolution 13C NMR in terms of
normalizedstereosequence distributions up to the heptad�nonad level
(6),has been well reproduced in terms of chain propagation
modelsbased on the limiting approximation of pure chain-end or
sitecontrol, the latter in a more or less elaborated version
dependingon the chirotopicity of the active sites (6). Without
going intodetails, let us define the following approximated
stochasticmatrix of chain propagation states:
The matrix formulation assumes complete regioselectivityand (up
to) first-order Markov configurational statistics. Therows are
addressed to the configuration of the last-insertedmonomeric unit,
and the columns are addressed to that of themonomeric unit to be
generated; therefore, � and � are theprobabilities that a monomeric
unit of R or S configuration,respectively, is followed by a new one
of R configuration. Pure
site control requires � � �, whereas � � (1 � �) correspondsto
pure chain-end control (6). Matrix multiplication codes forthe
statistical analysis of the 13C NMR stereosequence distri-bution,
ending up with the best-fit values of � and �, have beendescribed
(6).
Understandably, symmetry is a key issue for catalysts
withchirotopic sites. The evolution of metallocenes teaches
thatC2-symmetric active species with homotopic sites tend to
beisotactic-selective, whereas Cs-symmetric ones with
enantiotopicsites are often syndiotactic-selective (6, 7, 13). On
the otherhand, recent results in the fast-growing area of
nonmetallocenecatalysts (14), many of which are easier to make and
amenableto structural amplification with a parallel synthetic
approach (10,11), demonstrate that the relationship between
symmetry andstereoselectivity can be far less obvious (16).
Results and DiscussionIn the last few years, we have undertaken
a systematic study(17–19) of Zr(IV) catalysts bearing a stereorigid
tetradentatebis(phenoxy-amine) ancillary ligand (Fig. 5),
originally intro-duced by Kol and coworkers (20). We had been
attracted by thisligand framework because it closely mimics the
coordinationenvironment of the surface Ti atoms in the classical
heteroge-neous Ziegler–Natta systems for the industrial production
ofisotactic polypropylene (1, 4, 6). In the following, we shall
seehow a simple modulation in the steric demand of one
keysubstituent (R1 in Fig. 5) without altering the C2 symmetry of
thecomplex makes it possible to play with the mechanisms of
stereoregulation almost at will and to achieve polypropylenes
withdramatically different microstructures. We shall also discuss
theconcurrent effect of this fine-tuning on the propensity to
chaintransfer, which opened the door to the first highly
isotacticpropene block copolymerizations.
The activation of the neutral (ONNO)ZrBn2 precursors of Fig.5 to
[(ONNO)ZrR]� active cations can be carried out withmethylalumoxane,
or proper boranes or borate salts similarly towhat is done with
metallocenes (21). Solution NMR studies onthe resulting ion couples
indicate that the tetradentate coordi-
Fig. 2. Sawhorse representations of an isotactic (Upper) and
syndiotactic(Lower) poly(1-alkene) chain in the (hypothetical)
all-trans conformation.Consecutive monomeric units with equal or
opposite relative configurationsat the stereogenic tertiary C atoms
are denoted as meso (m) and racemo (r)diads, respectively,
according to a nomenclature originally introduced byBovey (5).
Fig. 3. The chirality of coordination of propene to a metal
center.
Fig. 4. Schematic representation of the ‘‘growing chain
orientation mech-anism of stereocontrol,’’ originally proposed by
Corradini et al. (15).
R S
R α 1-α
S β 1-β
15322 � www.pnas.org�cgi�doi�10.1073�pnas.0602856103 Busico et
al.
Dow
nloa
ded
by g
uest
on
July
3, 2
021
-
nation and the C2-symmetric trans(O,O), cis(N,N) configurationis
retained.
An extensive quantum mechanics (QM) computationalscreening of
the active species was aimed at predicting theirregioselectivity,
enantioselectivity, and molecular mass capabil-ity in propene
polymerization (for technical details see Materialsand Methods and
Supporting Text, which is published as support-ing information on
the PNAS web site). The results led us toconclude that in general
propene insertion with 1,2 regiochem-istry is dominant, for reasons
that have been described (22). Theyalso revealed that substituent
R2, far from the active pocket, hasvirtually no effect on the
catalytic behavior, whereas the stericbulk of R1, more to the
catalyst front, is the key to control at thesame time the
enantioselectivity of 1,2 insertion and the pro-pensity of the
growing chain undergo �-H elimination to themonomer, which is by
far the main chain transfer pathway in theabsence of Al-trialkyls
(17–19, 23). It is important to note thatthe above is at odds with
what is known for C2-symmetricansa-metallocene cations, whose
enantioselectivity and chaintransfer properties are independently
affected by different sub-stituents (7). According to our
computations (Table 1), site-controlled enantiodiscrimination in
1,2 propene insertion ap-pears for R1 � tert-butyl, but for a high
catalyst isotacticselectivity and molecular mass capability an even
higher stericdemand is needed to enforce the growing chain
orientationmechanism (Figs. 4 and 6 Left) and destabilize the
six-centertransition state of �-H elimination to propene (Fig. 6
Right) (24,25). On the other hand, the calculations indicate that
when R1is too bulky (e.g., R1 � trityl) direct interference occurs
betweenthe same R1 and the methyl group of a monomer
moleculeinserting with the correct regiochemistry and the
enantiofacefavored by the chain, which is detrimental to the
enantioselec-tivity, and also, less intuitively, to the
regioselectivity.
When we started our work, two (ONNO)ZrBn2 complexes,with R1 (�
R2) � methyl (1) and tert-butyl (3), had been reportedand upon
activation with B(C6F5)3 found to mediate the poly-merization of
1-hexene, the latter in isotactic and living fashion(20). These
systems appeared ideal entries for a validation of our
modeling; therefore, we prepared and tested them in
propenepolymerization. In agreement with the QM prediction, the
chiralpocket of 1 turned out to be too open to recognize the
twoenantiofaces of propene, and a syndiotactically
enrichedpolypropylene was obtained because of chain-end control.
Forpolymerizations carried out at 25°C in toluene, 13C
NMRstereosequence analysis on the polymers ended up indeed with� �
(1 � �) � 0.387, independently of propene concentration(Table 2,
entry 1). With R1 � tert-butyl, on the other hand, thechiral pocket
is tighter, and we were pleased to find out that QMmodeling had
been right to predict that site control would bedominant. In fact,
depending on its � or � configuration, at thetwo homotopic active
sites the active species of 3 has a fairlystrong preference for one
or the other monomer enantiofaces,which wipes out the influence of
chain end. Quantitatively, bystatistical analysis of the 13C NMR
stereosequence distributionfor polypropylene samples prepared with
3 at 25°C in toluene wemeasured � � � � 0.955; the polymer is
therefore isotactic,albeit with 4.5 mol% of randomly distributed
steroirregular units(. . . mmmrrmmm . . .), which results into a
fairly modest meltingtemperature (Table 2, entry 3).
The number average polymerization degree (Pn) of
thepolypropylene made with 1 and 3 is rather low (Table 2)
andindependent of propene concentration, as is the case when
�-Helimination to propene is the dominant chain transfer pathway(Pn
� Rp�Rt � kins[C3H6]�(kH,M[C3H6]), where Rp and Rt are theoverall
rates of chain propagation and transfer, and kins and kH,Mare the
kinetic constants of 1,2 monomer insertion and of �-Helimination to
monomer, respectively. These results are nicelyconsistent with the
QM indications.
Table 1. Main results of the QM modeling of
propenepolymerization at active cations derived from the
(ONNO)ZrBn2precursors of Fig. 5, with an iso-butyl group simulating
thegrowing chain
R1 R2 �Eregio# * �Eenantio
# † �Et-p# ‡
H H 3.5 �0 0Methyl Methyl 3.6 �0 0.2tert-butyl tert-butyl 3.6
1.7 1.81-Adamantyl Methyl 4.1 4.1 3.5�,�-dimethylbenzyl (cumyl)
Methyl 3.5 3.7 2.2Triphenylmethyl (trityl) Methyl 1.6 �0 3.8
Difference in internal energy (in kcal�mol�1) between the lowest
transitionstates for: *2,1 and 1,2 propene insertion; †1,2 propene
insertion with the twopossible enantiofaces; ‡chain transfer via
�-H elimination to propene andchain propagation via 1,2 propene
insertion. The latter is scaled to the case ofR1 � R2 � H set to
�Et-p
# � 0.
Fig. 5. The general structure of the (ONNO)ZrBn2 precatalysts
studied in thepresent work (Bn � benzyl).
Fig. 6. QM models of the transition states of 1,2 propene
insertion (Left) and �-H elimination to propene (Right) at an
active [(ONNO)ZrP]� cation, in whichthe growing polymer chain (P)
is simulated with an iso-butyl group (in bright green).
Substituents R1 of Fig. 5 are generically represented as yellow
spheres. Themain nonbonded interactions are indicated by dotted
lines.
Busico et al. PNAS � October 17, 2006 � vol. 103 � no. 42 �
15323
CHEM
ISTR
YSP
ECIA
LFE
ATU
RE
Dow
nloa
ded
by g
uest
on
July
3, 2
021
-
Moving from these facts, we undertook a program of
catalystfine-tuning aimed primarily at the achievement of highly
isotac-tic, high-molecular-mass polypropylene. We also wanted
tosynthesize one or more catalysts for which chain-end and
sitecontrol are both nonnegligible and verify the
QM-predictedexistence of a threshold in the steric bulk of R1 above
which thesubstituent effect on the regioselectivity and
enantioselectivitywould be detrimental.
Not surprisingly, we found out that the range in whichchain-end
and site control are concurrent corresponds to R1 ofa size
intermediate between methyl and tert-butyl, which is thecase, e.g.,
of R1 � cyclohexyl. The x-ray molecular structure ofthe dibenzyl
precatalyst 2 (R2 � methyl) is shown in Fig. 7 Left.The
coordination around Zr is distorted-octahedral with tran-s(O,O),
cis(N,N) configuration [C-Zr-C � 111.6(1), O-Zr-O �166.3(1),
N2-Zr-O2 � 74.8(1)], similarly to what was observedbefore for 3
(20). Apart from the distortions traceable to theconstraints
imposed by the chelation of the tetradentate(ONNO) ligand, the
local symmetry at the metal deviates fromC2 for the arrangement of
the two benzyl moieties, one of which
is bound in �1 fashion, whereas the other approaches �2
coor-dination [Zr-C1b-C13b � 97.1(2)°; Zr-C13b � 0.2880(4)
nm;C1a-Zr-C1b-C13b � 2.9(3)°]. The high-resolution 1H NMRspectrum
(Supporting Text) clearly indicates that the trans(O,O),cis(N,N)
configuration is maintained in solution, where themolecular motions
result into an average C2 symmetry.
The chiral active pocket of 2 is tight enough to moderatelyfavor
propene insertion with one enantioface, but this
preferenceconflicts with the preference of the growing chain for
analternation. As a result, the two controls can be either ‘‘in
phase’’or ‘‘out of phase,’’ depending on the configuration of
thelast-inserted monomeric unit, which is shown unambiguously
bypolymer stereosequence analysis ending up with � different from�
and � different from (1-�) (entry 2 of Table 2 and Table 3,which is
published as supporting information on the PNAS website); to the
best of our knowledge, this case was still unprece-dented in
Ziegler–Natta propene polymerization (6). The cor-responding
polypropylene configuration is really intriguing, withthe
copresence of relatively long isotactic and syndiotacticstrands
([mmmm] � 4.6%, [rrrrrr] � 7.0%).
Table 2. Experimental results of propene polymerization with
catalysts derived from complexes 1-7 (at 25°C in toluene;
forexperimental details, see Materials and Methods and Supporting
Text)
Precatalyst�entry R1 R2 mm, % rr, % � � 2,1 units, % Tm,°C Pn �
10�2 Rp
1 Methyl Methyl 12 30 0.387 1-� 0.2 Amorphous 1.0 6.42
Cyclohexyl Methyl 18 41 0.517 0.844 0.3 Amorphous 1.5 3.43
tert-butyl tert-butyl 87 4.3 0.955 � 0.3 120 1.4 2.54 1-Adamantyl
Methyl 98.8 0.4 0.996 � 0.4 152 18 4.85 Cumyl Methyl 97.6 0.8 0.980
� 0.8 140 50 736 Diphenylmethyl tert-butyl 86 4.0 0.93 � 1.0 114
9.1 7.47 Trityl tert-butyl 24 37 �0.5* �0.5* 10 Amorphous 20
1.8
mm and rr � 13C NMR fractions of mm and rr triads in the
polymer. � and � � stochastic probabilities of chain propagation,
as obtained by statistical analysisof 13C NMR stereosequence
distribution (for the model, see text). 2,1 units � 13C NMR
fraction of regioirregular (2,1) monomeric units in the polymer. Tm
� DSCmelting temperature of the polymer. Pn � 1H NMR number average
degree of polymerization. Rp � rate of polymerization, in
kg(polymer)�mol(Zr)�1 [C3H6]�1�h�1.*Difficult to quantify more
precisely by 13C NMR, caused by the extensive overlapping between
resonances arising from regio and stereo defects.
Fig. 7. X-ray molecular structures of the dibenzyl precatalysts
of Fig. 5 with R1 � cyclohexyl, R2 � methyl (2; Left) and R1 �
trityl, R2 � tert-butyl (7; Right). Thermalellipsoids are shown at
25% probability level. For clarity, all hydrogen atoms were
omitted. More details are provided in Supporting Text.
15324 � www.pnas.org�cgi�doi�10.1073�pnas.0602856103 Busico et
al.
Dow
nloa
ded
by g
uest
on
July
3, 2
021
-
At entries 4–7 of Table 2, in turn, we illustrate the effect
onthe extent of site control of increasing the steric demand of
R1beyond that of a tert-butyl. The catalyst obtained from
thedibenzyl precursor 4, with R1 � 1-adamantyl (and R2 �
methyl),has an exceedingly high enantioselectivity (at 25°C, we
measured� � � � 0.996, corresponding to an enantiomeric excess
�99.2%) and yields a highly isotactic polypropylene with a
meltingtemperature and enthalpy of 152°C and 130 J�g�1,
respectively;these values are among the highest ever reported for
propenepolymerization in homogeneous phase (6, 7, 13, 14). On
theother hand, in accordance with the computational analysis,
assoon as R1 becomes bigger the regioselectivity and
enantiose-lectivity start fading out. An extreme case is that of
complex 7,with R1 � trityl (and R2 � tert-butyl). The steric
crowding at theZr center is evident from the x-ray molecular
structure (Fig. 7Right). Although the configuration is the same as
2 and 3, thehuge trityl groups intrude into the volume occupied by
the twobenzyl moieties, which are both forced to bind in a ‘‘slim’’
�1fashion. The NMR characterization in solution (see Fig. 9,
whichis published as supporting information on the PNAS web
site),while confirming the trans(O,O), cis(N,N) configuration,
pointedout a diffuse substituent interlocking, as shown in
particular bythe broad resonances of most trityl protons in the
solution 1HNMR spectrum at room temperature. At the 13C NMR
charac-terization, the polypropylene prepared with this catalyst
turnedout to be practically atactic, and also poorly regioregular,
with�10 mol% of randomly distributed
head-to-head�tail-to-tailenchainments (Table 2, entry 7, and Fig.
10, which is publishedas supporting information on the PNAS web
site).
Fig. 8 shows the calculated transition states of the four
lowestenergy propene insertion paths identified for a model
activecation of 7 (R2 � Me), with the growing chain simulated with
aniso-butyl group (for details, see Computational Section in
Sup-porting Text). In Fig. 8 A and B, the first C-C bond of the
iso-butylpoints away from the nearest-in-space trityl substituent;
a pro-pene molecule that inserts with 1,2 regiochemistry cannot
avoidbumping with the methyl substituent either into the trityl
(Fig.8A) or, with the other enantioface, into the first chain C-C
bond
(Fig. 8B). The alternative 1,2 insertion shown in Fig. 8C is
morefavorable as far as the nonbonded contacts of propene
areconcerned, but it is the chain now to interfere with the trityl.
Allthree transition states are practically at the same energy,
whichgives reason for the observed poor enantioselectivity in
thedominant insertion regiochemistry; we may define this
situationas ‘‘internally conflicting site control,’’ in the sense
that the directand indirect (chain-mediated) nonbonded steering of
the(ONNO) ligand on the inserting monomer are divergent. Underthese
circumstances, it is not at all surprising that 2,1 insertioncan be
competitive; the transition state in Fig. 8D, in particular,was
calculated to be higher in energy than the other three by only�1.6
kcal�mol�1 only, in very nice agreement with the
poorregioselectivity observed experimentally.
A very important issue that still needs to be discussed is
theeffect of R1 on the ease of �-H elimination to propene.
Onceagain in agreement with the QM indication (Table 1 and Fig.
6Right), high-molecular-mass polypropylenes were obtained
withcatalysts derived from the encumbered complexes 4–7 (Table
2).As already observed for 1 and 3, Pn turned out to be
independentof propene concentration, which confirms that �-H
eliminationto propene is indeed the main chain transfer pathway for
thesecatalysts, too.
An interesting feature of the whole catalyst family is
anunusually low [for Ziegler–Natta catalysis (26–28)]
turnoverfrequency, as if the polymerization occurred ‘‘at slow
motion.’’In particular, for the most sterically congested catalysts
4, 6, and7 we measured average chain growth times of several hours
at25°C (under the conditions described in Materials and
Methods).This controlled kinetic behavior (29) is highly
advantageous forapplication in block copolymerization (29, 30); we
have dis-cussed specifically this aspect in ref. 19, where we
documentedthe use of catalyst systems based on complex 4 for the
firstsyntheses of well defined and fully characterized samples
of(highly isotactic polypropylene)-block-(polyolefin).
ConclusionsIn our opinion, the results presented in the previous
section arerelevant for (at least) two reasons. First, they
demonstrate how
Fig. 8. QM transition states for the lowest energy 1,2 and 2,1
propene insertions into a growing polypropylene chain (simulated
with an iso-butyl group) ata model active cation of 7 with �
configuration. (A) 1,2 insertion�re�anti. (B) 1,2 insertion�si�syn.
(C) 1,2 insertion�si�anti. (D) 2,1 insertion�si. The mainnonbonded
interactions relevant for the enantioselectivity are explicitly
indicated by dotted lines.
Busico et al. PNAS � October 17, 2006 � vol. 103 � no. 42 �
15325
CHEM
ISTR
YSP
ECIA
LFE
ATU
RE
Dow
nloa
ded
by g
uest
on
July
3, 2
021
-
far one can get in the molecular control of Ziegler–Nattapropene
polymerizations with applications of computer-aidedcatalyst design.
But we do not want to hide that some importantlimitations still
need to be overcome; in particular, the ab initioprediction of
catalytic activity is somehow at a pioneering stage,mainly because
most active species in single-center Ziegler–Natta olefin
polymerizations are cationic (21), and realisticcalculations of
counterion and solvent effects have becomefeasible only recently
(31–34). However, computational chem-istry is progressing so fast
that it is plausible to anticipate thatmost residual problems will
be solved in a few years.
Second, the results are direct proof that, at least in
favorablecases, a proper fine-tuning of a single basic catalyst
structurewithout altering the nature and the symmetry of the
ancillaryligand framework can provide access to practically the
full rangeof polypropylene microstructures. This finding is at odds
with thelesson of metallocene catalysts (6, 7, 13) and provides a
strongargument in support of catalyst diversification programs
basedon parallel synthesis and structural amplification (10,
11).
Altogether, we believe that the above points to high-throughput
computation as the next-generation tool for Ziegler–Natta catalyst
discovery (ref. 35 and references therein).
Materials and MethodsAll reactants and solvents used for
precatalyst synthesis werepurchased from Aldrich (St. Louis, MO)
and used as received.Air-sensitive compounds were manipulated under
argon, usingSchlenk techniques and�or a Braun (Melsungen,
Germany)LabMaster 130 glove box. Complexes 1 and 3–5 were
preparedas described (19, 20). Complexes 2, 6, and 7 were
synthesizedwith similar procedures, as described in Supporting
Text.
The molecular structures of complexes 2 and 7 were deter-mined
by x-ray diffraction on single crystals grown from asaturated
solution of the complex in benzene (for 2) or toluene(for 7) at 293
K. Data collection was performed in flowing N2at 173 K on a
Bruker-Nonius (Delft, The Netherlands) kap-paCCD diffractometer
(MoK� radiation, CCD rotation images,thick slices, � scans � �
scans to fill the asymmetric unit). Fulldetails can be found in
Supporting Text. Crystallographic data inCIF format are available
from the Cambridge CrystallographicData Centre.
Propene polymerization runs were carried out at 25°C intoluene
at two different concentrations, [C3H6] � 1.3 M and 5.3M, mainly to
check for the effect of this parameter on polymeraverage molecular
mass. The experiments at lower propeneconcentration were carried
out in a 250-ml magnetically stirred,two-necked, jacketed Pyrex
reactor; those at higher concentra-tion were carried out in a
2-liter magnetically stirred, stainless-steel reactor (Brignole
AU-2). In all cases, the precatalyst wasdissolved in toluene and
activated with methylalumoxane[Crompton (Middlebury, CT) 10%
(wt�wt) solution in toluene]added with 2,6-di-tert-butylphenol
(Aldrich) to trap ‘‘free’’ tri-methylaluminum (23). The general
procedures can be found inSupporting Text.
Quantitative 1H and 13C NMR spectra of all polypropylenesamples
were recorded at 120°C, on 35 mg�ml solutions
intetrachloroethane-1,2-d2, with a Bruker DRX 400 Avance
spec-trometer operating at 100.6 MHz with a 5-mm BBO
probe.Conditions for 1H NMR were: 90° pulse; acquisition time, 4.0
s;relaxation delay, 2.0 s; 32 transients. Conditions for 13C
NMRwere: 80° pulse; acquisition time, 1.6 s; relaxation delay, 3.0
s;10,000 transients. Broad-band proton decoupling was achievedwith
a modified WALTZ16 sequence (BI�WALTZ16�32 byBruker). Peak
integration by full spectral simulation and best-fitcalculations of
stereosequence distributions were carried out byusing the SHAPE2004
and CONFSTAT (version 3.1 for Win-dows) software packages,
respectively. For more detail onpolymer microstructures and
statistical analysis thereof, seeSupporting Text.
All possible transition states of propene insertion and
�-Htransfer to propene monomer at model active cations werecomputed
by means of full QM methods. Stationary points onthe potential
energy surface were calculated with the Amster-dam Density
Functional program system, release 2004.01 (doc-umentation,
including user manuals and licensing, is available atwww.scm.com).
More information can be found in SupportingText and Fig. 11, which
is published as supporting information onthe PNAS web site.
This research was supported by the Italian Ministry of
Education,University and Research (PRIN 2004) and the Regional
Government ofCampania (Legge 5, 2003). The NMR polymer
characterizations andQM calculations were carried out at the Centro
Interdipartimentale diMetodologie Chimico-Fisiche of the University
of Naples Federico II.
1. Moore EP, Jr, ed (1996) Polypropylene Handbook:
Polymerization, Character-ization, Properties, Applications
(Hanser, Munich).
2. Moore EP, Jr (1998) The Rebirth of Polypropylene: Supported
Catalysts (Hanser,Munich).
3. Scheirs V, Kaminsky W, eds (1999) Metallocene-Based
Polyolefins (Wiley,Chichester, UK), Vols 1 and 2.
4. Boor J, Jr (1979) Ziegler–Natta Catalysts and Polymerizations
(Academic, NewYork).
5. Bovey FA (1972) High-Resolution NMR of Macromolecules
(Academic, NewYork).
6. Busico V, Cipullo R (2001) Progr Polym Sci 26:443–533.7.
Resconi L, Cavallo L, Fait A, Piemontesi F (2000) Chem Rev
100:1253–1345.8. Ziegler K (1954) Brennst-Chem 35:321–325.9. Sinn
H, Kaminsky W (1980) Adv Organomet Chem 18:99–149.
10. Boussie TR, Diamond GM, Goh C, Hall KA, LaPointe AM, Leclerc
M, LundC, Murphy V, Shoemaker JAW, Tracht U, et al. (2003) J Am
Chem Soc125:4306–4317.
11. Boussie TR, Diamond GM, Goh C, Hall KA, LaPointe AM, Leclerc
MK,Murphy V, Shoemaker JAW, Turner H, Rosen RK, et al. (2006) Angew
ChemInt Ed 45:3278–3283.
12. Arriola DJ, Carnahan EM, Hustad PD, Kuhlman RL, Wenzel TT
(2006)Science 312:714–719.
13. Alt HG, Koppl A (2000) Chem Rev 100:1205–1222.14. Gibson VC,
Spitzmesser SK (2003) Chem Rev 103:283–315.15. Corradini P, Guerra
G, Cavallo L (2004) Acc Chem Res 37:231–241.16. Makio H, Kashiwa N,
Fujita T (2002) Adv Synth Catal 344:477–493.
17. Busico V, Cipullo R, Ronca S, Budzelaar PHM (2001) Macromol
RapidCommun 22:1405–1410.
18. Busico V, Cipullo R, Friederichs N, Ronca S, Togrou M (2003)
Macromolecules36:3806–3808.
19. Busico V, Cipullo R, Friederichs N, Ronca S, Talarico G,
Togrou M, Wang B(2004) Macromolecules 37:8201–8203.
20. Tshuva ET, Goldberg I, Kol M (2000) J Am Chem Soc
122:10706–10707.21. Chen, EY-X, Marks TJ (2000) Chem Rev
100:1391–1434.22. Talarico G, Busico V, Cavallo L (2003) J Am Chem
Soc 125:7172–7173.23. Busico V, Cipullo R, Cutillo F, Friederichs
N, Ronca S, Wang B (2003) J Am
Chem Soc 125:12402–12403.24. Deng L, Woo TK, Cavallo L, Margl P,
Ziegler T (1997) J Am Chem Soc
119:6177–6186.25. Margl P, Deng L, Ziegler T (1999) J Am Chem
Soc 121:154–162.26. Mori H, Terano M (1997) Trends Polym Sci
5:314–321.27. Busico V, Cipullo R, Esposito V (1999) Macromol Rapid
Commun 20:116–121.28. Song F, Cannon RD, Bochmann M (2003) J Am
Chem Soc 125:7641–7653.29. Busico V, Talarico G, Cipullo R (2005)
Macromol Symp 226:1–16.30. Coates GW, Hustad PD, Reinartz S (2002)
Angew Chem Int Ed 41:2236–2257.31. Chan MSW, Ziegler T (2000)
Organometallics 19:5182–5189.32. Lanza G, Fragalà IL, Marks TJ
(2002) Organometallics 21:5594–5612.33. Busico V, Van Axel Castelli
V, Aprea P, Cipullo R, Segre AL, Talarico G,
Vacatello M (2003) J Am Chem Soc 125:5451–5460.34. Correa A,
Cavallo L (2006) J Am Chem Soc 128:10952–10959.35. Miteva MA, Lee
WH, Montes MO, Villoutreix BO (2005) J Med Chem
48:6012–6022.
15326 � www.pnas.org�cgi�doi�10.1073�pnas.0602856103 Busico et
al.
Dow
nloa
ded
by g
uest
on
July
3, 2
021
