Quantitative effects of ion pairing and sterics on chain propagation kinetics for 1-hexene polymerization catalyzed by mixed Cp′/ArO complexes

Article abstract of DOI:10.1021/om8004993

A series of single-site catalysts with mixed cyclopentadienyl/aryloxide ligation were synthesized and used to polymerize 1-hexene. The effects of solvent, metal, counterion, and ligand structure were investigated by experiments and by DFT calculations. A solvent with a high dielectric constant led not only to an increase in the chain propagation rate but also to a change in the reaction order. Catalyst reactivity was found to be controlled by the difficulty of ion pair separation and steric congestion at the metal center, which were quantified from DFT simulation by SCF ion pair separation energies and ligand cone angles. A Cp*Ti(OC₆H₄-2-Br)Me ₂/B(C₆F₅)₃ catalyst exhibits unusually high reactivity, which was correlated to the formation of a partial bond between the aryloxide ortho substituent bromide and Ti upon ion pair separation. Natural bond orbital analysis was used to quantify the order of this opportunistic ligation for a series of catalysts. Kinetic analysis and a structure-activity correlation were applied to interpret the experimental results.

Full text of DOI:10.1021/om8004993

5504
Organometallics 2008, 27, 5504–5520
Quantitative Effects of Ion Pairing and Sterics on Chain
Propagation Kinetics for 1-Hexene Polymerization Catalyzed by
Mixed Cp′/ArO Complexes
Thomas A. Manz,† Shalini Sharma,‡ Khamphee Phomphrai,‡ Krista A. Novstrup,†
Andrew E. Fenwick,‡ Phillip E. Fanwick,‡ Grigori A. Medvedev,† Mahdi M. Abu-Omar,‡
W. Nicholas Delgass,† Kendall T. Thomson,† and James M. Caruthers*,†
School of Chemical Engineering and Department of Chemistry, Purdue UniVersity,
West Lafayette, Indiana 47907
ReceiVed June 1, 2008
A series of single-site catalysts with mixed cyclopentadienyl/aryloxide ligation were synthesized and
used to polymerize 1-hexene. The effects of solvent, metal, counterion, and ligand structure were
investigated by experiments and by DFT calculations. A solvent with a high dielectric constant led not
only to an increase in the chain propagation rate but also to a change in the reaction order. Catalyst
reactivity was found to be controlled by the difficulty of ion pair separation and steric congestion at the
metal center, which were quantified from DFT simulation by SCF ion pair separation energies and ligand
cone angles. A Cp*Ti(OC6H4-2-Br)Me2/B(C6F5)3 catalyst exhibits unusually high reactivity, which was
correlated to the formation of a partial bond between the aryloxide ortho substituent bromide and Ti
upon ion pair separation. Natural bond orbital analysis was used to quantify the order of this opportunistic
ligation for a series of catalysts. Kinetic analysis and a structure-activity correlation were applied to
interpret the experimental results.
the monomer to reach the catalytic site, the counterion must be
partially displaced; thus, a weakly bound counterion facilitates
monomer coordination and insertion. Ziegler et al. and Fragala
et al. have used computational chemistry to study the effects of
solvent on ion pair formation and separation energies8,9 as well
as on reaction barriers for chain initiation and propagation.10
The presence of solvent was found to dramatically lower the
ion pair separation energies and overall chain propagation
barriers. Since partial separation of the ion pair occurs in the
transition states for chain propagation, the lower barriers to chain
propagation in the presence of solvent were presumably due to
lower ion pair separation energies.
A key question is how to quantitatively model structure-
activity relationships in single-site olefin polymerization ca-
talysis. While it is known that lower ion pair separation energies
and higher solvent dielectric constants can enhance polymeri-
zation activity, only recently Manz et al. presented a quantitative
model explaining how much the chain propagation rate constant
increases for a given change in ligand structure.11 An important
question is whether the structure-activity correlation of Manz
et al. can be extended to include changes in metal, counterion,
and solvent.
Introduction
Homogeneous single-site olefin polymerization catalysts are
commercially important because the catalyst structure can be
engineered to control the polymer molecular weight distribution
and tacticity.1,2 These complexes contain a transition metal
bound to organic ligands and are often used to polymerize
olefins in the condensed phase. Most commonly these complexes
consist of a positively charged metal ion of group IV associated
with a bulky and weakly coordinating counteranion.
It has been established that the nature of the counterion has
a large effect on catalyst reactivity and stereoselectivity,3-5 as
reviewed by Chen and Marks6 and Pedeutour et al.7 Marks and
co-workers have shown that the counterion needs to be weakly
coordinating to achieve high catalyst activity.3,6 In order for
* To whom correspondence should be addressed. E-mail: caruther@
ecn.purdue.edu.
† School of Chemical Engineering.
‡ Department of Chemistry.
(1) (a) Alt, H. G.; Koppl, A. Chem. ReV. 2000, 100, 1205–1221. (b)
Coates, G. W.; Hustad, P. D.; Reinartz, S. Angew. Chem., Int. Ed. 2002,
41, 2236–2257.
(2) Chem. ReV. 2000, 100 (entire issue No. 4).
(3) Chen, M. C.; Roberts, J. A. S.; Marks, T. J. J. Am. Chem. Soc. 2004,
126, 4605–4625.
(4) (a) Wilson, P. A.; Hannant, M. H.; Wright, J. A.; Cannon, R. D.;
Bochmann, M. Macromol. Symp. 2006, 236, 100–110. (b) Deck, P. A.;
Beswick, C. L.; Marks, T. J. J. Am. Chem. Soc. 1998, 120, 12167–12167.
(c) Luo, L.; Marks, T. J. Top. Catal. 1999, 7, 97–106.
(5) (a) Yang, X. M.; Stern, C. L.; Marks, T. J. J. Am. Chem. Soc. 1994,
116, 10015–10031. (b) Jia, L.; Yang, X. M.; Ishihara, A.; Marks, T. J.
Organometallics 1995, 14, 3135–3137. (c) Chen, M. C.; Roberts, J. A. S.;
Marks, T. J. Organometallics 2004, 23, 932–935. (d) Metz, M. V.; Schwartz,
D. J.; Stern, C. L.; Marks, T. J.; Nickias, P. N. Organometallics 2002, 21,
4159–4168. (e) Metz, M. V.; Sun, Y. M.; Stern, C. L.; Marks, T. J.
Organometallics 2002, 21, 3691–3702.
(8) (a) Lanza, G.; Fragala, I. L.; Marks, T. J. J. Am. Chem. Soc. 1998,
120, 8257–8258. (b) Xu, Z. T.; Vanka, K.; Firman, T.; Michalak, A.; Zurek,
E.; Zhu, C. B.; Ziegler, T. Organometallics 2002, 21, 2444–2453.
(9) (a) Chan, M. S. W.; Vanka, K.; Pye, C. C.; Ziegler, T. Organome-
tallics 1999, 18, 4624–4636. (b) Lanza, G.; Fragala, I. L.; Marks, T. J.
J. Am. Chem. Soc. 2000, 122, 12764–12777. (c) Vanka, K.; Chan, M. S. W.;
Pye, C. C.; Ziegler, T. Organometallics 2000, 19, 1841–1849.
(10) (a) Lanza, G.; Fragala, I. L. Top. Catal. 1999, 7, 45–60. (b) Lanza,
G.; Fragala, I. L.; Marks, T. J. Organometallics 2002, 21, 5594–5612. (c)
Xu, Z. T.; Vanka, K.; Ziegler, T. Organometallics 2004, 23, 104–116.
(11) Manz, T. A.; Phomphrai, K.; Medvedev, G.; Krishnamurthy, B. B.;
Sharma, S.; Haq, J.; Novstrup, K. A.; Thomson, K. T.; Delgass, W. N.;
Caruthers, J. M.; Abu-Omar, M. M. J. Am. Chem. Soc. 2007, 129, 3776–
3777.
(6) Chen, E. Y. X.; Marks, T. J. Chem. ReV. 2000, 100, 1391–1434.
(7) Pedeutour, J. N.; Radhakrishnan, K.; Cramail, H.; Deffieux, A.
Macromol. Rapid Commun. 2001, 22, 1095–1123.
(12) Macchioni, A. Chem. ReV. 2005, 105, 2039–2073.
10.1021/om8004993 CCC: $40.75
2008 American Chemical Society
Publication on Web 10/08/2008

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5505
Ti/Zr mixed Cp′/ArO catalysts have been used to polymerize
1-hexene,11,20-23 propylene,16,21,24,25 ethylene,16,21,24-28 and
styrene,22,24,29 and these catalysts are well suited for copolym-
erization involving ethylene/1-hexene,21,28,30 ethylene/1-butene,21,26
ethylene/styrene,31 ethylene/cyclohexene,32 and ethylene/nor-
bornene.33 Most of these studies focused on activation of the
dichloride precursors by methylaluminoxane (MAO). Changes
in the ligand structure were found to affect the polymerization
rate, polymer molecular weight, and degree of comonomer
incorporation. Nomura and co-workers, who have done exten-
sive work on these catalyst systems and written a review of
them,34 have recently developed a method for tethering mixed
Cp′/ArO catalysts and found that the tethered catalysts have
catalytic properties similar to those of the untethered species.35
Polymerization catalysts have also been synthesized with a
bridge connecting the Cp′ and ArO ligands.24,36 On the basis
of this data, mixed Cp′/ArO ligated catalysts afford a number
of advantages for detailed kinetic studies. First, they are active
for the polymerization of 1-hexene, which allows the monomer
to be added in the liquid phase close to ambient temperature
and pressure. Second, the structure of the aryloxide ligand can
be conveniently varied, owing to the availability of different
phenols. Finally, changing the structure of the aryloxide ligand
allows the electronic and steric properties of these catalysts to
be systematically tuned.
Figure 1. Example of a single-site olefin polymerization catalyst
containing mixed cyclopentadienyl/aryloxide ligation: the front-
side ethylene π-complex of [CpTi(OC6H3-2,6-Me2-4-Br)C3H7]+
[MeB(C6F5)3]-.
Another important question is what roles different forms of
the ion pair play in the polymerization kinetics. Macchioni
reviewed ion pairing in transition-metal organometallic chem-
istry, including olefin polymerization.12 Four types of transition-
metal ion pairs were discussed: (a) an inner-sphere ion pair
(ISIP) in which the counterion exists in the first coordination
sphere of the metal, (b) an outer-sphere ion pair (OSIP) in which
the counterion exists in the second coordination sphere of the
metal, (c) a solvent-shared ion pair or solvated cation-anion
pair (SCAP) in which a single solvent layer separates the cation
from the counterion, and (d) a solvent-separated ion pair (SSIP)
in which two or more solvent layers separate the cation from
the counterion. All four of these ion pair types have been
experimentally and computationally observed for some single-
site olefin polymerization systems.9,13-17 Which ion pair type
is the most favored depends upon the catalyst structure, solvent,
and other reaction conditions. In addition, aggregates consisting
of four, eight, or more ions in a cluster are possible and have
been experimentally observed at higher catalyst concentrations,15,17
but some studies indicate that aggregates are not likely to form
at the low catalyst concentrations typically employed in po-
lymerization reactions.17,18 Different forms of each of these ion
pair types are also possible. For example, an OSIP may exist
with or without agostic bonding between the growing polymer
chain and metal center, and several papers have investigated
the details of these agostic interactions.15,19
There are five basic ways we will vary the ion pair behavior
in the Ti/Zr mixed Cp′/ArO systems: (a) change the metal, (b)
change the solvent, (c) change the ligands to alter steric and
electronic properties, (d) change the activator/counterion, and
(20) (a) Nomura, K.; Fudo, A. Inorg. Chim. Acta 2003, 345, 37–43. (b)
Nomura, K.; Fudo, A. J. Mol. Catal. A: Chem. 2004, 209, 9–17.
(21) Nomura, K.; Naga, N.; Miki, M.; Yanagi, K.; Imai, A. Organo-
metallics 1998, 17, 2152–2154.
(22) Nomura, K.; Tanaka, A.; Katao, S. J. Mol. Catal. A: Chem. 2006,
254, 197–205.
(23) Phomphrai, K.; Fenwick, A. E.; Sharma, S.; Fanwick, P. E.;
Caruthers, J. M.; Delgass, W. N.; Abu-Omar, M. M.; Rothwell, I. P.
Organometallics 2006, 25, 214–220.
(24) Chen, Y. X.; Fu, P. F.; Stern, C. L.; Marks, T. J. Organometallics
1997, 16, 5958–5963.
(25) Ewart, S. W.; Sarsfield, M. J.; Jeremic, D.; Tremblay, T. L.;
Williams, E. F.; Baird, M. C. Organometallics 1998, 17, 1502–1510.
(26) Nomura, K.; Naga, N.; Miki, M.; Yanagi, K. Macromolecules 1998,
31, 7588–7597.
(27) (a) Antinolo, A.; Carrillo-Hermosilla, F.; Corrochano, A.; Fernan-
dez-Baeza, J.; Lara-Sanchez, A.; Ribeiro, M. R.; Lanfranchi, M.; Otero,
A.; Pellinghelli, M. A.; Portela, M. F.; Santos, J. V. Organometallics 2000,
19, 2837–2843. (b) Zhang, Y. T.; Mu, Y.; Lu, C. S.; Li, G. H.; Xu, J. S.;
Zhang, Y. R.; Zhu, D. S.; Feng, S. H. Organometallics 2004, 23, 540–546.
(28) (a) Rau, A.; Schmitz, S.; Luft, G. J. Organomet. Chem. 2000, 608,
71–75. (b) Zhang, Y. T.; Wang, J. H.; Mu, Y.; Shi, Z.; Lu, C. S.; Zhang,
Y. R.; Qiao, L. J.; Feng, S. H. Organometallics 2003, 22, 3877–3883.
(29) (a) Byun, D. J.; Fudo, A.; Tanaka, A.; Fujiki, M.; Nomura, K.
Macromolecules 2004, 37, 5520–5530. (b) Nomura, K.; Fudo, A. Catal.
Commun. 2003, 4, 269–274. (c) Nomura, K.; Komatsu, T.; Imanishi, Y.
Macromolecules 2000, 33, 8122–8124.
(30) (a) Nomura, K.; Oya, K.; Imanishi, Y. J. Mol. Catal. A: Chem.
2001, 174, 127–140. (b) Nomura, K.; Komatsu, T.; Imanishi, Y. J. Mol.
Catal. A: Chem. 2000, 159, 127–137. (c) Nomura, K.; Komatsu, T.;
Imanishi, Y. J. Mol. Catal. A: Chem. 2000, 152, 249–252. (d) Nomura, K.;
Oya, K.; Komatsu, T.; Imanishi, Y. Macromolecules 2000, 33, 3187–3189.
(31) (a) Nomura, K.; Okumura, H.; Komatsu, T.; Naga, N. Macromol-
ecules 2002, 35, 5388–5395. (b) Nomura, K.; Okumura, H.; Komatsu, T.;
Naga, N.; Imanishi, Y. J. Mol. Catal. A: Chem. 2002, 190, 225–234.
(32) Wang, W.; Fujiki, M.; Nomura, K. J. Am. Chem. Soc. 2005, 127,
4582–4583.
In order to address these two key questions, we now report
experimental and DFT results for a series of Ti and Zr catalysts
containing mixed cyclopentadienyl/aryloxide ligation. Dimethyl
catalyst precursors were activated with B(C6F5)3 or
Ph3CB(C6F5)4. An example of a catalyst is shown in Figure 1,
where the counterion is displaced from the metal center and
turned sideways to allow room for the monomer to dock.
(13) Bochmann, M.; Cannon, R. D.; Song, F. Kinet. Catal. 2006, 47,
160–169.
(14) Eisch, J. J.; Pombrik, S. I.; Zheng, G. X. Organometallics 1993,
12, 3856–3863.
(15) Song, F. Q.; Lancaster, S. J.; Cannon, R. D.; Schormann, M.;
Humphrey, S. M.; Zuccaccia, C.; Macchioni, A.; Bochmann, M. Organo-
metallics 2005, 24, 1315–1328.
(16) Tremblay, T. L.; Ewart, S. W.; Sarsfield, M. J.; Baird, M. C. Chem.
Commun. 1997, 831–832.
(17) Zuccaccia, C.; Stahl, N. G.; Macchioni, A.; Chen, M. C.; Roberts,
J. A.; Marks, T. J. J. Am. Chem. Soc. 2004, 126, 1448–1464.
(18) Stahl, N. G.; Zuccaccia, C.; Jensen, T. R.; Marks, T. J. J. Am. Chem.
Soc. 2003, 125, 5256–5257.
(33) (a) Nomura, K.; Tsubota, M.; Fujiki, M. Macromolecules 2003,
36, 3797–3799. (b) Wang, W.; Tanaka, T.; Tsubota, M.; Fujiki, M.;
Yamanaka, S.; Nomura, K. AdV. Synth. Catal. 2005, 347, 433–446.
(34) Nomura, K.; Liu, J. Y.; Padmanabhan, S.; Kitiyanan, B. J. Mol.
Catal. A: Chem. 2007, 267, 1–29.
(19) (a) Brookhart, M.; Green, M. L. H.; Parkin, G. Proc. Natl. Acad.
Sci. USA 2007, 104, 6908–6914. (b) Nifant’ev, I. E.; Ustynyuk, L. Y.;
Laikov, D. N. Organometallics 2001, 20, 5375–5393.
(35) Kitiyanan, B.; Nomura, K. Organometallics 2007, 26, 3461–3465.
(36) Xu, J. S.; Zhang, Y. T.; Li, G. H.; Mu, Y.; Feng, S. H. Polyhedron
2006, 25, 3071–3076.

5506 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
Scheme 1. Catalyst Precursor Structures
Cp*TiCl3 (0.500 g, 1.73 mmol) in 30 mL of Et2O at 0 °C. The
mixture was stirred at 0 °C for 1 h and at room temperature for
2 h. The mixture was then cooled back to 0 °C and a solution of
2-bromophenol (0.299 g, 1.73 mmol) in 10 mL of Et2O was added
dropwise. The mixture was slowly warmed to room temperature
and was stirred overnight. The solvent was removed under vacuum,
and benzene was added to the solid residue. The suspension was
filtered through a plug of Celite over fritted glass to remove the
lithium salts. The filtrate was then evacuated to dryness, yielding
a yellow solid (0.52 g, 78%). Anal. Calcd for C18H25BrOTi: C,
56.16; H, 6.49. Found: C, 55.84; H, 6.52. 1H NMR (C6D6, 25 °C):
δ 7.45 (d, 1H, Ar H); 6.93 (t, 1H, Ar H); 6.85 (d, 1H, Ar H); 6.51
(t, 1H, Ar H); 1.76 (s, 15H, Cp*); 0.86 (s, 6H, Ti-Me). Selected
13C NMR (C6D6, 25 °C): δ 161.1 (Ti-O-C); 123.0 (C5Me5); 57.3
(Ti--Me); 11.5 (C5Me5).
Cp*Ti(OC6H4-2-Ph)Me2 (21). LiMe (3.24 mL, 1.6 M sol. in
diethyl ether, 5.18 mmol) was added dropwise to a suspension of
Cp*TiCl3 (0.500 g, 1.73 mmol) in 30 mL of Et2O at 0 °C. The
mixture was stirred at 0 °C for 1 h and at room temperature for
2 h. The mixture was then cooled back to 0 °C, and a solution of
2-phenylphenol (0.294 g, 1.73 mmol) in 10 mL of Et2O was added
dropwise. The mixture was slowly warmed to room temperature
and was stirred overnight. The solvent was removed under vacuum,
and benzene was added to the solid residue. The suspension was
filtered through a plug of Celite over fritted glass to remove the
lithium salts. The filtrate was evacuated to dryness, yielding a yellow
solid (0.62 g, 94%). Anal. Calcd for C24H30OTi: C, 75.43; H, 7.85.
Found: C, 75.68; H, 7.91. 1H NMR (C6D6, 25 °C): δ 7.68 (d, 2H,
Ar H); 6.90-7.40 (m, Ar H); 1.56 (s, 15H, Cp*); 0.68 (s, 6H,
Ti-Me). Selected 13C NMR (C6D6, 25 °C): δ 161.9 (Ti-O-C);
122.9 (C5Me5); 55.7 (Ti-Me); 11.6 (C5Me5).
(e) change the chain-initiating group. Each of these five variants
was studied experimentally and computationally; however, the
effects of chain initiating group will be presented in a later paper
dealing with chain initiation kinetics. Herein, we focus on the
chain propagation kinetics.
Experimental Section
Catalyst Precursor Synthesis. Scheme 1 shows the chemical
structure of a mixed Cp′/ArO dimethyl catalyst precursor. The
particular precatalysts synthesized are 1-24, 30, and 31 of Table
1. Extensive precautions were taken to exclude air and moisture
and other sources of contamination during catalyst preparation and
use. All synthesis and polymerization experiments were carried out
using either standard Schlenk line techniques or a circulating
nitrogen-filled glovebox operating at less than 0.2 ppm of oxygen.
The hydrocarbon solvents were distilled from sodium/benzophenone
or purified using an Innovative Technologies solvent purification
system and were stored over sodium ribbons under nitrogen until
use.
Synthesis details for most of the precatalysts used here were
previously reported.11,23,26,37 In addition, the following new catalyst
precursors were synthesized: 20-24, 30, and 31. Additional
synthesis details for 24 can be found in the thesis by Sharma.38
The precursors were activated with 1 equiv of B(C6F5)3 or
Ph3CB(C6F5)4 and used to polymerize 1-hexene. Activation with
B(C6F5)3 is denoted by adding a after the precursor (e.g., 10a),
while activation with Ph3CB(C6F5)4 is denoted by adding b after
the precursor (e.g., 10b).
IndTi(OC6H-2,3,5,6-Ph4)Me2 (22). A solution of IndTiCl3 (1.00
g, 3.71 mmol) in 50 mL of dichloromethane was cooled to -78
°C in a dry ice/acetone bath. To this solution was added MeMgBr
(3.71 mL, 3.0 M in diethyl ether, 11.1 mmol) dropwise via syringe
under a flush of nitrogen. After the mixture was stirred for 1 h at
-78 °C, a solution of 2,3,5,6-tetraphenylphenol (1.48 g, 3.71 mmol)
in dichloromethane was added dropwise. The mixture was slowly
warmed to room temperature and stirred for 2 days. The solvent
was removed under vacuum, and benzene was added to the solid
residue. The suspension was filtered through a plug of Celite over
fritted glass to remove magnesium salts. The filtrate was evacuated
to dryness, yielding a yellow powder (1.71 g, 78%). Crystals
suitable for single-crystal X-ray crystallography were obtained by
layering hexane on a benzene solution. Anal. Calcd for C41H34OTi:
1
C, 83.42; H, 5.76. Found: C, 83.51; H, 6.11. H NMR (C6D6, 25
°C): δ 6.80-7.40 (m, Ar H); 5.72 (d, J ) 3.0 Hz, 2H, Ind); 5.11
(t, J ) 3.0 Hz, 1H, Ind); 0.00 (s, 6H, Ti-Me). Selected 13C NMR
(C6D6, 25 °C): δ 161.8 (Ti-O-C); 116.6, 104.7 (Ind); 61.5
(Ti-Me).
The following measures were taken to ensure that the precatalysts
were of acceptable purity. In general, the amounts of LiMe and
other chemicals have to be quite exact to get clean products. Care
was taken to use a relatively fresh bottle of LiMe so that there
could be certainty regarding its concentration and purity. Precata-
IndTi(OC6H3-2,6-iPr2)Me2 (23). A solution of IndTiCl3 (1.00
g, 3.72 mmol) in 30 mL of dichloromethane was cooled to -78
°C in a dry ice/acetone bath. To this solution was added MeMgCl
(3.72 mL, 3.0 M in THF, 11.2 mmol) dropwise via syringe under
a flush of nitrogen. After the mixture was stirred for 1 h at -78
°C, a solution of 2,6-diisopropylphenol (0.663 g, 3.72 mmol) in
dichloromethane was added dropwise. The mixture was slowly
warmed to room temperature and stirred for 1 h. The solvent was
removed under vacuum, and benzene was added to the solid residue.
The suspension was filtered through a plug of Celite over fritted
glass to remove magnesium salts. The filtrate was evacuated to
dryness, yielding a dark red viscous liquid (1.16 g, 84%). Anal.
Calcd for C23H30OTi: C, 74.64; H, 8.11. Found: C, 74.15; H, 7.74.
1H NMR (C6D6, 25 °C): δ 7.28 (m, 2H, Ind); 7.12 (d, J ) 7.8 Hz,
2H, m-H); 7.01 (t, J ) 7.8 Hz, 1H, p-H); 6.89 (m, 2H, Ind); 6.18
(d, J ) 3.3 Hz, 2H, Ind); 5.87 (t, J ) 3.3 Hz, 1H, Ind); 3.28 (m,
J ) 6.9 Hz, 2H, CHMe2); 1.25 (d, J ) 6.9 Hz, 12H, CHMe2); 0.67
1
lysts were checked for purity by H NMR and elemental analysis
before use. The whole batch of precatalyst was discarded and
1
remade if the purity was less than ca. 90% by H NMR. All solid
precatalysts were stored inside a drybox at room temperature and
recrystallized before use. (Microcrystals were grown by placing a
concentrated hexane or toluene solution in a freezer overnight.)
All liquid precatalysts were stored at -20 °C in a drybox freezer
and discarded 2-3 weeks after synthesis.
Cp*Ti(OC6H4-2-Br)Me2 (20). LiMe (3.24 mL, 1.6 M sol. in
diethyl ether, 5.18 mmol) was added dropwise to a suspension of
(37) Fenwick, A. E.; Phomphrai, K.; Thorn, M. G.; Vilardo, J. S.; Trefun,
C. A.; Hanna, B.; Fanwick, P. E.; Rothwell, I. P. Organometallics 2004,
23, 2146–2156.
(38) Sharma, S. Ph.D. Thesis, Chemistry Department, Purdue University,
2005.

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5507
Table 1. Computed Ion Pair Separation Energies and Ligand Cone Angles for [Cp′m(OAr)Me]+[A]- a
SCF EIPS (kcal/mol)
fitted param (kcal/mol)
precat no.
m
Cp′ OAr substituent
[A]-
θCp′ (deg) θOAr (deg)
vac
C6H5-Me C6H5-Br C6H4-1,2-Cl2
B
C
rms err
1
2
3
4
5
6
7
8
9
Ti Cp
Ti Cp
Ti Cp
Ti Cp* none
Ti Cp* 4-F
Ti Cp* 4-Cl
Ti Cp* 4-Br
Ti Cp* 4-Ph
Ti Cp* 4-tBu
Ti Cp* 2,6-Me2
2,6-Me2-4-Br
2,6-Et2
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[BQ4]-
128
129
130
155
156
156
156
156
156
154
158
155
158
154
159
155d
156p
155
155d
156p
154
156d
155p
156
130
130
154d
155p
154d
154p
136
136
149
149
151
156d
155p
155d
155p
154d
156p
155
155
154
129
132
136
96
96
96
96
96
95
74.01
72.16
69.83
66.02
67.10
67.06
67.06
65.27
64.70
62.00
52.24
61.26
51.44
59.53
50.67
63.56d
64.89p
63.25
57.30d
65.46p
46.57
65.14d
65.19p
63.75
55.19
45.72
56.88d
57.70p
59.23d
59.54p
53.04
63.99
53.26
69.73
57.38
63.22d
63.36p
59.22d
60.06p
53.37d
54.51p
64.66
53.70
61.12
35.74
34.63
34.07
31.42
32.08
32.09
31.80
31.20
30.95
28.88
23.38
28.39
22.71
26.64
22.34
30.17d
30.91p
29.61
23.86d
31.10p
14.06
30.72d
31.06p
30.24
20.89
17.12
23.63d
24.19p
25.36d
26.02p
19.61
28.97
20.21
36.25
27.81
29.92d
29.63p
26.03d
26.09p
20.26d
21.04p
29.79
21.26
27.92
19.65
18.75
19.27
16.98
17.81
17.80
17.15
17.04
16.67
15.37
12.15
14.86
11.41
13.42
11.33
16.71d
16.95p
15.72
10.57d
16.86p
0.78
13.70
12.94
13.90
11.69
12.49
12.47
11.75
11.84
11.40
10.52
8.18
7.23
6.59
7.76
5.80
6.51
6.50
5.75
6.03
5.71
4.78
3.05
4.31
2.32
2.95
2.41
6.06d
6.02p
4.91
66.92
65.72
62.13
60.32
60.63
60.60
61.40
59.32
59.13
57.24
49.08
56.99
49.04
56.56
48.14
57.48d
58.92p
58.39
0.2
0.3
0.1
0.2
0.1
0.1
0.2
0.1
0.2
0.0
0.2
0.1
0.1
0.0
2,6-iPr2
10
127
130
132
140
136
138
125d
127p
127
113d
111p
163
96d
11
12
13
Ti Cp* 2,6-Et2
[MeBQ3]-
[BQ4]-
9.98
7.40
8.69
7.46
Ti Cp* 2,6-iPr2
Ti Cp* 2-cyclohexyl
[MeBQ3]-
[BQ4]-
0.2
[MeBQ3]-
11.88d
11.85p
10.70
5.85d
11.63p
-4.04
11.24d
11.80p
10.98
1.54
0.0d
0.1p
0.1
14
15
Ti Cp* 2,6-Me2-4-Br
Ti Cp* 2-CH2Ph
[MeBQ3]-
[MeBQ3]-
-0.03d 57.27d
0.1d
0.1p
0.0
5.77p 59.77p
16
17
Ti Cp* 2,3,5,6-Ph4
Ti Cp* 3-OMe
[MeBQ3]-
[MeBQ3]-
-9.65 56.24
16.48d
18.17p
16.17
6.70
5.37d
6.51p
5.31
59.85d
58.70p
58.56
0.1d
0.5p
0.2
96p
96
18
19
Ti Cp* 4-OMe
Ti Cp
[MeBQ3]-
[MeBQ3]-
[BQ4]-
2,3,5,6-Ph4
167
188
118d
117p
127d
128p
169
138
155
134
133
105d
105p
114d
114p
120d
120p
96
-4.34 59.60
-3.32 48.98
-0.93d 57.88d
-0.03p 57.72p
0.1
0.1
5.76
1.71
20
21
Ti Cp* 2-Br
Ti Cp* 2-Ph
[MeBQ3]-
9.66d
10.65p
11.78d
12.59p
5.80
4.87d
5.82p
6.92d
7.78p
0.75
0.1d
0.0p
0.1d
0.1p
0.1
[MeBQ3]-
1.00d
1.92p
58.20d
57.59p
22
23
24
25
Ti Ind 2,3,5,6-Ph4
Ti Ind 2,6-iPr2
Zr Cp* 2,3,5,6-Ph4
Zr Cp* 2,6-iPr2
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
[BQ4]-
-4.97 58.07
3.46 60.55
-4.26 57.61
14.68
6.43
9.52
1.36
0.0
0.2
0.0
0.1
22.60
16.03
16.54d
15.97p
12.63d
12.53p
6.85d
7.48p
15.10
7.95
17.66
11.83
11.78d
11.09p
7.86d
7.69p
2.08d
2.62p
9.68
11.87
6.64
57.88
50.69
26
27
28
Ti Cp* 2-F
Ti Cp* 2-Cl
Ti Cp* 2-I
[MeBQ3]-
5.95d
5.21p
2.06d
1.74p
57.24d
58.14p
57.13d
58.27p
0.1d
0.0p
0.0d
0.1p
0.0d
0.0p
0.2
[MeBQ3]-
[MeBQ3]-
-3.70d 57.06d
-3.20p 57.71p
29
30
31
Ti Cp* 4-I
[MeBQ3]-
[MeBQ3]-
[MeBQ3]-
3.79
-2.43 56.16
3.87 57.25
61.01
Ti Cp* 2,6-(OMe)2
135
135
3.18
9.64
0.1
0.0
Ti Cp* 2,6-iPr2-4-Br
14.47
a Legend: d, aryloxide substituent distal to initiating group; p, aryloxide substituent proximal to initiating group. Q ) C6F5.
(s, 6H, Ti-Me). 13C NMR (C6D6, 25 °C): δ 161.6 (Ti-O-C);
138.6, 127.0, 125.9, 125.7, 123.8, 122.9 (aromatics); 117.3, 104.6
(Ind); 58.2 (Ti-Me); 27.6 (CHMe2); 24.1 (CHMe2).
solid (0.21 g, 74%). The solid (22 mg) was dissolved in a minimum
amount of hot benzene, and this solution was layered with hexane
and allowed to sit overnight to give white needle-shaped crystals
of the title compound as a benzene solvate (15 mg, 68%). Anal.
Calcd for C42H42OZr. C6H6: C, 78.75; H, 6.61. Found: C, 78.44;
Cp*Zr(OC6H-2,3,5,6-Ph4)Cl2. A sample of Li(OC6H-2,3,5,6-
Ph4) · Et2O (0.65 g, 1.36 mmol) was added to a suspension
containing Cp*ZrCl3 (0.45 g, 1.35 mmol) in 50 mL of benzene.
The resulting solution was stirred at room temperature overnight
and filtered through a plug of Celite over fritted glass to remove
LiCl. The solvent was removed under vacuum to give a white solid
(0.86 g, 91%). A small amount of this solid (20 mg) was dissolved
in a minimum amount of hot benzene and layered with hexane,
affording white crystals of Cp*Zr(OC6H-2,3,5,6-Ph4)Cl2 as a
benzene solvate (11 mg, 55%). Anal. Calcd for C40H36OZrCl2: C,
69.15; H, 5.22. Found: C, 69.87; H, 5.64. 1H NMR (C6D6, 25 °C):
δ 6.90-7.40 (aromatics); 1.53 (s, 15H, Cp*). Selected 13C NMR
(C6D6, 25 °C): δ 179.1 (Zr-O-C); 126.3 (Cp*); 11.0 (Me, Cp*).
Cp*Zr(OC6H-2,3,5,6-Ph4)Me2 (24). A diethyl ether solution of
Cp*Zr(OC6H-2,3,5,6-Ph4)Cl2 (0.3 g, 0.43 mmol) was cooled to 0
°C in an ice bath. To this solution was added MeLi · LiBr (0.57
mL, 0.86 mmol) via syringe under a flush of nitrogen. The solution
was stirred at 0 °C in ice bath for an additional 3 h, and the solvent
was removed under vacuum and replaced with benzene. The
suspension was filtered through a plug of Celite over fritted glass
to remove lithium salts. The filtrate was evacuated to afford a white
1
H, 6.50. H NMR (C6D6, 25 °C): δ 6.90-7.40 (aromatics); 1.53
(s, 15H, C5Me5); -0.13 (s, 6H, Zr-Me). Selected 13C NMR (C6D6,
25 °C): δ 158.1 (Zr-O-C); 120.0 (C5Me5); 42.7 (Zr-Me), 11.1
(C5Me5).
Cp*Ti(OC6H3-2,6-(OMe)2)Me2 (30). LiMe (3.24 mL, 1.6 M
solution in diethyl ether, 5.18 mmol) was added dropwise to a
suspension of Cp*TiCl3 (0.500 g, 1.73 mmol) in 30 mL of Et2O at
0 °C. The mixture was stirred at 0 °C for 1 h and at room
temperature for 2 h. The mixture was then cooled back to 0 °C,
and a solution of 2,6-dimethoxyphenol (0.266 g, 1.73 mmol) in 10
mL of Et2O was added dropwise. The mixture was slowly warmed
to room temperature and stirred overnight. The solvent was removed
under vacuum, and benzene was added to the solid residue. The
suspension was filtered through a plug of Celite over fritted glass
to remove lithium salts. The filtrate was evacuated to dryness,
yielding a yellow liquid (0.52 g, 82%). Anal. Calcd for C20H30O3Ti:
1
C, 65.62; H, 8.19. Found: C, 65.45; H, 8.02. H NMR (C6D6, 25
°C): δ 7.76 (t, J ) 8.1 Hz, 1H, p-H); 6.43 (d, J ) 8.4 Hz, 2H,
m-H); 3.42 (s, 6H, OMe); 1.86 (s, 15H, Cp*); 0.89 (s, 6H, Ti-Me).

5508 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
13C NMR (C6D6, 25 °C): δ 152.0 (Ti-O-C); 145.1, 119.8, 106.0
(aromatics); 122.6 (C5Me5); 55.9 (OMe); 55.0 (Ti-Me); 11.7
(C5Me5).
Cp*Ti(OC6H2-2,6-iPr2-4-Br)Me2 (31). LiMe (3.24 mL, 1.6 M
solution in diethyl ether, 5.18 mmol) was added dropwise to a
suspension of Cp*TiCl3 (0.500 g, 1.73 mmol) in 30 mL of Et2O at
0 °C. The mixture was stirred at 0 °C for 1 h and at room
temperature for 2 h. The mixture was then cooled back to 0 °C,
and a solution of 4-bromo-2,6-diisopropylphenol (0.445 g, 1.73
mmol) in 10 mL of Et2O was added dropwise. The mixture was
slowly warmed to room temperature and was stirred overnight. The
solvent was removed under vacuum, and benzene was added to
the solid residue. The suspension was filtered through a plug of
Celite over fritted glass to remove lithium salts. The filtrate was
evacuated to dryness, yielding an orange solid (0.75 g, 92%). X-ray-
quality crystals were obtained by placing a concentrated hexanes
solution in a freezer. Anal. Calcd for C24H37BrOTi: C, 61.46; H,
7.89. Found: C, 61.26; H, 8.02. 1H NMR (C6D6, 25 °C): δ 7.40 (s,
2H, m-H); 3.06 (m, J ) 6.9 Hz, 2H, CHMe2); 1.68 (s, 15H, Cp*);
1.14 (d, J ) 6.9 Hz, 12H, CHMe2); 0.70 (s, 6H, Ti-Me). 13C NMR
(C6D6, 25 °C): δ 158.1 (Ti-O-C); 141.3, 127.0, 115.1 (aromatics);
123.0 (C5Me5); 55.4 (Ti-Me); 27.5 (CHMe2); 23.9 (CHMe2); 11.8
(C5Me5).
Figure 2. Effect of basis set on computed EIPS values for catalysts
1a-18a. The solid line is y ) x.
1-Hexene Polymerization Reactions. 1-Hexene polymerization
reactions were followed by measuring monomer consumption versus
time using 1H NMR or by weighing dried polymer samples at
regular intervals of time during the polymerization reaction to
compute yields. The polymer samples were purified by extracting
into THF and passing the extracts through a short plug of alumina
to remove the catalyst. Molecular weight distributions were
measured using a Waters Alliance GPCV 2000. Experimental details
and data analysis are given in the Supporting Information.
A representative procedure is provided here for the 200:1
monomer-catalyst trial at 0 °C. A solution of precatalyst 13 in
toluene-d8 (25.0 µL, 0.10 M, 2.5 µmol) was added to a mixture of
1-hexene (62.5 µL, 0.500 mmol) and internal standard Ph2CH2 (10
µL) in 0.39 mL of toluene-d8 in a screw-cap NMR tube with a
PTFE/silicone septum. This solution was cooled to 0 °C using a
cooler attached to the NMR spectrometer, and an NMR spectrum
was taken. A solution of B(C6F5)3 activator in toluene-d8 (25.0 µL,
0.10 M, 2.5 µmol) was then added through the septum at 0 °C.
The NMR tube was shaken vigorously and quickly placed back
into the spectrometer. The conversion of 1-hexene was monitored
by integrating the proton signal in the olefinic region versus the
internal standard Ph2CH2. After the polymerization proceeded to
over 95% completion, the solution was poured into excess methanol
to precipitate the polymer. The excess methanol was decanted, and
the polymer was dried under vacuum overnight.
Figure 3. Effect of exchange-correlation functional on computed
EIPS values for catalysts 1a-18a. The solid line is y ) x + 10.
optimization is too computationally expensive. In order to assess
the validity of this approach, PCM-geometry optimization was
performed for catalysts 1a and 3a in a toluene-like solvent, which
gave EIPS values within 0.6 kcal/mol of the vacuum-optimized
geometries. The computational approach used here was found to
give EIPS values that correlate well to experimental reactivity
trends.11
Computational Methods. To compute ion pair separation
energies (EIPS), the geometries of the contact ion pair, bare cation,
and bare counterion were first optimized under vacuum using
Gaussian 03. The GDIIS method was used to optimize ground-
state geometries to better than 0.005 Å for the atom displacements,
0.0025 au for the forces, and 0.02 kcal/mol for the total energies.
Several initial conditions were explored whenever there was any
doubt about which catalyst conformation had the lowest energy.
Once the geometries converged, the lowest energy conformation
was utilized.
6-311++G** basis sets were used on all atoms, except for I
and Zr, which used the Stuttgart/Dresden (SDD) triple-ꢀ basis set
with effective core potentials. This basis set was sufficiently large
that basis set superposition error and size limitations could be
neglected. The counterpoise correction (which is widely used but
known to usually overestimate BSSE) was found to be ap-
proximately 2 kcal/mol for the vacuum EIPS value of catalyst 1a
using the OLYP/6-311++G** method. As shown in Figure 2,
for catalysts 1a-18a the 6-311++G** EIPS values for vacuum or
toluene were 1 (average) ( 1 (standard deviation) kcal/mol higher
than the LANL2DZ values, irrespective of the exchange-correlation
functional. These results show that basis set size limitations were
not appreciable in the calculation of EIPS. The OLYP exchange-
correlation functional was used because it afforded reliable
performance at reasonable computational cost. Comparisons were
made to the more computationally expensive B3LYP functional
for catalysts 1a-18a in a series of solvents. As shown in Figure 3,
the B3LYP EIPS values were 10 (average) ( 2 (standard deviation)
kcal/mol higher than the OLYP values irrespective of the solvent.
Once the vacuum geometries were determined, single-point
calculations were performed using the polarization continuum model
(PCM) to obtain the self-consistent field (SCF) energies in solution.
PCM calculations were performed in Gaussian 03 using the default
solute cavity generation parameters (UA0 radii).39 Vacuum geom-
etry optimization was used, because solvent-phase geometry
(39) Frisch, M. J., et al. Gaussian 03, Revision C.02; Gaussian, Inc.,
Wallingford, CT, 2004.

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5509
Figure 4. Process of ion pair separation.
These results show EIPS values computed using different exchange-
correlation functionals and/or basis sets are linearly correlated to
each other.
Figure 5. SCF EIPS values in toluene (KS ) 2.38), bromobenzene
(KS ) 5.40), and 1,2-dichlorobenzene (KS ) 9.93) for the 29
Cp′Ti(OAr)Me2/B(C6F5)3 catalysts in Table 1.
Results of Experimental and DFT Studies
Ion Pair Separation Energies and Ligand Cone Angles.
Figure 4 illustrates the process of ion pair separation. In the
ISIP, also called the contact ion pair, the cation and counterion
are in direct contact with no solvent in between. Inside the
solvent-accessible surface the dielectric constant felt by the ion
pair is equal to 1, whereas outside the solvent-accessible surface
the dielectric constant felt by the ion pair is equal to the solvent
dielectric constant KS. The charge separation distance between
the cation and counterion is denoted as D, where D0 is the charge
separation distance in the contact ion pair. If the counterion is
displaced by a very small amount, i.e. by D0 < D < Dc, there
is not enough room for a solvent molecule between the cation
and counterion. At some critical charge separation distance, Dc,
the solvent-accessible surfaces of the cation and counterion touch
at a single point. For D > Dc, the solvent-accessible surfaces
of the cation and counterion are separated by a layer of solvent.
An ion pair separation energy descriptor is needed that
measures the intrinsic difficulty of separating the cation from
the anion while allowing the cation and anion geometries to
relax. In order to have the most utility, the ion pair separation
energy descriptor should scale in predictable ways with KS and
D. Using electrostatic arguments to derive scaling relationships,
a reliable prediction will be made (see eq 3) of how the
electronic ion pair separation energy of the system changes with
KS. The “thermodynamic” energies (vibrational, translational,
rotational) do not follow this scaling relationship; thus, it is
important for the EIPS descriptor to exclude them. The electronic
ion pair separation energy, EIPS, is defined as the SCF energy
of the isolated cation plus the SCF energy of the isolated
counterion minus the SCF energy of the contact ion pairsall
computed in the dielectric medium of interest.
between the maximum at KS ) 1, as determined from the
vacuum energy and the minimum as KS approaches infinity:
specifically
C ) EIPS[vacuum] - B
(2)
From Coulomb’s law, the solvent-dependent EIPS portion must
be inversely proportional to KS; thus
1
KS
EIPS[solvent] ) B + C ⁄ KS ) C
- 1 + EIPS[vacuum]
(
)
(3)
where each catalyst has its own B and C values. Once the
constants B and C are determined, the EIPS value for that catalyst
in different organic solvents can be estimated.
Activation with B(C6F5)3 was studied for a series of 31
catalyst precursors (Table 1). Activation with Ph3CB(C6F5)4 was
studied for 5 of these catalyst precursors. In each case, the PCM
model was used to compute EIPS values in vacuum, toluene (KS
) 2.38), bromobenzene (KS ) 5.40), and 1,2-dichlorobenzene
(KS ) 9.93). Results for the 29 Cp′Ti(OAr)Me2/B(C6F5)3
systems are plotted in Figure 5. For a given catalyst, the value
of EIPS decreases as KS increases. For a given metal and
counterion, the value of EIPS in a particular solvent decreases
as EIPS in vacuum decreases. A least-squares fit of eq 3 to these
EIPS values was performed to obtain the optimized B and C
parameters for each catalyst conformation (Table 1). The rms
error between eq 3 and the DFT-computed EIPS values was less
than 0.5 kcal/mol for each of the catalysts, indicating that eq 3
is the proper functional form.
There are several factors that influence the values of B and
C. For a given catalyst precursor, the EIPS, B, and C values are
slightly smaller for [B(C6F5)4]- than for [MeB(C6F5)3]-, reflect-
ing the fact that [B(C6F5)4]- is more weakly bound than
[MeB(C6F5)3]-. This decrease in EIPS is partially due to the
slightly larger size of [B(C6F5)4]-, which leads to a larger initial
separation distance. For example, the Ti-B distance is 6.65 Å
in 12b compared to 4.19 Å in 12a. The B and C values tend to
decrease as the electron-donating ability of the ligand increases.
Specifically, the electron-donating abilities of the cyclopenta-
dienyl ligands follow the order Cp* > Ind > Cp, and for a
given aryloxide ligand the B and C values follow the order Cp
> Ind > Cp*. In a similar manner, for the aryloxide ligand
both B and C decrease with the addition of electron-donating
substituents such as Me, Et, iPr, etc. On the other hand, addition
What type of scaling relationship should be followed by EIPS
as the solvent dielectric constant KS is changed? When no
solvent is present, KS ) 1 and EIPS is at its maximum value,
and as KS approaches infinity, EIPS will be at its mimimum value.
Because solvent does not penetrate between the cation and
counterion in the contact ion pair, some energy B is required to
separate them even as KS approaches infinity:
lim
8 EIPS ) B
S f ∞
(1)
K
In general, EIPS has two parts: (a) the part that is screened by
solvent and (b) the part that is not screened by solvent. The
part not screened by solvent is equal to B, since no matter how
much we increase KS the energy B remains constant. The
fraction of EIPS screened by solvent is equal to the difference

5510 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
Table 2. Effects of Counterion on Activity for [Cp*Ti(OC6H3-2,6-R2)Me]+[A]-a
[B(C6F5)4]- to [MeB(C6F5)3]- activity
ratio
R
[A]-
EIPS(toluene) (kcal/mol) monomer ratio tquench (min) yield (%) activity (g/(mmol h))
exptl
predictionb
adjustedc
Me [MeB(C6F5)3]-
28.9
23.9
23.9
28.4
22.7
22.7
26.6
22.3
22.3
2000:1
2000:1
4000:1
2000:1
2000:1
4000:1
2000:1
2000:1
4000:1
20
10
3
20
10
3
20
10
3
55
100
100
14
100
52
3.7
40
8
280
g24
16
35
[B(C6F5)4]-
high
g6720
71.4
high
3500
18.6
404
[B(C6F5)4]-
Et [MeB(C6F5)3]-
[B(C6F5)4]-
49
23
11
51
24
[B(C6F5)4]-
iPr [MeB(C6F5)3]-
[B(C6F5)4]-
22-30
[B(C6F5)4]-
552
a Reactions at 25 °C with 2.0 M 1-hexene in toluene solvent. b Prediction based upon change in EIPS assuming A, R, and f are the same for both
counterions. c Based upon change in EIPS assuming R and f are the same for both counterions and A is 2.2 times larger for [B(C6F5)4]- than for
[MeB(C6F5)3]-.
of electron-withdrawing groups such as F, Cl, Br, and I in the
para position increases B and C. As reflected by the B and C
values, for a given ligand structure it is much harder to separate
the [MeB(C6F5)3]- counterion from a Zr catalyst than from the
corresponding Ti catalyst. Surprisingly, some of the catalysts
have negative B values, which implies that the global ground
state of these catalysts as KS approaches infinity is an SSIP.
This unusual behavior is due to opportunistic coordination of
the ligand upon ion pair separation, as explained in the next
section.
Also shown in Table 1 are the computed ligand cone angles
for each of the catalyst systems. The ligand cone angle, θlig,
was defined as the largest X-m-Y angle, where m is the center
of the metal atom and X and Y are two points on the van der
Waals surface of the ligand outside the metal’s van der Waals
sphere. θlig lies between the angle θmax of a circular cone (with
the apex at the metal center) that completely encompasses the
ligand’s van der Waals surface and the angle θmin ) 2 acos[1
- Ω/(2π)] of a circular cone that subtends a solid angle equal
to the solid angle Ω subtended by the ligand’s van der Waals
surface:40
Zr is slightly less sterically hindering than the same ligand on
Ti, because the larger diameter of Zr causes the ligand to be
located slightly farther from the metal center, which in turn
decreases the ligand cone angle. In summary, the degree of steric
hindrance at the metal center is primarily determined by (i) the
type of cyclopentadienyl ring (i.e., Cp, Ind, or Cp*), (ii)
the number of ortho substituents on the aryloxide ring, and (iii)
the metal. The growing polymer chain also contributes to steric
hindrance at the metal center.
Table 2 compares the activity of catalysts 10-12 activated
with B(C6F5)3 and Ph3CB(C6F5)4. For each of these catalysts,
the activity for the [B(C6F5)4]- counterion was higher than for
the [MeB(C6F5)3]- counterion. The increase in activity can be
attributed to the lower EIPS value for [B(C6F5)4]- compared to
that for the [MeB(C6F5)3]- counterion.
Opportunistic Coordination of Some Ortho Substituents
to the Metal Center. On examination of the data in Table 1,
some catalysts have comparatively low B values, which we
postulate is caused by opportunistic coordination of a ligand
substituent to the metal upon ion separation. The formation of
this partial bond between the metal and ligand substituent in
the bare cation lowers the energy required to separate the ion
pair. The parameter B becomes negative when the energy gain
from the formation of this partial bond exceeds the energy
required to break the cation-counterion bond as KS approaches
infinity. To quantify these effects, selected geometric parameters
and bond orders were computed for each catalyst.
For all of the contact ion pairs in Table 3, the Ti-O-Ar
bond angle is nearly straight (163.7-178.0°). For the cations,
about half of the structures (group 1) have a Ti-O-Ar bond
angle in the range of 172.6-177.7°, while the other half (group
2, boldface) lie in the range of 122.8-151.3°. The bending of
the Ti-O-Ar angle in the cations of group 2 allows the ortho
substituent to move closer to and form a partial bond with the
metal. Figure 6 shows the geometries of the proximal and distal
cations and ion pairs for catalyst 20a. One can clearly see that
the Ti-O-Ar bond angle is nearly straight in the contact ion
pairs but substantially bent in the bare cations.
θmin e θlig e θmax
(4)
The predicted solid angle (Ωlig ) 4π sin2(θlig/4)) blocked by
the ligand lies between the actual solid angle Ω subtended by the
ligand’s van der Waals surface and the solid angle subtended
by the circular cone of angle θmax. Because the van der Waals
surface of the ligand may contain inaccessible crevices, one
expects the true measure of solid angle blocked by the ligand
to satisfy the same bounds.
The cone angles reveal that the Ind ligand (136°) is more
sterically hindering than the Cp ligand (∼129°) but less sterically
hindering than the Cp* ligand (∼155°). There is a very large
difference in steric hindrance between catalysts with zero, one,
or two ortho substituents on the aryloxide ligand. For the Ti
catalysts, we note that θOAr varies from 95 to 96° for no ortho
substituents, from 105 to 128° for one ortho substituent, and
from 127 to 188° for two ortho substituents. Also, a ligand on
The Wiberg bond index array was computed using natural
bond orbital (NBO) analysis in the Gaussian 03 program. The
Wiberg bond index array is a measure of the effective bond
order between atoms in a molecular structure.41 The largest
partial bond orders between any ortho substitutent group atom
and the metal center are given in the last two columns of Table
3. For all of the contact ion pairs and the cations in group 1 the
effective bond order was 0, implying there was no bonding
between the ortho substituent group and the metal center.
(40) Consider the solid angle Ω of the shadow cast when a point source
of light is placed at the metal center m by the portion of the ligand’s van
der Waals surface lying outside the metal’s van der Waals radius onto a
sphere surrounding the ligand. Among all surfaces that subtend a solid angle
Ω, the circular cone has the smallest maximum angle X-m-Y (aka θmin),
where X and Y are any two points on the surface; therefore, the maximum
X-m-Y angle of the ligand’s van der Waals surface (aka θlig) must be
greater than or equal to θmin. The maximum angle X-m-Y of the ligand’s
van der Waals surface (i.e. θlig) lies on the surface of the circular cone
(with apex at the metal center) completely enclosing the ligand’s van der
Waals surface; therefore, the angle θmax across this cone must necessarily
be greater than or equal to θlig
.
(41) Wiberg, K. B. Tetrahedron 1968, 24, 1083–1096.

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5511
Table 3. Properties of Selected [Cp*Ti(OAr)Me]+[MeB(C6F5)3]-a
Ti-O-Ar (deg)
Ti-X distb (Å)
Ti-X Wiberg bond indexb,c
OAr substituent
EIPS(toluene) (kcal/mol)
A (kcal/mol)
B (kcal/mol)
cation
IP
cation
IP
cation
IP
none (X ) H)
31.42
29.92d
29.63p
32.08
60.32
57.24d
58.14p
60.63
57.24d
58.14p
60.60
57.88d
57.72p
61.40
57.06d
57.71p
61.01
58.20d
57.59p
59.32
57.48d
58.92p
56.24
56.16
5.80
175.6
125.8d
126.0p
176.4
135.3d
135.3p
176.6
138.5d
138.3p
177.3
144.3d
143.7p
175.0
148.9d
146.5p
177.7
174.6d
172.6p
151.3
122.8
174.4
175.8d
172.7p
174.6
170.9d
170.2p
173.7
170.0d
167.7p
173.6
166.8d
166.6p
174.7
163.7d
164.4p
175.1
164.2d
166.2p
178.0
173.4
3.94
3.94
0.0
0.0
2-F
5.95d
5.21p
6.51
2.32d
2.33p
7.26
3.96d
3.99p
7.28
0.2d
0.2p
0.0
0.0d
0.0p
0.0
4-F
2-Cl
26.03d
26.09p
32.09
5.95d
5.21p
6.50
2.67d
2.67p
7.69
4.27d
4.26p
7.67
0.4d
0.4p
0.0
0.0d
0.0p
0.0
4-Cl
2-Br
23.63d
24.19p
31.80
-0.93d
-0.03p
5.75
2.78d
2.78p
7.84
4.63d
4.41p
7.85
0.5d
0.5p
0.0
0.0d
0.0p
0.0
4-Br
2-I
20.26d
21.04p
29.79
-3.70d
-3.20p
3.79
3.00d
2.99p
8.07
4.57d
4.56p
8.08
0.6d
0.6p
0.0
0.0d
0.0p
0.0
4-I
2-Ph
25.36d
26.02p
31.20
1.00d
1.92p
6.03
2.62d
2.50p
7.46
4.17d
4.31p
7.46
0.2d
0.2p
0.0
0.0d
0.0p
0.0
4-Ph
2-cyclohexyl
30.17d
30.91p
14.06
6.06d
6.02p
-9.65
-2.43
3.21d
3.34p
2.67
3.61d
3.50p
4.07
0.0d
0.0p
0.2
0.0d
0.0p
0.0
2,3,5,6-Ph4
2,6-(OMe)2
21.26
2.22
3.83
0.2
0.0
a Legend: d, aryloxide substituent distal to initiating group; p, aryloxide substituent proximal to initiating group. b The smallest Ti-X distance and
largest Ti-X bond index for any atom in the OAr substituent. c In toluene-like solvent computed using natural bond orbital analysis.
or growing chain, and (iv) the counterion. When the counterion
is displaced, this creates an empty docking site at the metal
center. The metal center seeks to fill this docking site with
monomer or other ligands to regain its four-coordinate environ-
ment. It can partially fill this empty coordination site by forming
a partial bond to an ortho ligand substituent capable of donating
electron pairs. Ortho substituents that have electron pairs
available for coordination to the metal include halogens,
aromatic rings, and methoxy groups. Ortho substituents that do
Figure 6. [Cp*Ti(OC6H4-2-Br)Me]+ geometry in the (a) distal
contact ion pair, (b) distal bare cation, (c) proximal contact ion
pair, and (d) proximal bare cation. For clarity, the [MeB(C6F5)3]-
counterion is not shown in the ion pairs.
not have any electron pairs available for coordination to the
metal include H and aliphatic groups (e.g., Me, Et, Pr).
i
Determination of Chain Initiation and Propagation
Rate Constants. Table 4 gives an overview of the experimental
1-hexene polymerization data. The experimental conditions,
polymer molecular weight, polydispersity index (PDI) defined
as Mw/Mn, kp, and ki/kp are given. The PDI values are between
1 and 3, which is consistent with a single-site behavior. For
polymerizations in toluene or bromobenzene solvent, the
monomer versus time concentration profiles were fit to a simple
kinetic model containing chain initiation, propagation, and
catalyst deactivation steps to determine the rate constants:
However, for all the cations in group 2 there was clear evidence
for the formation of a partial bond between the ortho substituent
and the metal center, where the computed effective bond order
ranged from 0.2 to 0.6.
The Ti-X distance was taken to be the distance between the
Ti center and the closest ortho substituent group atom. For
catalysts in group 1, the Ti-X distance was similar for the cation
and ion pair. For catalysts in group 2, the Ti-X distance was
substantially shorter in the cation than in the ion pair, due to
the formation of a partial Ti-X bond in the cation.
In addition to the catalysts 16a, 20a, 28a, and 30a discussed
above, Table 1 shows that the Cp (19a,b), Ind (22a), and Zr
(24a) analogues of 16a also have negative B values. For these
complexes, the bare cations have m-O-Ar angles smaller than
150° and display partial coordination between the ortho sub-
stituent and metal center.42 These results show that opportunistic
ligand coordination occurs irrespective of the type of metal (Ti
or Zr), counterion, and ancillary Cp′ ligand (Cp, Ind, or Cp*).
There is a simple chemical explanation for the occurrence of
opportunistic ligand bonding in catalysts 16, 19-22, 24, 26-28,
and 30 activated with either counterion. The positively charged
metal prefers a four-coordinate pseudotetrahedral environment.
In the ion pair, these four sites are filled by (i) the cyclopen-
tadienyl ligand, (ii) the aryloxide ligand, (iii) the initiating group
ki
initiation: C + M
9
8 CM
(5)
(6)
kp
propagation: CMN + M
98 CMN+1 for Ng1
kd
kd
deactivation: CMN + M f DMN+1 or CMN f
DMN for Ng1 (7)
The fitted kinetic profiles and rate constants for catalysts 1a-18a
were previously reported.11 The same procedure was also used
to determine rate constants for catalysts 19a-24a, 30a, and 31a
from the monomer concentration versus time data and is
described in the Supporting Information. (Precatalysts 25-29
were not synthesized.) The catalyst deactivation step was ignored
if catalyst deactivation was negligible. In cases where catalyst
deactivation was appreciable, the deactivation step was made
either first or zero order in monomer concentration according
to what was needed to fit the monomer concentration versus
(42) The m-O-Ar angles, m-X Wiberg bond indexes, and m-X
distances for the cations and ion pairs of all catalysts are presented in the
Supporting Information.

5512 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
Table 4. Experimental 1-Hexene Polymerization Data for [Cp′m(OAr)Me]+[MeB(C6F5)3]- a
b
c
precat. no.
m
Cp′ OAr substituent θCp′ + θOAr (deg)
S
T (°C) [M]0 (M) [M]0:[C]0 kp (M-1 s-1)
ki/kp
init. Mn (kDa) PDI
1
2
3
4
5
6
7
8
9
10
11
12
13
Ti Cp 2,6-Me2-4-Br
Ti Cp 2,6-Et2
Ti Cp 2,6-iPr2
Ti Cp* none
Ti Cp* 4-F
Ti Cp* 4-Cl
257
261
265
251
251
252
251
251
251
282
286
290
280d
282p
282
268d
267p
317
252d
251p
251
297
272d
272p
280d
282p
305
274
Tol
Tol
Tol
Tol
Tol
Tol
Tol
Tol
Tol
Tol
Tol
Tol
Tol
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
1
1
1
200
100
200
200
200
200
200
200
200
200
200
200
200
0.16-0.22
0.25-0.34
0.42
0.42
0.27
0.24
0.26
0.38
0.48
g0.02
g0.03
g0.02
g0.02
g0.02
g0.02
g0.02
g0.02
g0.02
0.0014
0.0003
0.0001
g0.02
facile
facile
facile
facile
facile
facile
facile
facile
facile
slow
3.2
5.2
6.2
4.4
4.7
5.3
5.2
4.7
5.1
28.6
44.8
62.4
17.3
h
1.4
1.3
1.9
1.9
2.0
2.2
1.9
h
1.1
1.2
1.5
1.5
Ti Cp* 4-Br
Ti Cp* 4-Ph
Ti Cp* 4-tBu
Ti Cp* 2,6-Me2
Ti Cp* 2,6-Et2
Ti Cp* 2,6-iPr2
Ti Cp* 2-cyclohexyl
0.51
0.56-0.78
0.60-0.92
0.74
slow
slow
facile
14
15
Ti Cp* 2,6-Me2-4-Br
Ti Cp* 2-CH2Ph
Tol
Tol
0
0
1
1
200
200
0.28
0.27
0.0014
g0.02
slow
facile
27.9
14.3
1.6
1.4
16
17
Ti Cp* 2,3,5,6-Ph4
Ti Cp* 3-OMe
Tol
Tol
0
0
1
1
200
200
1.36
0.25
g0.02
g0.02
facile
facile
9.5
3.1
h
1.7
18
19
20
Ti Cp* 4-OMe
Ti Cp 2,3,5,6-Ph4
Ti Cp* 2-Br
Tol
Tol
Tol
0
0
0
1
1
1
200
100
500
0.20
2.2
5.0-6.1
g0.02
g0.03
g0.006
facile
facile
facile
5.4
9.7
32.4
1.9
1.5
1.4
21
Ti Cp* 2-Ph
Tol
0
1
200
3.1
g0.02
facile
14.7
1.3
22
23
Ti Ind 2,3,5,6-Ph4
Ti Ind 2,6-iPr2
Tol
Tol
Tol
Tol
BrB
BrB
DCB
Tol
Tol
25
0
1
1
1
1
0.5
1
1
1
1
100
100
100
100
100f
200f
100f
200
200
g3e
2.4
0.6
negl.
0.045f
0.045f
0.73f
0.05
0.55
g0.01
g0.03
g0.03
facile
facile
facile
7.8
4.7
4.2
1.4
1.7
1.6
-20
25
25
25
25
0
24
Zr Cp* 2,3,5,6-Ph4
304
g0.03
g0.02
g0.005
g0.02
facile
facile
g
11.8
17.6
13.2
7.2
1.1
1.1
1.1
2.9
1.8
30
31
Ti Cp* 2,6-(OMe)2
Ti Cp* 2,6-iPr2-4-Br
290
289
facile
0
9 × 10-5 slow
44.9
a Legend: d, aryloxide substituent distal to initiating group; p, aryloxide substituent proximal to initiating group. b kp values for catalysts 1-18 were
taken from ref 11. c ki values for catalysts 10-12 and 14 were taken from the Supporting Information of ref 11. d Distal conformation of catalyst. e All
monomer consumed within 2 min. f The Mn values with extremely narrow molecular weight distributions indicate 65-70% of the metal sites are active
in these experiments, giving M:C ) 140-150 (for nominal 100:1) and 280 (for nominal 200:1) and kp ) 0.063 (bromobenzene) and 1.1
(1,2-dichlorobenzene) on a per active site basis. g Undetermined because the kinetic profile shows an induction period but this induction period does not
have the same form as slow initiation kinetics first order in monomer. h Experimental data not available.
time profile. Chain initiation was considered facile if there was
no visible induction period at the beginning of the log monomer
concentration versus time plot, and in this case only a lower
bound can be placed on the value of ki. If there was a visible
induction period, chain initiation was considered slow and
separate ki and kp values were determined via simultaneous
optimization to fit the monomer concentration versus time profile
and the Mn value.
For the moment, consider the case of a catalyst that does not
deactivate and has chain initiation and propagation that are both
first-order in monomer as given in eqs 5 and 6. The rate
expression for monomer consumption is given by
d[M]
dt
d[C]
dt
) -ki[M][C] - kp([C]0 - [C])[M] where
) -
Figure 7. Dimensionless kinetic profiles for a batch polymerization
reaction with first order in monomer chain initiation and propagation
in the absence of catalyst deactivation.
ki[M][C] (8)
which can be written in the dimensionless form
negligible whenever [C]0/[M]0 e 0.01, the numerical solution
to eq 9 was found to be independent of [C]0/[M]0 whenever
[C]0/[M]0 e 0.01. Equation 9 was solved numerically and is
shown in Figure 7.
The parameter ꢀ that defines the character of the kinetics
depends upon both the ki to kp ratio and the initial concentration
of monomer to catalyst, i.e. [M]0/[C]0. Examining Figure 7, one
can see that when ꢀ > 3 the induction period at the beginning
of the reaction is negligible and chain initiation is facile. On
j
where τ ) kp[C]0t, ꢀ ) (ki/kp)([M]0/[C]0), M ) [M]/[M]0, and
j
C ) [C]/[C]0. The term in eq 9 containing the ratio [C]0/[M]0
accounts for monomer consumption due to chain initiation. Since
the total amount of monomer consumed by chain initiation
compared to chain propagation during the batch reaction is

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5513
Figure 8. Different forms of π-complexes and ion pairs for catalyst 24a: (red) metal; (green) growing chain; (lavender) counterion; (orange)
ligands; (yellow) monomer; (brown) solvent.
Table 5. Effect of Solvent on
will be 1000 mers, assuming the absence of chain transfer and
termination. Thus, slow chain initiation can lead to longer
polymer chains. For 100:1, 200:1, and 500:1 monomer to
catalyst ratios, living polymerization of 1-hexene with facile
initiation corresponds to 8.4, 16.8, and 42 kDa, respectively.
Table 4 shows that in every case where chain initiation is slow
the polymer Mn was more than 1.5 times these values. On the
other hand, for every case with facile chain initiation, the
polymer Mn was found to be either much less than or
approximately equal to these values. The experimentally
measured Mn values confirm the fact that only a few of the sites
are active when chain initiation is slow. The cutoff for slow
versus facile chain initiation does not occur at a ki/kp ratio of 1;
rather, it occurs at the value ꢀ ) 3, which corresponds to a
ki/kp ratio of 3[C]0/[M]0. For a 200:1 monomer:catalyst ratio,
the cutoff for facile versus slow initiation occurs at ki/kp ) 0.015.
[Cp*Zr(OC6H-2,3,5,6-Ph4)CH2CH2CH3]+[MeB(C6F5)3]- Ethylene
π-Complex and Product Formation Energies
rel SCF energy (kcal/mol)
1,2-
Zr-B (Å) vacuum toluene bromobenzene dichlorobenzene
Reactants
ISIP + ethylene 4.32
0
0
0
0
π-Complexes
FS π-complex
BS π-complex
SS π-complex
7.91
9.09
∼13
8.7
14.4
n.a.
4.4
6.9
11.2
2.2
3.6
4.6
1.2
2.0
2.0
Products
ISIP
4.32
-17.7 -17.5
-10.5 -17.1
-16.7
-19.8
-4.9
-16.5
-21.1
n.a.
ꢁ-agostic OSIP 7.55
SCAP
8.65 (Tol) n.a.
8.11 (BrB)
ꢁ-agostic SSIP ∼13
-4.4
n.a.
-12.8
-19.4
-22.0
The catalysts can be divided into the following basic
categories, according to whether chain initiation and chain
transfer/termination are facile or slow.
the other hand, when ꢀ < 3 the induction period at the beginning
of the batch reaction is appreciable and chain initiation is slow.
In this case, some of the catalyst sites remain uninitiated even
as the reaction proceeds to completion. For example, when ꢀ
) 1, 0.1, and 0.01 the fractions of sites initiated as t approaches
infinity, s∞, are 0.84, 0.38, and 0.13, respectively. When ꢀ <
0.1, more than 62% of the sites remain uninitiated throughout
the course of the batch reaction.
When a significant number of the sites remain uninitiated
when the monomer is completely consumed and when chain
transfer/termination steps are negligible, the final number of
mers in a chain will exceed the initial monomer to catalyst ratio.
For example, if only 10% of the catalyst sites are initiated
throughout the course of a batch reaction having a 100:1 initial
monomer to catalyst ratio, the average chain length produced
(1) Chain initiation is facile and chain transfer/termination
is facile. Facile chain transfer/termination causes the Mn value
to be substantially below that expected for a living polymeri-
zation. This category included catalysts 1a-9a, 17a, 18a, and
23a, for which Mn values were 6 kDa or less. These are the
most sterically open catalysts, with θCp′ + θOAr ranging from
251 to 274°. This category contained (a) all of the Cp* catalysts
with two trivial ortho substituents (i.e., 2,6-H2) and (b) all of
the Cp and Ind catalysts except for those containing the largest
ArO ligand (i.e., 2,3,5,6-Ph4).
(2) Chain initiation is facile, chain transfer/termination is slow,
and all of the metal sites are active. This category contained

5514 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
Figure 9. Effects of solvent on the reaction rate and order: monomer concentration versus time profiles for 1-hexene polymerization catalyzed
by [Cp*Zr(OC6H-2,3,5,6-Ph4)Me]+[MeB(C6F5)3]- in bromobenzene and 1,2-dichlorobenzene at room temperature. Solid lines are the kinetic
model fits described in the text.
Scheme 2. Chain Initiation and Propagation Reactions
catalysts for which the Mn value was between 43 and 103% of
that which would be expected from a living polymerization.
Category 2 included catalysts 13a, 15a, 16a, 19a-22a, and 30a,
which have θCp′ + θOAr in the range of 267-317°. This category
contained the Ti-based catalysts with (a) Cp* plus one nontrivial
ortho aryloxide substituent, (b) Cp, Ind, or Cp* with the 2,3,5,6-
Ph4 substituted ligand, and (c) Cp* catalysts for which one or
both ortho ayloxide substituents can opportunistically coordinate
to the metal center upon ion pair separation.
(3) Chain initiation is slow and chain transfer/termination is
slow. This category included the catalysts 10a-12a, 14a, and
31a. For these catalysts, the Mn values ranged from 166 to 371%
of the value expected for a living polymerization. The total
ligand cone angles θCp′ + θOAr ranged from 282 to 290°. This
category contained all Cp* catalysts with two nontrivial ortho
substituents where neither of these substituents opportunistically
coordinates to the metal center.
the Zr-based catalyst 24a. For this catalyst, the Mn value was
approximately 140% of that expected for living polymerization.
Combined with the narrow molecular weight distribution (PDI
) 1.1), this suggests only 65-70% of the metal sites are active
for polymerization. The percentages of active sites were
approximately the same in bromobenzene and 1,2-dichloroben-
zene solvents. The reason only some of the metal sites are active
is not fully understood at this time.
Different Types of Ion Pair and π-Complex States. As
mentioned in the Introduction, ion pairs can exist as either ISIP,
OSIP, SCAP, or SSIP. There is competition between counterion,
monomer, solvent, ligand, and initiating group or growing
polymer chain for the docking site at the metal center. The
counterion occupies the docking site in an ISIP, while the solvent
occupies the docking site in an SCAP. In the outer-sphere and
solvent-separated states, monomer, ligand, solvent, initiating
group, or growing polymer chain occupies the docking site. The
ligand, initiating group, or growing polymer chain can poten-
tially occupy the docking site only if it has a chemical structure
(4) Chain initiation is facile, chain transfer/termination is slow,
but not all of the metal sites are active. This category contained

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5515
that allows it to opportunistically, fluxionally, or agostically bond
to the metal center upon partial ion pair separation.
Effects of Solvent on Reaction Order. Figure 9 shows
monomer concentration versus time profiles for catalyst 24a in
bromobenzene and 1,2-dichlorobenzene solvents at room tem-
perature. Kinetic data are not available for toluene, since the
reaction rate was negligible. For each solvent, normal and
semilog plots are shown. First order in monomer chain propaga-
tion with facile chain initiation and negligible catalyst deactiva-
tion leads to a straight line on the semilog plot, while a zero
order in monomer propagation with facile initiation and
negligible deactivation leads to a linear monomer concentration
versus time plot. In bromobenzene, the kinetic profile is linear
on a semilog plot; thus, the propagation is apparently first order
in monomer and initiation is facile. In 1,2-dichlorobenzene, the
chain propagation starts out as zero order in monomer, since
the linear monomer concentration versus time plot is observed,
but the kinetics then transitions to first order as monomer is
consumed.
A candidate mechanism for transition from zero- to first-
order monomer consumption during a batch reaction is
developed here. Consider a simple model of chain propaga-
tion in which the monomer reversibly coordinates to an ion
pair to form a π-complex followed by insertion of the
coordinated monomer to elongate the polymer chain (Scheme
2); specifically
The geometries (Figure 8) and relative energies (Table 5) of
different ion pair and π-complex states were computed for
catalyst 24a. For this catalyst, the OSIP and SSIP contained
opportunistic bonding from the aryloxide ligand and ꢁ-agostic
bonding from the growing polymer chain to the metal center.
The Zr-B distance increased from 4.32 Å in the ISIP to 7.55
Å in the OSIP and 8.11 Å in the bromobenzene SCAP. The
charge separation distance in the SSIP state was estimated as
approximately 13 Å by adding twice the solvent radius (2.82
Å) to the Zr-B distance in the OSIP state and rounding to the
nearest angstrom.
Geometry optimization of the SCAP states was started with
an initial guess of η6 coordination by a solvent molecule to the
metal center. In the case of 1,2-dichlorobenzene the solvent
molecule was displaced to the outer coordination sphere,
indicating that an SCAP state is not formed. In the case of
toluene and bromobenzene, the converged geometry contained
a solvent molecule in the inner coordination sphere with η1
coordination, which suggests η6 coordination of the solvent to
the metal is not favorable for 24a. In toluene and bromobenzene,
the SCAP states were 8.4-14.9 kcal/mol higher in energy than
the ISIP, OSIP, and SSIP. Because the SCAP is the least favored
of the four different ion pair forms, it follows that solvent is a
poor competitor for the docking site.
k2
98 IP
IP + M yk1z Π
(10)
k-1
The relative SCF energies of different ion pair forms were
found to be a strong function of solvent. Under vacuum, the
ISIP was much lower in energy than any of the other ion pairs.
In toluene, the ISIP and OSIP were approximately tied for the
lowest energy, while the SCAP and SSIP were substantially
higher in energy. In bromobenzene, the computed relative
energies of the ISIP, OSIP, and SSIP were within 3 kcal/mol
of each other, while the SCAP was much higher in energy. Since
the absolute accuracy of the DFT method is ca. (5 kcal/mol, it
is not clear which of these three ion pair forms is actually the
most stable in bromobenzene. In 1,2-dichlorobenzene, the SSIP
and OSIP were approximately tied for the lowest energy while
the ISIP was approximately 5 kcal/mol higher in energy. These
results clearly show the trend that ISIP and/or OSIP are preferred
in lower dielectric constant solvents while SSIP and/or OSIP
are preferred in higher dielectric constant solvents.
The coordination and insertion steps are repeated many times
to produce the polymer chain. A similar kinetic mechanism for
a steady-state reactor has been previously considered.43 Now
the batch reactor kinetics are developed. The terms IP and Π
in these equations refer to the kinetically most important ion
pair and π-complex states. Which particular ion pair (ISIP,
OSIP, SCAP, or SSIP) and π-complex (FS, BS, or SS) is the
kinetically most important depends upon the catalyst structure,
solvent, and other reaction conditions. The concentration of live
catalyst sites is given by
[Clive] ) [IP] + [Π]
(11)
Since the catalyst undergoes many turnovers during the course
of polymerization, we can utilize the pseudo-steady-state
approximation:
The counterion can occupy different positions in the π-com-
plexes: (a) the front-side (FS) π-complex has the monomer and
counterion on the same side of the cation and (b) the back-side
(BS) π-complex has the monomer and counterion on oppo-
site sides of the cation. In the FS and BS π-complexes, the
counterion is in the outer coordination sphere of the metal, while
the counterion is fully solvated in the solvent-separated (SS)
π-complex. In each solvent, the FS π-complex was lower in
energy than the SS and BS π-complexes. This is explained by
the smaller ion pair separation distance in the FS (Zr-B ) 7.91
Å, Table 5) versus the BS (9.09 Å) and SS (∼13 Å)
π-complexes. In general, the energy required to form a
π-complex is the sum of two competing effects: (1) the energy
cost associated with partial separation of the ion pair and (2)
the energy gain associated with the monomer forming a bond
to the metal center. As KS increases, the energy cost of partial
ion pair separation decreases, causing the π-complex formation
energy to also decrease. For example, the computed energy to
form the FS π-complex decreased from 8.7 kcal/mol under
vacuum to only 1.2 kcal/mol in 1,2-dichlorobenzene, while that
for the BS π-complex decreased from 14.1 kcal/mol under
vacuum to 2.0 kcal/mol in 1,2-dichlorobenzene.
d[Π]
0 ≈
) k1[IP][M] - (k-1 + k2)[Π]
(12)
dt
which combined with the site balance gives
k1[M][Clive (k-1 + k2)[Clive
k1[M] + k-1 + k2
]
]
[Π] )
and [IP] )
k1[M] + k-1 + k2
(13)
The propagation rate for a batch reaction is thus given by
k1k2[Clive][M]
d[M]
dt
-
) k2[Π] )
(14)
k1[M] + k-1 + k2
Equation 14 is integrated to give
k1[M] + k-1 + k2
[M]0
t dt ) t ) -
d[M] )
∫
∫
0
[M]
k1k2[Clive][M]
1
1
1
kp
[M]
[M]0
([M]0 - [M]) - ln
(15)
(
(
))
[Clive] k2
where the effective propagation rate constant is given by

5516 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
The activation energy Ea and pre-exponential factor k0 are
experimentally measurable quantities; thus, their values should
be independent of any of the computational methods employed
to generate the catalyst descriptors. Since the slope of the linear
correlation between B3LYP and OLYP EIPS values is ap-
proximately 1, irrespective of basis set, it follows from eq 19
the adjustable parameter R is the same for both functionals and
independent of basis set. From the average EIPS differences, it
follows from eq 19 that the value of the adjustable parameter
E0 is approximately R × 10 kcal/mol ) 3 kcal/mol higher when
using OLYP than when using B3LYP EIPS values. For each
catalyst, the computed ligand cone angles were virtually
unaffected by the choice of exchange-correlation functional and
basis set. For example, catalyst 1a gave θCp′ ) 128° (OLYP/
6-311++G**), 128° (OLYP/LANL2DZ), 129° (B3LYP/6-
311++G**), 128° (B3LYP/LANL2DZ) and θOAr ) 129°
(OLYP/6-311++G**), 129° (OLYP/LANL2DZ), 130° (B3LYP/
6-311++G**), 130° (B3LYP/LANL2DZ). Because the com-
puted ligand cone angles are independent of basis set and
functional choice, it follows from eqs 20 and 21 that the
adjustable parameters a0 and f are likewise unaffected by a
change in basis set or exchange-correlation functional.
k1k2
kp )
(16)
k-1 + k2
Equation 15 provides a good fit to the experimental data, as
shown in Figure 9, where the fitted parameters for 24a in 1,2-
dichlorobenzene are kp ) 0.73 M-1 s-1 and k2 ) 0.1 s-1, and
normalizing for the percentage of active sites at 65% gives kp
) 1.1 M-1 s-1 and k2 ) 0.15 s-1
.
On examination of the denominator in eq 14, the value of
the dimensionless group
k1[M]
[Π]
G )
)
(17)
k-1 + k2 [IP]
is related to whether chain propagation is zero or first order in
monomer. Chain propagation is zero order when G .1 and first-
order when G , 1. If the π-complex formation energy is
substantially positive, one predicts [Π] , [IP] and thus first-
order in monomer propagation. If the π-complex formation
energy is substantially negative, one predicts [Π] . [IP] and
thus zero-order in monomer propagation. If the π-complex
formation energy is close to 0, one predicts [Π] ≈ [IP] and
chain propagation may transition between zero and first order.
The decrease in π-complex formation energy as KS increases
leads to increasing G, which can cause the reaction to transition
from first to zero order in monomer as KS increases. This
provides a possible explanation for why the chain propagation
reaction for catalyst 24a was first order in monomer for
bromobenzene solvent but combined zero and first order for
1,2-dichlorobenzene solvent.
The structure-activity correlation for kp depends upon (i) the
quantitative descriptors EIPS, θCp′, and θOAr that are computed
theoretically and (ii) adjustable parameters R, f, E0, and a0 that
are fit using a training set of experimental data. Similar catalysts
are grouped into families such that all catalysts in a single family
have the same values of the adjustable parameters. The
parameters R and E0 could potentially be computed by purely
theoretical means by regressing a series of DFT-computed Ea
values to EIPS values to determine the slope R and intercept E0
for a given family. The steric factor f accounting for space
blocked by the growing chain and partially displaced counterion
should be constant for a given monomer and counterion, while
a0 just scales the reaction rate for the whole catalyst family.
Thus, the parameters are actually a very small set and all have
a clear physical meaning. Since all of the catalysts in a single
family have the same values of the adjustable parameters and
the quantitative descriptors can be computed theoretically, it is
possible to predict the kp value ahead of time for a new catalyst
in an established family.
Structure-Activity Correlation for Chain Propagation
Previously we reported a structure-activity correlation for
1-hexene chain propagation rate constants for a series of Ti
mixed Cp′/ArO catalysts activated with B(C6F5)3 in toluene
solvent.11 This structure-activity correlation was based on the
observation that monomer access to the metal center is the key
factor controlling the reactivity of these catalysts. Monomer
access to the metal is in turn controlled by steric congestion
and the difficulty of partial ion pair separation. Ligand cone
angles were used to quantify the amount of steric congestion at
the metal, while EIPS was used to quantify the difficulty of ion
pair separation. The structure-activity correlation took the form
of an Arrhenius-like equation:
Since the reactions were all performed at 0 °C, a0 and E0
were absorbed into a parameter A ) a0 exp(-E0/RT) that was
optimized for each catalyst family. We propose that mixed Cp′/
ArO catalyst families are determined according to the following
factors: (a) the metal center, i.e. Zr or Ti, (b) the structure of
the Cp′ ligand, i.e. Cp, Ind, Cp*, etc., (c) the initiating group,
i.e. methyl, benzyl, etc., (d) the counterion, i.e. [MeB(C6F5)3]-,
[B(C6F5)4]-, etc., (e) whether or not the aryloxide ligand
contains non-H substituents in both ortho positions, (f) whether
or not the aryloxide ligand contains at least one ortho substituent
that can opportunistically bond to the metal, and (g) whether
or not the aryloxide ligand contains substituents that can
potentially lead to undesirable coordination to the activator or
a second catalyst center, e.g. methoxy substituents.
-1
RT
kp ) k0e-E ⁄RT ) γa0 exp
(E0 + REIPS
)
(18)
a
[
]
The activation energy was linearly correlated to EIPS in solution
by
Ea ) E0 + REIPS
(19)
The pre-exponential factor was linearly correlated to the free
solid angle 4πγ for monomer approach to the metal center
k0 ) γa0
(20)
Catalysts in the same family have similar features in
categories a-g above. The division of catalysts into different
families is necessary in part because basic structural changes
cause the catalyst’s kinetic behavior to change. For example,
Cp* catalysts with two ortho alkyl substituents on the aryloxide
ring (e.g., 10a-12a, 14a, 31a) had slow chain initiation and
slow chain transfer. Catalysts with opportunistically bonding
ligands (e.g., 16a, 19a-22a, 24a, 30a) had facile chain initiation
irrespective of the amount of steric congestion. Cp* catalysts
where the free solid angle for monomer approach was computed
by subtracting the solid angles occupied by each of the ligands:
γ ) the larger of (1 - sin2(θCp’ ⁄ 4) - sin2(θOAr ⁄ 4) - f) and 0
(21)
The parameter f accounts for space blocked by the growing
polymer chain and partially displaced counterion and was fitted
from the data.

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5517
with zero or one ortho substituent on the aryloxide ring (4a-9a,
13a, 15a, 17a, 18a, 20a, 21a) had facile chain initiation but
slow chain transfer. Cp catalysts with two ortho Me, Et, or iPr
substituents on the aryloxide ring (1a-3a) had facile chain
initiation and facile chain transfer.
[B(C6F5)4]- is approximated by exp(R(EIPS[MeB(C6F5)3] -
EIPS[B(C6F5)4])/RT) and is shown in Table 2. This initial
prediction underestimates the experimental ratios by a factor
of ca. 2.2, and this suggests the family parameter A ) a0
exp(-E0/RT) for the [Cp*Ti(OC6H3-2,6-R2)Me]+[B(C6F5)4]-
family is about 2.2 times that for the [Cp*Ti(OC6H3-2,6-
R2)Me]+[MeB(C6F5)3]- family. This adjustment is taken into
consideration to give the refined predictions displayed in the
last column of Table 2, and these values are in good agreement
with experiment.
The experimental data for catalysts 1a-18a were previously
fit to the above equations to yield optimized values R ) 0.300
and f ) 0.187.11 The new catalyst 31a provides a test of this
correlation, because it falls into a family for which the parameter
A is known. Using the parameter A ) 2.65 × 107 M-1 s-1 for
this family and the computed EIPS and ligand cone angles, the
predicted propagation rate constant is 0.64 M-1 s-1 for 1-hexene
polymerization in toluene at 0 °C, which is in good agreement
Effects of Opportunistic Ligand Coordination. Computa-
tional investigation into the experimentally observed high
reactivity of 16a led to the discovery that an opportunistically
coordinating group in the ortho position of the aryloxide ring
with the experimental value of kp ) 0.55 M-1 s-1
.
Effect of Solvent. A change in solvent affects EIPS but not
the ligand cone angles. If the parameters R, f, E0, and a0 are
presumed to be similar in different solvents, this corresponds
to the following ratio of rate constants:
can greatly enhance the catalyst reactivity by lowering EIPS
.
Additional experiments and computations showed that, to
minimize steric congestion, this group should only be placed
in one of the ortho positions. Placement of the group in both
ortho positions (e.g., 2,3,5,6-Ph4 16a) can lead to lower reactivity
compared to placement of the group in one ortho position (e.g.,
2-Ph 21a) due to the very high steric congestion associated with
Cp* 2,6-substituted aryloxide complexes. Synthesis of catalyst
20a was motivated by computations showing that an ortho Br
substituent exhibits stronger opportunistic coordination and
lower EIPS than an ortho phenyl substituent. The kp value of
20a was subsequently found to be the highest of any of the
catalysts we investigated. This is a clear demonstration that the
proposed structure-activity correlation can be used to design
a more reactive catalyst.
The system Cp*Ti(OC6H4-2-X)Me2/B(C6F5)3 for X ) F, Cl,
Br, I, Ph forms a family of similar catalysts that have lower
EIPS and presumably higher reactivity than the corresponding
Cp*Ti(OC6H4-4-X)Me2/B(C6F5)3 catalysts. For example, cata-
lyst 20a (2-Br) has EIPS 7.89 kcal/mol lower44 and a kp value
about 21 times higher than for 7a (4-Br). Catalyst 21a (2-Ph)
has EIPS 5.51 kcal/mol lower and a kp value about 8.2 times
higher than for 8a (4-Ph). At the present time, there is not
enough experimental data to accurately determine the parameters
R and A for this family; however, anticipated reactivity trends
can still be predicted for different catalysts in this family.
Catalyst 28a (2-I), which has not been synthesized, is anticipated
to have even higher reactivity than catalyst 20a (2-Br), due to
its lower EIPS and similar steric congestion. The remaining
catalysts in this series, 26a (2-F) and 27a (2-Cl), have higher
EIPS values and are anticipated to have lower kp values than
20a (2-Br).
Effect of Metal. When the structure of the Cp′/ArO catalyst
is held constant except for a change in metal from Ti to Zr,
there is a large increase in EIPS but only a slight decrease in
steric congestion. For example, the Ti-based catalyst 16a has
an EIPS value in toluene that is 6.1 kcal/mol lower than for the
Zr analogue (catalyst 24a), but θCp′ + θOAr is only 13° higher
for 16a than for 24a. Due to the tighter counterion binding,
monomer coordination is more difficult for 24a compared to
16a. Catalyst 16a is reactive in toluene, while 24a is not. To
compensate for the tighter counterion binding in the Zr-based
system, it was necessary to use the Zr-based catalyst in a more
polar solvent such as bromobenzene or 1,2-dichlorobenzene.
In bromobenzene, catalyst 16a consumed all monomer within
kp[solvent2]
kp[solvent1]
∆Ea
= exp
(22)
(
)
RT
where
∆Ea = R(EIPS[solvent1] - EIPS[solvent2])
(23)
Equation 22 can be used to predict the dependence of kp on
solvent if the reaction mechanisms are the same in both solvents.
This requires that the metal docking site be occupied by the
same electron donor system (i.e., counterion, solvent, ligand,
or growing chain) in the kinetically dominant form of IP in
both solvents; however, the overall reaction order need not be
the same in both solvents, since the same reaction mechanism
can give rise to different overall reaction orders.
For catalyst 24a in bromobenzene and 1,2-dichlorobenzene,
the docking site is occupied by an H atom from the growing
polymer via ꢁ-agostic bonding. In both of these solvents, the
monomer must displace this H atom from the docking site in
order to form the π-complex and the counterion acts only as a
spectator. Since the reaction mechanisms are similar, the ratio
of kp values in bromobenzene and 1,2-dichlorobenzene can be
estimated using eq 23. Using R ) 0.3, the predicted lowering
of activation energy for 1-hexene polymerization by catalyst
24a upon changing from bromobenzene to 1,2-dichlorobenzene
solvent is 1.52 kcal/mol. At 25 °C this corresponds to a predicted
increase in kp by a factor of 13. As shown in Table 4, the
experimentally measured ratio was 16, which is in reasonable
agreement.
Effect of Counterion. As a result of weaker counterion
binding of [B(C6F5)4]- versus [MeB(C6F5)3]-, the rate of
polymerization is expected to increase significantly upon
changing [MeB(C6F5)3]- to [B(C6F5)4]-. This agrees with the
experimental results given in Table 2, where the 1-hexene
polymerization activity increased by a factor of 20-50 for
catalysts 10-12 upon changing [MeB(C6F5)3]- to [B(C6F5)4]-.
A rigorous prediction of the dependence of polymerization
rate on counterion should take into consideration changes in
the counterion size and other factors. These factors can be
included implicitly in the above structure-activity correlation
by using different catalyst families for [MeB(C6F5)3]-- and
[B(C6F5)4]--based catalysts. However, a rough prediction of the
rate increase upon switching from [MeB(C6F5)3]- to [B(C6F5)4]-
can be made by assuming the parameters R, f, E0, and a0 are
similar for the two counterions. Using these simplifications, the
increase in activity upon changing from [MeB(C6F5)3]- to
(43) Marques, M. M.; Dias, A. R.; Costa, C.; Lemos, F.; Ribeiro, F. R.
Polym. Int. 1997, 43, 77–85.
(44) For this catalyst family, the average of EIPS values for the proximal
and distal conformations have been used, since these conformations have
similar total energies and EIPS values.

5518 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
2 min, while catalyst 24a had a slow but measurable polym-
erization rate. In addition, the rate constant k2 for insertion proper
is lower for the Zr-based catalyst 24a than for the Ti-based
analogue. Specifically, for catalyst 24a in 1,2-dichlorobenzene
k2 ) 0.1-0.15 s-1, and for catalyst 16a in toluene the initial
turnover rate was 0.9 s-1, giving k2 g 0.9 s-1. The value of k2
should be relatively insensitive to a change in the dielectric
constant KS of solvent, since the ion pair separation distance is
nearly constant during insertion proper.
The slower rate of 24a compared to that of 16a is due to
both higher EIPS leading to slower monomer coordination and
lower k2 leading to slower insertion proper. These two factors
combine to make 24a orders of magnitude less reactive than
its Ti-based analogue 16a. Because the intrinsic rate of monomer
insertion, k2, is different for these two metals, Ti and Zr catalysts
shouldnotbeclassifiedintothesamefamilyinthestructure-activity
correlation described above.
with the same ligands and consequently the rate of propagation
for Ti catalysts is greater than for the analogous Zr catalysts.
These findings provide additional evidence supporting the
proposed quantitative structure-activity model for chain propa-
gation, where the solvent affects the interaction between the
metal center and counterion via dielectric shielding. We believe
that this is the most extensive development of a quantitative
structure-activity model for single-site polymerization catalysts
to date and shows the value of quantitative modeling for these
systems.
Other groups have also pursued the development of structure-
activity models, which are consistent with the findings above.
Chen and Marks report an inverse relationship between polym-
erization activity for five constrained-geometry catalysts and
the free energy of activation, ∆Gqreorg, required for the coun-
terion to separate from the metal center and redock following
a chain swing.6 They used two different metals (Ti, Zr) and
three different counterions to generate the five catalysts, and
all five catalysts had the same ligands. Because counterion
binding strength is a major contribution to ∆Gqreorg, there is a
Discussion
strong relationship between the ∆Gq
and EIPS descriptors.
reorg
A systematic study of the relationship between the rate of
chain propagation and the catalyst structure as computed via
DFT has been reported for a series of aryloxide-ligated Ti and
Zr single-site catalysts for the polymerization of 1-hexene, where
the effects of solvent, metal, counterion, and ligand structure
were investigated. Significant findings include the following.
(1) The data support a previously developed structure-activity
model.11 The rate of propagation is determined by the difficulty
of the monomer to gain access to the metal center and form a
π-complex, and the rate of propagation is a function of the ion
pair separation energy and the steric congestion at the catalytic
site, as described by the cone angles of the ligands and growing
polymer chain.
A major difference is that ∆Gq
is an experiment quantity;
reorg
in contrast, EIPS can be computed via DFT for a catalyst not
yet synthesized. Counterion Lewis acidity strength, counterion
size (i.e., van der Waals volume), ion pair formation energy
from the precatalyst, and NMR shifts of the methyl initiating
group are other quantitative descriptors that have been proposed
to correlate the effects of counterion structure on the reactivity
of single-site olefin polymerization catalysts.4,6 These descriptors
are limited because they require aspects of the catalyst structure
to be held constant. Specifically, the first two descriptors do
not consider the nature of the cation and thus cannot explain
any changes in reactivity observed when the cation structure is
changed. Ion pair formation energies and NMR shifts depend
upon a variety of factors, including the metal and initiating
group, in addition to counterion binding strength. In summary,
EIPS is a more appropriate quantitative descriptor of counterion
binding strength, because (i) it is directly related to the energy
required for the separation of the counterion required for the
monomer to form the π-complex and (ii) it naturally accounts
for variations in ligands, counterion, metal, and solvent.
Conclusion 5 above concerning the effect of the solvent’s
dielectric constant on the formation of different kinds of ion
pairs is consistent with the experiments by Eisch et al. using
multinuclear NMR.14 Their experiments indicate that more polar
solvents and higher dilution favor formation of an SSIP over
an ISIP. They observed that small arene solvents favored
formation of the SCAP, where σ or η1 binding is postulated.
Alternatively, larger arene solvents cannot form the SCAP due
to steric limitations. These results are in general agreement with
our computations showing that high-polarity solvents favor the
SSIP over the ISIP and that smaller arenes may lead to the
formation of η1-bound SCAP. The relative reactivities of these
ion pairs were found to be SSIP > ISIP > SCAP.14
(2) The effect of the type of solvent on the ion pair separation
energy, and hence the rate of propagation, is determined by
electrostatics, where EIPS is a linear function of KS-1 (see eq 3)
and KS is the solvent’s dielectric constant.
(3) For catalyst 24a the monomer consumption was first order
in bromobenzene, but when the solvent was changed to 1,2-
dichlorobenzene with a higher dielectric constant, the reaction
was initially zero order but changed to first order as the reaction
proceeded. This change in reaction order with solvent and during
the course of reaction can be quantitatively explained, assuming
formation of the π-complex is a reversible reaction and the
insertion of the monomer to increase the chain length is an
irreversible reaction. The overall reaction (i) is zero order in
monomer if the π-complex formation energy is substantially
negative, (ii) is first order if the π-complex formation energy is
substantially positive, and (iii) exhibits a transition from zero
order to first order if the π-complex formation energy is
approximately 0.
(4) It was observed that ligands with ortho substituents
capable of donating electrons that can opportunistically bond
with the metal center, as determined by the Wiberg bond index,
have a significantly lower EIPS and consequently a much faster
rate of propagation.
(5) Consistent with simple electrostatic arguments, DFT
simulations indicate that in lower dielectric constant solvents
the catalyst/counterion system exists as either an inner-sphere
ion pair (ISIP) or an outer-sphere ion pair (OSIP), while in high
dielectric constant solvents the system prefers to be an OSIP
or solvent-separated ion pair (SSIP).
There is no clear consensus in the literature regarding the
effects of solvent polarity (KS) on catalyst reactivity. In some
cases, an increase in KS leads to orders of magnitude increase
in rate; in other cases, an increase in KS leads to only a modest
increase in rate; in yet other cases, an increase in KS has almost
no effect on the polymerization rate.7,14,45 These results imply
a threshold beyond which increasing KS does not further enhance
the catalyst reactivity, and this threshold is dependent upon the
(6) DFT simulations indicate that Zr catalysts more tightly
bind the counterion (i.e., the EIPS is greater) than Ti catalysts
(45) Klamo, S. B. Ph.D. Thesis; Chemistry Department, California
Institute of Technology, 2005.

Polymerization Catalyzed by Mixed Cp′/ArO Complexes
Organometallics, Vol. 27, No. 21, 2008 5519
catalyst system under investigation. One potential explanation
for this behavior is that increased KS is expected to enhance
the reaction rate only if an ISIP is the global ground state, since
formation of monomer coordination and insertion proper transi-
tion states from ISIP requires partial ion pair separation;
however, for solvents where an OSIP or SSIP is the global
ground state, any further increase in KS may have little effect
on reactivity, since formation of monomer coordination and
insertion proper transition states requires little (if any) further
ion pair separation. Additional experiments are needed to
validate this hypothesis
contrast, a detailed microkinetic model can describe the effect
of catalyst, monomer, cocatalyst, and/or activator concentration
on the catalyst performance, where the various rate constants
are subsequently correlated with the molecular structure of the
various species. The development of a QSAR for the chain
propagation rate constant discussed in this article is a significant
improvement over the more traditional use of QSAR to model
activity. To the best of our knowledge, there is only one other
QSAR study for homogeneous group 4 single-site olefin
polymerization catalysts that correlates quantitative descriptors
to the chain propagation rate constant, kp. Stovneng et al.
developed a QSAR for ethylene polymerization for 9 unbridged
zirconocenes activated with MAO.49 Principal-component analy-
sis followed by partial least-squares regression was used to
correlate kp and activity for 9 catalysts to 17 descriptors. Better
correlations were obtained for kp than for catalyst activity.
Several steric and electronic parameters had significant regres-
sion coefficients; thus, Stovneng et al. concluded that catalyst
reactivity is a complex function of molecular structure, but they
did not report the actual optimized coefficients for the individual
QSAR descriptors.
An important descriptor is the steric congestion at the catalytic
site due to the presence of the ligands, growing polymer chain,
and the counterion. Mo¨hring and Coville reviewed various types
of ligand cone angles, solid angles, and other steric descriptors
used in a QSAR description.48 For zirconocenes, ligand cone
angles, solid angle, and coordination gap aperture were found
to be functionally equivalent, because the catalyst activity
decreased linearly with increasing ligand cone angle, increasing
solid angle, or decreasing coordination gap aperture.48,50
Consequently, the monomer accessibility factor γ in eq 20 can
also be expanded in terms of the combined ligand solid angle
Ω instead of ligand cone angles, where Ω is the solid angle
simultaneously subtended by the van der Waals surfaces of all
ligands except the growing polymer chain. The monomer
accessibility factor γ can be rewritten in terms of solid angle
as
In addition to changing counterion and solvent polarity, we
have shown that opportunistic ligand coordination can decrease
EIPS and thereby increase catalyst reactivity. Although flexible
ligand bonding has been described in the literature for several
group 4 single-site olefin polymerization catalysts,6,46 to the best
of our knowledge the specific use of flexible ligand bonding to
decrease EIPS has not been reported previously. In our picture,
opportunistic ligand coordination involves a group attached to
the ligand backbone that does not coordinate to the metal center
in the ISIP, but the group partially coordinates to the metal
center as the counterion leaves, thereby reducing EIPS. The
opportunistically coordinating group must be capable of exhibit-
ing partial coordination to the metal center when the counterion
is only partially displaced; otherwise, the opportunistic coor-
dination does not facilitate the early stage of ion pair separation
crucial for monomer approach to the metal center. The early
stage of ion pair separation is crucial, because computations
show only partial ion pair separation occurs in the monomer
coordination transition states. Moreover, the opportunistically
coordinating group should be located off to the side of the
catalytic site in order to allow room for the approaching
monomer to enter. We believe opportunistic ligand coordination
is a potential strategy for decreasing the EIPS of various of single-
site olefin polymerization catalysts containing a counterion that
must be displaced from the metal to allow formation of an
olefinic π-complex. This provides a potential strategy for
increasing the reactivity of a wide range of different single-site
olefin polymerization catalysts.
γ ) the larger of (1 - Ω ⁄ (4π) - fΩ) or 0
(24)
where fΩ is a fitted parameter analogous to the parameter f in
eq 21 that accounts for space blocked by the growing polymer
chain and partially displaced counterion. Several papers have
discussed the merits of ligand cone angles and solid angle as
steric descriptors.48,51 The solid angle offers two potential
advantages over ligand cone angles: (a) the solid angle properly
accounts for the overlap between the van der Waals surfaces of
two or more ligands52 and (b) the solid angle can be used for
bridged ligands where ligand cone angles may not be as useful.
A potential disadvantage of solid angle compared to ligand cone
angles is that the van der Waals surfaces of the ligands may
contain inaccessible crevices which are properly counted as
inaccessible space by the ligand cone angles but improperly
Quantitative structure-activity relationships (QSARs) can
potentially be used both to explain reactivity trends of existing
catalysts and to intelligently design new catalysts. A key
drawback of most existing QSARs for single-site olefin polym-
erization catalysis is that they correlate catalyst descriptors to
activity rather than to rate constants.47,48 For many olefin
polymerization systems, catalyst initiation may not be facile,
catalyst deactivation may be appreciable, and the percentage
of active sites may be less than 100%; thus, the kp value per
active site can be significantly different from that indicated by
simple activity measurements. Consequently, QSARs based on
activity are of limited value, because they can only predict
catalyst performance for a fixed set of reaction conditions. In
(46) (a) Froese, R. D. J.; Musaev, D. G.; Matsubara, T.; Morokuma, K.
J. Am. Chem. Soc. 1997, 119, 7190–7196. (b) Froese, R. D. J.; Musaev,
D. G.; Morokuma, K. Organometallics 1999, 18, 373–379. (c) Vanderlinden,
A.; Schaverien, C. J.; Meijboom, N.; Ganter, C.; Orpen, A. G. J. Am. Chem.
Soc. 1995, 117, 3008–3021.
(47) (a) Yao, S.; Shoji, T.; Iwamoto, Y.; Kamei, E. Comput. Theor.
Polym. Sci. 1999, 9, 41–46. (b) Cruz, V. L.; Ramos, J.; Martinez, S.; Munoz-
Escalona, A.; Martinez-Salazar, J. Organometallics 2005, 24, 5095–5102.
(c) Cruz, V. L.; Martinez, S.; Martinez-Salazar, J.; Polo-Ceron, D.; Gomez-
Ruiz, S.; Fajardo, M.; Prashar, S. Polymer 2007, 48, 4663–4674. (d) Cruz,
V.; Ramos, J.; Munoz-Escalona, A.; Lafuente, P.; Pena, B.; Martinez-
Salazar, J. Polymer 2004, 45, 2061–2072.
(49) Stovneng, J. A.; Stokvold, A.; Thorshaug, K.; Rytter, E. In
Metalorganic Catalysts for Synthesis and Polymerization; Kaminsky, W.,
Ed.; Springer-Verlag: Berlin, 1999; pp 274-282.
(50) Janiak, C.; Versteeg, U.; Lange, K. C. H.; Weimann, R.; Hahn, E.
J. Organomet. Chem. 1995, 501, 219–234.
(51) (a) Hirota, M.; Sakakibara, K.; Komatsuzaki, T.; Akai, I. Comput.
Chem. 1991, 15, 241–248. (b) White, D.; Taverner, B. C.; Leach, P. G. L.;
Coville, N. J. J. Comput. Chem. 1993, 14, 1042–1049. (c) Tolman, C. A.
Chem. ReV. 1977, 77, 313–348.
(52) Because the van der Waals surfaces of different ligands may overlap,
the solid angle subtended by multiple ligands is not necessarily equal to
the sum of solid angles subtended by each ligand separately; therefore, the
solid angle simultaneously subtended by the multiple ligands should be
computed.
(48) Mohring, P. C.; Coville, N. J. Coord. Chem. ReV. 2006, 250, 18–
35.

5520 Organometallics, Vol. 27, No. 21, 2008
Manz et al.
counted as accessible space in the solid angle. Since ligand cone
angles are easier to compute, they are the preferred steric
descriptor if ligand geometry is uncomplicated. Both eqs 21
and 24 contain a steric threshold. Specifically, if
structure-activity correlation using DFT-computed ligand cone
angles as a measure of steric congestion and EIPS as a measure
of ion pair separation difficulty was utilized to explain reactivity
changes as a function of metal, ligand, counterion, and solvent
for a series of Cp′/ArO-ligated complexes. However, more
extensive experimental and theoretical analysis is needed to
determine whether the structure-activity correlation of eqs 19
and 20 is applicable to various kinds of group 4 single-site olefin
polymerization catalysts in addition to mixed Cp′/ArO complexes.
(1 - sin2(θCp’ ⁄ 4) - sin2(θOAr ⁄ 4) - f) < 0 or (1 - Ω ⁄ (4π) -
fΩ) < 0 (25)
there is not enough room to accommodate the incoming
monomer and growing polymer chain, and the predicted
polymerization rate drops to 0. For 1-hexene polymerization
by Cp′/ArO-ligated Ti catalysts, the steric parameter was f )
0.187, which predicts Ti catalysts containing Cp* combined with
a ligand sterically larger than OC6H-2,3,5,6-Ph4 should not be
active for the polymerization of 1-hexene. Janiak et al. reported
experimental evidence that qualitatively supports the idea of a
steric threshold for zirconocenes, where the activity dropped to
slightly above 0 for highly congested catalysts.48,53
Acknowledgment. Financial support was provided by the
U.S. Department of Energy by Grant No. DE-FG-
0203ER15466. The computational resources for the DFT
calculations were provided by the National Center for
Supercomputing Applications Grant No. MCA04N010 and
Information Technology at Purdue (ITaP).
Supporting Information Available: Text, tables, and figures
giving experimental details and kinetic analysis and DFT-optimized
geometries and energies and CIF files giving X-ray crystal structures
for precatalysts 14, 22, 24, and 31. This material is available free
of charge via the Internet at http://pubs.acs.org.
In summary, ion pair separation energy and steric congestion
at the metal center appear to be important descriptors for a wide
range of group 4 single-site olefin polymerization catalysts. A
(53) Janiak, C.; Lange, K. C. H.; Versteeg, U.; Lentz, D.; Budzelaar,
P. H. M. Chem. Ber. 1996, 129, 1517–1529.
OM8004993