Modeling of the active site in TiCl4MgCl2 based ZieglerNatta heterogeneous catalysts

1
Modeling of the active site in TiCl4/MgCl2 based
Ziegler-Natta heterogeneous catalysts

• L. Petitjean and T. Ziegler
• University of Calgary
• 2500 University Drive, NW
• Calgary, AlbertaT2N 1N4 CANADA

2
1. Introduction

• The catalytic polymerization of ?-olefin, known
  as Ziegler-Natta catalysis, is of high interest
  for industry.1 This reaction is indeed able to
  produce polymer from ?-olefin under mild
  conditions with high activity and is the only way
  to get regular polymers, such as isotactic
  polypropylene (iso-PP), which have a wide range
  of applications. Since the structure of the
  active species involved in this catalysis has
  still not been yet clearly identified
  experimentally, molecular modeling seems to be a
  good tool to understand how this catalytic system
  works. Some theoretical works 2-13 have already
  attempted to give some insight on this topic, but
  the complexity of the active site has been for a
  long time a significant problem for the
  calculations.
• Fortunately, recent advances in theoretical
  methods (i.e. QM/MM methods) give us now access
  to larger molecular models, by treating only a
  small part of the system (reaction site) by
  quantum mechanical method and the other part by
  molecular mechanics potential (environment). The
  objective of this work is to apply such hybrid
  QM/MM method to investigate the structure of the
  active site in order to bring a new insight to
  this problem.

3

• The catalytic systems used in industrial process
  are basically composed by molecular TiCl4
  supported on MgCl2 crystal to which is added an
  AlEt3 as co-catalysts.
• In order to improve the activity and
  stereo-selectivity of the catalyst two Lewis
  bases have been added to initial catalytic
  system. (called ILB and ELB). These two compounds
  have a dramatic impact on the catalyst
  performance. Their mode of action is not clearly
  determined at this time but they are supposed to
  be a part of, or at least to interfere with, the
  actual active center.
• It is consequently very interesting for both
  theoretical and experimental point of view to
  imagine and calculate a model, which includes the
  presence of a Lewis Base.

Typical ILB
Typical ELB
4
2. Choice of the model

• An important part of our work was to define a
  potential model of the active site. We have
  chosen to describe it as the product of the
  adsorption of a single TiCl4 molecule on the
  100 surface of the MgCl2 support. In this
  structure the octahedral sphere of Ti is not
  filled (5-fold species). The main reasons that
  brought us to this choice are developed
  hereafter.
• We have considered the formal oxidation state of
  the catalysts to be Ti (IV). Even if other
  oxydation states of Ti may be present on the
  active catalysts, we believe, like other authors
  on the basis of experimental evidences, 16 that
  the dominant catalytic species adopts this
  oxydation state.

TiCl4 on 100 MgCl2 surface
5

• The MgCl2 crystal gives rise principally to
  hexagonal shaped crystallites. This means that
  there is mainly two host surfaces for the
  adsorption of TiCl4 molecules 100 and 110.
• On the 110 surface, it has been shown that the
  adsorption of TiCl4 is more likely to form 6-fold
  species (octahedral sphere of Ti filled).4 The
  reason being that a 6-fold structure is able to
  form 4 bonds with the surface while the 5-fold
  only forms 3. This stabilizes the 6-fold
  structure by 8,1 kcal/mole with respect to the
  other, even if the deformation necessary to form
  this 6-fold species is higher than for the 5-fold
  one.

6-fold
5-fold
6

• On the 100 surface, there can be also formation
  either of 6-fold or 5-fold structures. The 6-fold
  structure is formed via adsorption of Ti2Cl8
  dimeric molecules on the surface.11 However,
  the number of shared bonds with the surface per
  titanium atom is only 5/2 in the case of dimeric
  Ti2Cl8 while it is 3 for the mononeric TiCl4.
  Since, the deformation necessary to adsorb the
  original TiCl4 molecule is also larger in the
  case of the dimeric species, we have concluded
  that a 5-fold structure would be favored on the
  100 surface. Since, as explained in the
  following a 5-fold species is more likely to form
  a polymerization center, we have chosen this
  surface as a model for MgCl2 support.

6-fold
5-fold
7

• After the adsorption on the surface and prior to
  the olefin insertion, the active center must get
  in some way its first Ti-C bond. Experimentally,
  in most of the case this activation step is done
  by reaction with AlEt3.
• In the case of 6-fold models,11-12 in which the
  octahedral sphere of the Ti is filled, this step
  involves the abstraction of a Cl atom before the
  chlorine/alkyl exchange with AlEt3 can take
  place. This reaction is quite energetically
  expensive.
• With a 5-fold structure, the activation involves
  only a chlorine/alkyl exchange, which is quite
  cheap.
• A 5-fold species is consequently more likely to
  form an actual polymerization center.

AlEt3
DE -9,3 kcal/mole
DE -11,3 kcal/mole
AlEt2Cl
Alkyl exchange with AlEt3
8

• It is known experimentally that by adding an ELB
  before the AlEt3 the molecule instead of
  activating the catalyst kills its activity.This
  effect is very difficult to understand with a
  6-fold site model in which the octahedral sphere
  of the Ti is filled. In that case, no easy
  reaction is possible before the activation.
• By contrast, with a 5-fold model we can imagine
  that a ELB will be able to complex easily the
  TiCl4 center. This will fill the octahedral
  sphere of the Ti and hinder further activation by
  AlEt3.

DE -16,8 kcal/mole
9
3. Building the QM/MM model

• High Resolution 13C NMR spectra can be best
  fitted using a 3-sites type statistics.15 This
  statistical model corresponds to the mixing of an
  enantiomorphic model site (ES), a chain-end model
  site (CE), and a more general model site for
  polymerization called C1. It corresponds to the
  fact that the polymer is composed by 3 different
  stereo-blocks highly-isotactic, syndiotactic
  and isotactoïd (i.e. isotactic with low
  regularity) corresponding respectively to ES, CE
  and C1 models.

10

• We have tried to interpret the NMR features in
  terms of molecular structure and imagined a
  molecular model, which is able to have 3 possible
  reaction centers. The idea is that the presence
  of ELB or ILB molecules around the Ti active site
  would create an asymmetry, which would be at the
  origin of 3 reactive centers (A, B, C). The
  purpose of the QM/MM calculations has been to
  check if those 3 sites could be related to the
  stereo-blocks described by NMR.

Bulk
Bulk
Bulk
Bulk
Reaction Center A
Reaction Center B
Reaction Center C
11

• Our hypothesis can be associated to many
  molecular structures. Indeed, it is generally
  accepted that the active center lies in many
  different environment leading to a distribution
  of properties of the product. Unfortunately, for
  the calculations we had to choose one. We
  decided to pick up the one in which reaction
  centers B and C would be most differentiating.

QM system
QM system
12
4. Computational details

• We have used a QM/MM code implemented at the
  University of Calgary18 using the ADF density
  functional package developed by Baerends et
  al.19 The molecular mechanics part has been
  treated using the Sybyl force field 20 for C,
  N, O, H and Cl atoms. Van der Waals parameters,
  which are not present in Sybyl, have been added
  from Dreiding 21 for Ti and Mg atom type.
• For the quantum part, the electronic
  configuration of the molecular systems were
  described by a triple-? basis set on Ti atom for
  3s, 3p, 3d and 4s, while double-? STO basis set
  with polarization functions were applied for Mg,
  Cl, C and H atoms. 22The 1s electron of C atoms
  and the 1s-2p electrons for Mg, Cl and Ti atoms
  were treated as frozen core. The auxiliary s, p,
  d, f and g STO functions, 23 centered on all
  nuclei, were used to fit the electron density and
  the Coulomb and exchange potentials in each SCF
  cycle. The B-LYP exchange-correlation functional
  24was used in all the calculations.
• In all the calculations, the atoms of the surface
  model have been fixed using the X-ray parameters
  published in the literature 17 for the crystal
  bulk. For the QM part, the surface model is
  Mg2Cl4, which may be seen as modest. However, the
  use of a larger and much more realistic cluster
  model (Mg9Cl4H6) for the MgCl2 layer didnt
  affect dramatically the results.

13
5. Results

• Here is a summary of the QM/MM calculations
  carried out on the active site model with 2,2,6,6
  tetra-methyl piperidine (TMP) as ELB (p.11). We
  want to be able to compare qualitatively the
  results of the calculations with the NMR data.
  This means that we have to determine the type
  polymer produced by each reaction center A,B,C.
• A first step to evaluate the type polymer
  produced by each center is to determine the
  selectivity of the insertion at each of the
  potential position of coordination of the olefin
  (A1,A2,B1,B2,C1,C2). For each position, this
  involves the calculations of two paths
  corresponding to the two face of the olefin (re
  or si). By comparing the energy of the transition
  states we will know which path is favorable.
• In a second step, we have to imagine a succession
  of insertions. In the case of center B and C, we
  have postulated that the polymerization occurs
  via a chain migratory insertion, which means that
  the insertion takes place alternatively in
  position 1 or position 2. Indeed, by looking at
  the ?-H agostic resting states it can be seen
  that the Ti stays in a pseudo octahedral
  conformation between two insertions.

14

• However, this is not true in the case of center
  A, in which the Ti adopts a tetrahedral
  conformation. This involves a movement of the
  chain in an intermediate position between the 1
  and 2 (which correspond to octahedral
  environment). Moreover the two possible
  conformations found for the resting-state have
  the same energy and can be considered to
  interconvert easily.

Two conformations of canter A resting-state
15

• Selectivity of Center A
• The presence of TMP molecule induces an asymmetry
  between position 1 and position 2. It turns out
  that the transitions states corresponding to
  insertion with olefin in position 2 are lower in
  energy. It seems then reasonable to think that a
  majority of insertion will occur through a path
  which involves insertion with olefin in position
  2.
• Additionally, due to the presence of surface
  atoms the chain cannot rotate (pointing down in
  the figure). When the olefin is in position 2,
  the re insertion will occur by trans approach of
  propylene while si one will happen by cis
  approach. The insertion will be favored when
  occurring through a trans approach (here re).
• Consequently, we believe that Center A will
  produce isotactic polymer.

re (E3.4)
si (E5.6 )
re (E 3.0)
si (E 0)
Position A1
Position A2
16

• Selectivity of Center B
• When the olefin is in position 1, due to the
  presence of surface atoms nearby, the chain
  cannot rotate to avoid the pressure of the
  incoming olefin. The re insertion will occur by
  trans approach of propylene while si one will
  happen by cis approach. The more favorable
  corresponds to the trans approach insertion (here
  re).
• By contrast, when the olefin is in position 2,
  the chain can easily rotate to avoid the pressure
  of the olefin. Both re and si insertions will
  occurs favorably by the less sterically demanding
  trans approach. The insertion is almost not
  selective.
• Consequently, one can think that B would produce
  a hemi-isotactic type of polymer.

re (E0)
si (E2.6)
re (E0)
si (E0.5)
Position B1
Position B2
17

• Selectivity of Center C
• As in the case of B and due to the presence of
  atoms from both surface and TMP molecule nearby,
  the chain cannot rotate to avoid the pressure of
  the incoming olefin when the olefin is in
  position 1. As already explained, the more
  favorable insertion corresponds to the trans
  approach insertion, in that case for the si face.
• When the olefin is in position 2, the presence of
  TMP molecule hinders the rotation of the chain.
  Again it cannot avoid the pressure of the
  incoming olefin. The favorable trans approach
  insertion is here the re one.
• Consequently, center C can be considered to
  produce syndiotactic polymer.

re (E4.6)
si (E0)
re (E0)
si (E2.9)
Position C1
Position C2
18
6. Conclusions

• We have presented here a new approach for the
  modeling of the active site of heterogeneous
  Ziegler-Natta catalysts. This model consists in a
  TiCl4 molecule adsorbed on a 100 MgCl2 surface
  surrounded by two molecules of ELB. These
  molecules create an asymmetric environment that
  makes possible to imagine a structure with 3
  possible reaction centers.
• QM/MM calculations have been carried out on this
  type of model with 2,2,6,6 tetramethyl piperidine
  (TMP) as ELB. We have been able to identify the
  type of polymer produced by 3 reaction centers
  (A, B and C). A produces higlhy isotactic
  polymer, B seems to produce hemi-isotactic
  polymer, while C could produce syndiotactic
  polymer.
• Our model seems in fairly good quliative
  agreement with NMR datas describing the formation
  of stereo-blocks in polypropylene. By this way,
  it gives some insight on the influence of the ELB
  on the catalysts selectivity. The presence of ELB
  molecules is at the origin of the formation
  isotactic polymer by reaction center A.
  Syndiotactic blocks formed by center C would not
  exist without the presence of the ELB.
• In our future works, we will focus on describing
  the evolution of stereospecificity of the
  catalyst using a series of ELB molecules. With
  our model we will try to understand what are the
  mechanism that explain the variations of
  stereospecificy in the family of R,R'-dimethoxy
  propane molecules.

19
7. References

• 1 Albizzati, E. Galimberti, M. Catal. Today,
  1998, 41, 159.
• 2 Colbourn, E. A.Cox, P. A. Carruthers, B.
  Jones, P. J. V. J. Mater. Chem., 1994, 805.
• 3 Puhakka, E. Pakkanen, T. T. Pakkanen, T. A.
  Surf. Sci., 1995, 334, 289.
• 4 (a) Boero, M. Parinello, M. Terakura,K. J.
  Am. Chem. Soc. , 1998, 120, 2746. (b) Boero, M.
  Parinello, M. Hüffer, S. Weiss, H. J. Am. Chem.
  Soc., 2000, in print.
• 5 Lin, J. S. Catlow, C. R. A. J. Mater.
  Chem., 1993, 3, 1217.
• 6 Cavallo, L. Guerra, G.Corradini, P. J. Am.
  Chem. Soc., 1998, 120, 2428.
• 7 Fujimoto, H. Yamasaki, T. Mizutani, H.
  Koga, N. J. Am. Chem. Soc., 1985, 107, 6157.
• 8 Sakai, S. J. Phys. Chem., 1994, 98, 12053.
• 9 Novaro, O. Int. J. Quantum. Chem., 1992, 114,
  5832.
• 10 Jensen, V.R. Børve, K.J. Ystens, M. J.
  Am. Chem. Soc., 1995, 117, 4109.
• 11 Corradini, P. Busico, V., Cavallo,L.
  Guerra, G. Vacatello, M. Venditto, V. J. Mol.
  Catal., 1992, 74, 433.
• 12 G. Allegra Makromol. Chem., 1971, 145, 235.
• 13 Barino, L. Scordamaglia Macromol. Theory
  Simul., 1998, 7, 407.
• 14 Armstrong, D. R. Perkins, P.G. Stewart,
  J.J.P. J. Chem. Soc., Dalt. Trans.,1972, 1972.
• 15 (a) Busico, V. Cipullo, R. Monaco, G.
  Talarico, G. Vacatello, M. Chadwick, J. C.
  Segre, A. L. Sudmeijer, O. Macromolecules, 1999,
  32, 4173 and references therein.
• 16 Jones, P. J. V. Oldman, R.J. Transition
  Metals and Organometallics as Catalysts for
  Olefin Polymerization Kaminsky, W. Sinn, H.,
  Eds, Springer-Verlag Berlin, 1988 p. 223.
• 17 W.G. Wyckoff, Crystal Structure, vol. 2
  Wiley, New-York, 1972.
• 18 Woo, T. K. Cavallo, L. Ziegler, T. Theor.
  Chem. Acc., 1998, 100, 307.
• 19 (a) Baerends, E.J. Ellis D. E. Ros, P.
  Chem.Phys., 1973, 2, 41. (b) te Velde, B.
  Baerends, E. J. J. Comput. Phys., 1992, 99, 84.
  (c) Boeririgter, P.M. te Velde, G. Baerends,
  E.J. Int. J. Quantum Chem., 1988, 33, 87. (d)
  Verluis, L. Ziegler, T. J. Chem. Phys., 1988,
  88, 322.

20

• 20 Clark, M. Cramer III, R.D. Van Opdenbosch,
  N. J. Comput. Chem., 1989, 10(8), 982.
• 21 Mayo, S.L.Olafson, B.D. Goddard, W.A. III
  J. Phys. Chem. 1990, 94, 8897.
• 22 (a) Snijders, J.G. Baerends, E. J.
  Vernoijs. P. At. Nucl. Data Tables, 1982, 26,
  483. (b) Vernoijs, P. Snijders, J.G. Baerends,
  E. J. Slater Type Basis Functions for the Whole
  Periodic System Departemnt of Theoretical
  Chemistry, Free University, Amsterdam, The
  Netherlands, 1981.
• 23 Krijn, J. Baerends, E.J. Fit Functions in
  the HFS Method Department of Theoretical
  Chemistry, Free University, Amsterdam, The
  Netherlands, 1984.
• 24 (a) Becke, A.D. Phys. Rev. A, 1988, 38,
  3098. (b) Lee, C. Yang, W. Parr, R. G. Phys.
  Rev. B, 1988, 37, 785.
• Acknowledgments
• Authors would like to thank Elf-Atochem and Elf
  Aquitaine for their financial support. We would
  like to thankalso Dr. T. K. Woo (University of
  Calgary) for his help in providing the QM/MM code
  for this work. L. Petitjean would like to thank
  Dr J. Malinge, Dr T. Saudemont and Dr D. Pattou
  from the Groupement de Recherches de Lacq
  (Elf-Atochem) for their collaboration in this
  work.