1588
Organometallics 2010, 29, 1588–1602 DOI: 10.1021/om900917k
Role of 1,2-Dimethoxyethane in the Transformation from Ethylene Polymerization to Trimerization Using Chromium Tris(2-ethylhexanoate)-Based Catalyst System: A DFT Study Yuan Qi,† Qi Dong,*,† Lei Zhong,† Zhen Liu,† Pengyuan Qiu,† Ruihua Cheng,† Xuelian He,† Jeffrey Vanderbilt,‡ and Boping Liu*,† †
State Key Laboratory of Chemical Engineering, East China University of Science and Technology, Meilong Road 130, Shanghai 200237, People’s Republic of China, and ‡Westlake Chemical Corporation, P.O. Box 8388, Longview, Texas 75607 Received October 19, 2009
Selective ethylene trimerization into 1-hexene, which is an important comonomer for production of copolymers with high performance, was first commercialized in 2003 using a chromium tris(2ethylhexanoate) (Cr(EH)3)-based catalyst. One milestone during the development of this catalyst is the addition of 1,2-dimethoxyethane (DME) to Cr(EH)3/partially hydrolyzed tri-isobutyl aluminum (PIBAO) catalyst, resulting in a transformation from ethylene polymerization into trimerization. However, the mechanism of this transformation is unknown. In this work, the role of DME in the switching mechanism from ethylene polymerization to trimerization was investigated by a density functional theory (DFT) method based on the 10 most plausible molecular models for the active species of Cr(EH)3/PIBAO catalyst with or without DME coordination. For neutral models (Cr(I)OR; Cr(I)OR/DME; Cr(I)R; Cr(I)R/DME; R refers to isobutyl ligand), those without DME coordination were found to be dominantly ethylene trimerization sites. After DME coordination, both the weak electron-donating effect and steric effect of DME increased the total energy barriers of metallocycle growth, leading to ethylene trimerization with dimerization as a side reaction. So neutral models mainly presented ethylene trimerization with or without DME coordination and were not consistent with experimental evidence of transformation. For cationic models (Cr(I)þ; Cr(I)þ/DME; Cr(II)þOR; Cr(II)þOR/DME; Cr(II)þR; Cr(II)þR/DME), metallocycle growth leading to ethylene polymerization occurred on those models without DME. After DME coordination, the strong electron-donating effect of DME was found to facilitate metallocycle expansion, while the dominant role of the steric effect of DME made nine-membered ring formation unfavorable, and thus 1-hexene was liberated from the seven-membered ring. So all the cationic models presented a transformation from ethylene polymerization to trimerization after DME coordination and might be the most plausible active sites for the Cr(EH)3/PIBAO catalyst system. The results provided a much deeper understanding of the highly selective trimerization mechanism and for further development of new catalysts with high performance as well.
Introduction Linear low-density polyethylene (LLDPE) products made from 1-hexene comonomer present much improved mechanical properties compared with products made from 1-butene comonomer. The development of high-performance LLDPE and other polyolefin copolymers brings a growing commercial demand for 1-hexene. The conventional method to obtain 1-hexene was derived from ethylene oligomerization processes producing various R-olefins with a Schulz-Flory1 or Poisson distribution. This provides a serious challenge to subsequent separation processes to economically obtain comonomer-grade 1-hexene. On the contrary, selective trimerization of ethylene fundamentally avoids this disadvantage. *Corresponding authors. Tel/Fax: þ86-021-64253627. E-mail: tungqi@ hotmail.com;
[email protected]. (1) Flory, P. J. J. Am. Chem. Soc. 1940, 62, 1561. pubs.acs.org/Organometallics
Published on Web 03/02/2010
Ethylene trimerization to 1-hexene was first discovered by Manyik et al.2 together with copolymerization of ethylene and 1-hexene using chromium tris(2-ethylhexanoate) (Cr(EH)3) activated by partially hydrolyzed tri-isobutyl aluminum (PIBAO) in 1967. They found that this Cr(EH)3/ PIBAO catalyst system is mainly an ethylene polymerization catalyst with low ethylene trimerization activity. Many years later, Manyik et al.3 found that adding 1,2-dimethoxyethane (DME) to Cr(EH)3/PIBAO could increase both the activity and selectivity of ethylene trimerization into 1-hexene. Following on this clue, further work on Cr(EH)3/PIBAO/DME catalyst by Briggs4 reported that 1-hexene selectivity and (2) Manyik, R. M. (Union Carbide Corporation) U.S. Patent 3300458, 1967. (3) Manyik, R. M.; Walker, W. E.; Wilson, T. P. J. Catal. 1977, 47, 197. (4) Briggs, J. R. J. Chem. Soc., Chem. Commun. 1989, 11, 674. r 2010 American Chemical Society
Article
Organometallics, Vol. 29, No. 7, 2010
1589
Figure 1. Metallocycle mechanism for ethylene trimerization proposed by Manyik.
Figure 2. Metallocycle mechanism for ethylene trimerization proposed by Briggs.
catalyst activity of ethylene trimerization could be drastically increased to 73% and 2086 g/gCr 3 h, respectively. This work was regarded as a milestone for the development of Cr(EH)3based ethylene trimerization catalysts. It is interesting that upon adding DME, the original polymerization catalyst Cr(EH)3/PIBAO could be transformed into a trimerization catalyst. Such a significant phenomenon implies that DME plays the most important role in this switching process, which deserves further basic research. In the following decades, numerous investigations5 have covered different activators or ligands for various ethylene trimerization catalyst systems. In 2003, the Cr(EH)3/2,5-dimethylpyrrole/ Et3Al/Et2AlCl catalyst system obtained by Phillips Petroleum Company6 has been finally commercialized for producing comonomer-grade 1-hexene with very high activity and selectivity. Paralleling the development of new ethylene trimerization catalyst systems, many investigations have been focused on the mechanism of ethylene trimerization. Because the mechanism proposed by Cossee and Arlman7 could not reasonably explain the high selectivity for 1-hexene through ethylene trimerization, Manyik et al.3 first proposed a metallocycle-based mechanism in 1977 (see Figure 1). This mechanism introduced a metallocyclopentane formed by the coupling of two coordinated ethylene monomers. Then a β-hydrogen in the metallocyclopentane was transferred to a third coordinated ethylene, giving a chromium butenyl ethyl species, from which 1-hexene was liberated by reductive elimination. In 1989, Briggs4 modified the previous mechanism and proposed that the insertion of a third ethylene molecule into the metallocyclopentane yields a
metallocycloheptane instead of the formation of a dialkyl intermediate (see Figure 2). After that, many experiments and density functional theory (DFT) studies have been conducted to prove the metallocycle mechanism proposed above. Emrich et al.8 isolated and characterized crystal structures of the two metallocycles involved in the mechanism and demonstrated that the Cr-cyclopentane was more stable than Cr-cycloheptane and the latter decomposed with the liberation of 1-hexene. Agapie et al.9 successfully distinguished between the metallocycle mechanism for selective trimerization on chromium/PNP catalyst and the Cosseetype mechanism for nonselective ethylene oligomerization on a nickel-based Shell Higher Olefin Process (SHOP) catalyst by deuterium labeling experiments. Arteaga-M€ uller et al.10 detected a tantalacyclopentane intermediate using variabletemperature NMR spectroscopy. These results confirmed that metallocyclopentane and metallocycloheptane are important intermediates responsible for ethylene trimerization into 1-hexene, supporting the mechanism as shown in Figure 2. There still exist some problems and inconsistencies for the metallocycle mechanisms of ethylene trimerization. For example, researchers have not reached an agreement on the specific mechanism for the reductive decomposition of metallocycloheptane to give 1-hexene. Briggs4 suggested that the mechanism on thermal decomposition of platinum metallocycles11 could be applied to chromium-catalyzed trimerization, so a chromium hydride species should be an intermediate before the liberation of 1-hexene. Deckers et al.12 proposed a similar trimerization mechanism for a titanium-based catalyst, in which an alkyl titanium hydride was formed and then 1-hexene was liberated from this hydride species. Overett et al.13 and Elowe et al.14 inferred that 1-hexene was formed through a two-step path involving a hexenyl hydride species in order to explain the formation of
(5) (a) Araki, Y.; Nakamura, H.; Nanba, Y.; Okano, T. (Mitsubishi Chemical Corporation) U.S. Patent 5856612, 1999. (b) Yang, Y.; Kim, H.; Lee, J.; Paik, H.; Jang, H. G. Appl. Catal., A 2000, 193, 29. (c) Carter, A.; Cohen, S. A.; Cooley, N. A.; Murphy, A.; Scutt, J.; Wass, D. F. Chem. Commun. 2002, 858. (d) McGuinness, D. S.; Wasserscheid, P.; Keim, W.; Hu, C.; Englert, U.; Dixon, J. T.; Grove, C. Chem. Commun. 2003, 3, 334. (e) McGuinness, D. S.; Wasserscheid, P.; Keim, W.; Morgan, D.; Dixon, J. T.; Bollmann, A.; Maumela, H.; Hess, F.; Englert, U. J. Am. Chem. Soc. 2003, 125, 5272. (f) Dixon, J. T.; Green, M. J.; Hess, F. M.; Morgan, D. H. J. Organomet. Chem. 2004, 689, 3641. (g) Luo, H.; Li, D.; Li, S. J. Mol. Catal. A: Chem. 2004, 221, 9. (h) van Rensburg, W. J.; van den Berg, J.-A.; Steynberg, P. J. Organometallics 2007, 26, 1000. (i) de Bruin, T.; Raybaud, P.; Toulhoat, H. Organometallics 2008, 27, 4864. (j) Moulin, J. O.; Evans, J.; McGuinness, D. S.; Reid, G.; Rucklidge, A. J.; Tooze, R. P.; Tromp, M. Dalton Trans. 2008, 1177. (k) Wass, D. F. Dalton Trans. 2007, 816. (l) Tobisch, S.; Ziegler, T. J. Am. Chem. Soc. 2004, 126, 9059. (m) Tobisch, S.; Ziegler, T. Organometallics 2004, 23, 4077. (n) Tobisch, S.; Ziegler, T. Organometallics 2005, 24, 256. (6) Freeman, J. W.; Buster, J. L.; Knudsen, R. D. (Phillips Petroleum Company) U.S. Patent 5856257, 1999. (7) (a) Arlman, E. J.; Cossee, P. J. Catal. 1964, 3, 99. (b) Cossee, P. J. Catal. 1964, 3, 80.
(8) Emrich, R.; Heinemann, O.; Jolly, P. W.; Krueger, C.; Verhovnik, G. P. J. Organometallics 1997, 16, 1511. (9) (a) Agapie, T.; Schofer, S. J.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 2004, 126, 1304. (b) Agapie, T.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 2007, 129, 14281. (10) Arteaga-M€ uller, R.; Tsurugi, H.; Saito, T.; Yanagawa, M.; Oda, S.; Mashima, K. J. Am. Chem. Soc. 2009, 131, 5370. (11) (a) McDermott, J. X.; White, J. F.; Whitesides, G. M. J. Am. Chem. Soc. 1973, 95, 4451. (b) McDermott, J. X.; White, J. F.; Whitesides, G. M. J. Am. Chem. Soc. 1976, 98, 6521. (12) (a) Deckers, P. J. W.; Hessen, B.; Teuben, J. H. Angew. Chem., Int. Ed. 2001, 40, 2516. (b) Deckers, P. J. W.; Hessen, B.; Teuben, J. H. Organometallics 2002, 21, 5122. (13) Overett, M. J.; Blann, K.; Bollmann, A.; Dixon, J. T.; Haasbroek, D.; Killian, E.; Maumela, H.; McGuinness, D. S.; Morgan, D. H. J. Am. Chem. Soc. 2005, 127, 10723. (14) Elowe, P. R.; McCann, C.; Pringle, P. G.; Spitzmesser, S. K.; Bercaw, J. E. Organometallics 2006, 25, 5255.
1590
Organometallics, Vol. 29, No. 7, 2010
C6 cyclic byproducts during ethylene tetramerization. But K€ ohn15 experimentally confirmed that alkyl hydride elimination via the two-step path was a “spin blocking” reaction and suggested that the two-step path was a minor path leading to the C6 cyclic products, and the one-step path was dominating and led to ethylene trimerization into 1-hexene. Also the alternative one-step mechanism for 1-hexene liberation was considered in several theoretical studies. Yu and Houk16 reported a one-step mechanism in a tantalum catalyst system called agostic-assisted hydride shift affording 1-hexene liberation and concluded that the nonexistence of a β-agostic interaction in smaller ring sizes led to the disfavor of dimerization. DFT studies on titaniumcatalyzed trimerization found that 1-butene elimination occurred via the two-step mechanism with the formation of the titanium hydride species, while 1-hexene was directly liberated through one-step path.17 For the chromium catalyst system, 1-hexene elimination via one-step agosticassisted hydrogen transfer was obtained.18 However, another computational investigation on a chromium catalyst supported a two-step path rather than a one-step path to give 1-hexene, and this could be possibly attributed to the adoption of a different computational method during molecular modeling according to the authors.19 Another problem for the proposed mechanism of ethylene trimerization is that the oxidation state of the active species during the metallocycle catalytic cycle has not been elucidated yet. van Rensburg et al.18a supported the suggestion that CrII intermediates account for the initial activated catalytic species by molecular modeling. Temple et al.20 reported that CrIII catalyzed ethylene into 1-hexene, and CrII produced a mathematical distribution of oligomers. Br€ uckner et al.21 reported that CrI was responsible for ethylene oligomerization using an in situ ESR spectroscopy technique, but could not totally exclude the possibility of other chromium valence states under the experimental conditions. Jabri et al.22 proposed that CrI led to selective trimerization, CrII to polymerization, and CrIII to nonselective oligomerization. Therefore, further investigation is needed in order to confirm the real oxidation state of the active species for ethylene trimerization using Cr-based catalyst systems. In summary, the ethylene trimerization mechanism has not been understood thoroughly. In our opinion, the most interesting phenomenon in terms of a transformation from ethylene polymerization into ethylene trimerization on the significant Cr(EH)3/PIBAO catalyst system after adding DME is an ideal model system for further investigation of the ethylene trimerization mechanism. The other transformation (15) K€ ohn, R. D. Angew. Chem., Int. Ed. 2008, 47, 245. (16) Yu, Z.-X.; Houk, K. N. Angew. Chem., Int. Ed. 2003, 42, 808. (17) (a) Blok, A. N. J.; Budzelaar, P. H. M.; Gal, A. W. Organometallics 2003, 22, 2564. (b) de Bruin, T. J. M.; Magna, L.; Raybaud, P.; Toulhoat, H. Organometallics 2003, 22, 3404. (c) Tobisch, S.; Ziegler, T. Organometallics 2003, 22, 5392. (18) (a) van Rensburg, W. J.; Grove, C.; Steynberg, J. P.; Stark, K. B.; Huyser, J. J.; Steynberg, P. J. Organometallics 2004, 23, 1207. (b) Bhaduri, S.; Mukhopadhyay, S.; Kulkarni, S. A. J. Organomet. Chem. 2009, 694, 1297. (c) Klemps, C.; Payet, E.; Magna, L.; Saussine, L.; Le Goff, X. F.; Le Floch, P. Chem. Eur. J. 2009, 15, 8259. (19) Blom, B.; Klatt, G.; Fletcher, J. C. Q.; Moss, J. R. Inorg. Chim. Acta 2007, 360, 2890. (20) Temple, C. N.; Gambarotta, S.; Korobkov, I.; Duchateau, R. Organometallics 2007, 26, 4598. (21) Br€ uckner, A.; Jabor, J. K.; McConnell, A. E. C.; Webb, P. B. Organometallics 2008, 27, 3849. (22) Jabri, A.; Mason, C. B.; Sim, Y.; Gambarotta, S.; Burchell, T. J.; Duchateau, R. Angew. Chem., Int. Ed. 2008, 47, 9717.
Qi et al.
phenomena have been reported in the literature. Deckers et al.12a reported that ethylene polymerization was transformed into ethylene trimerization on the [(η5-C5H4CMe2-R)TiCl3]/MAO catalyst system when the ligand substituent R was changed from a methyl to a phenyl group. Later on, the same research group made it clear for such ligand-induced site transformation by examining the catalytic performance of a series of [(η5-C5H3R-(bridge)Ar)TiCl3]/MAO complexes and indicated that the cyclopentadienyl-arene ligand possessed a hemilabile nature during trimerization, which means that the arene group can stabilize intermediates by η6-coordination and make room for incoming ethylene molecules by slipping away.12b The above mechanism was proved using the DFT method,17,23 and similar hemilabile behavior of the pyrrole ligand in the commercial Phillips trimerization catalyst system was also identified by DFT.18a On the basis of experimental and computational results of ethylene oligomerization catalyzed by a chromium complex bearing a R0 N(CH2PR2)2 ligand, Klemps et al.18c reported that bulky groups R, dicyclohexylphosphinyl substituents, led to high selectivity for 1-hexene, whereas less bulky groups R, di-n-butylphosphinyl groups, increased the formation of 1-octene. Duchateau and co-workers24 conducted a series of experiments investigating the interaction between trivalent chromium catalyst precursors with Al-based activators. Results showed that the treatment of deliberately prepared catalysts with or without different cocatalysts led to the switch of catalyst selectivity among ethylene trimerization, nonselective oligomerization, and polymerization. Also as the reaction temperature was increased, a single-component trimerization catalyst became an ethylene polymerization catalyst.24b Monoi et al.25 prepared high-activity ethylene polymerization catalysts by supporting Cr[N(SiMe3)2]3 and alumoxane on silica and found poly[ethylene-co-(1-hexene)] produced at high calcination temperatures of silica. Meanwhile further investigations on silica-supported Cr[N(SiMe3)2]3/isobutylalumoxane catalyst by Monoi and co-workers26 showed that the main product was polyethylene at low calcination temperature, while at high calcination temperature ethylene trimerized into 1-hexene. The authors also found that addition of DME as electron donor could increase 1-hexene selectivity; after all, the silica calcination temperature was regarded as the key factor for transformation from ethylene polymerization to ethylene trimerization. In this contribution, we are interested in studying the role of DME in the transformation from ethylene polymerization into ethylene trimerization on the Cr(EH)3/PIBAO catalyst with or without DME coordination. The role of DME in the switching mechanism from ethylene polymerization to trimerization was investigated by the DFT method based on 10 plausible molecular models for the active species of the Cr(EH)3/PIBAO catalyst system with or without DME (23) de Bruin, T.; Raybaud, P.; Toulhoat, H. Organometallics 2008, 27, 4864. (24) (a) Crewdson, P.; Gambarotta, S.; Djoman, M. C.; Korobkov, I.; Duchateau, R. Organometallics 2005, 24, 5214. (b) Albahily, K.; AlBaldawi, D.; Gambarotta, S.; Duchateau, R.; Koc-, E.; Burchell, T. J. Organometallics 2008, 27, 5708. (c) Albahily, K.; Al-Baldawi, D.; Gambarotta, S.; Koc-, E.; Duchateau, R. Organometallics 2008, 27, 5943. (d) Albahily, K.; Koc-, E.; Al-Baldawi, D.; Savard, D.; Gambarotta, S.; Burchell, T. J.; Duchateau, R. Angew. Chem., Int. Ed. 2008, 47, 5816. (25) Monoi, T.; Ikeda, H.; Ohira, H.; Sasaki, Y. Polym. J. 2002, 34, 461. (26) Monoi, T.; Sasaki, Y. J. Mol. Catal. A: Chem. 2002, 187, 135.
Article
Organometallics, Vol. 29, No. 7, 2010
1591
Scheme 1. Proposed Catalytic Cycle for the Cr(EH)3/PIBAO Catalyst System in this Work
coordination. A much deeper mechanistic understanding of ethylene trimerization over the Cr(EH)3-based catalyst system will be demonstrated.
Theoretical Methods The catalyst system in the current study contained three components: the precursor Cr(EH)3, the ligand DME, and the cocatalyst PIBAO. Considering possible oxidation states of the chromium centers, the 10 most plausible molecular models including four neutral and six cationic active sites for the active species of the Cr(EH)3/PIBAO catalyst system with or without DME coordination are shown in Table 1. All geometry optimizations and transition state searches were performed with the program GAUSSIAN 0327 using the nonlocal method B3LYP, a combination of Becke’s three-parameter hybrid exchange functional28 and the correlation functional of Lee, Yang, and Parr.29 The LANL2DZ (Los Alamos National Laboratory second double-zeta) basis set30 was used for Cr atoms, and the 6-31G(d,p) basis set was used for C, H, and O atoms. All transition states were verified using intrinsic reaction coordinate (IRC) calculations,31 which indicate each transition state is directly connected to the involved reactant and product geometries. All energies reported referred to Gibbs free energy corrections to the total electronic energies at 298.15 K. Geometry optimization of each structure in each model was conducted under all possible spin states to identify the ground (27) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; T. V.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, Revision E.01; Gaussian, Inc.: Wallingford, CT, 2004. (28) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 1372. (29) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B: Condens. Matter 1988, 37, 785. (30) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (31) Fukui, K. Acc. Chem. Res. 2002, 14, 363.
Table 1. Molecular Models Used in This Work (R denoting isobutyl) oxidation states Cr(I)/(III) Cr(II)/(IV)
no. A C E G I
without DME þ
Cr CrOR CrR CrþOR CrþR
no.
with DME
B D F H J
Crþ/DME CrOR/DME CrR/DME CrþOR/DME CrþR/DME
spin states. For Cr(I)/(III) active site models (six models from A to F), three possible spin states (doublet, quartet, and sextet) of Cr(I) species 1X and two spin states (doublet and quartet) of Cr(III) species were considered. For Cr(II)/(IV) models (four models from G to J), three spin states (singlet, triplet, and quintet) of Cr(II) species 1X and two spin states (singlet and triplet) of Cr(IV) species were considered. The relative free energies for different spin states of chromium ethylene coordination species 1X and chromium metallocyclopentane 2X are summarized in Table S.1 included in the Supporting Information. For ethylene coordination species 1X in Cr(I)/(III) models, sextet 1X were the lowest compared to doublet and quartet spin states, except in model CrOR/DME, where 1D was lowest in energy under a quartet spin state. All metallocyclopentane 2X in Cr(I)/(III) models were lower in energy under a quartet spin state. So for Cr(I)/(III) models the quartet spin state was chosen as the ground spin state in order to avoid a “spin blocking” reaction15 and locate the transition states TS[1X-2X] under the same spin state. Similarly, for Cr(II)/(IV) active site models, the triplet spin state represented the ground spin state for all equilibrium and transition-state geometries, which is consistent with other previous theoretical studies on chromium(II)/(IV)catalyzed ethylene trimerization.5h,18a,19
Results and Discussion In this work, the catalytic cycle was designated on the basis of the metallocycle mechanism proposed by Briggs (Figure 2).4 As illustrated in Scheme 1, X signifies the labeling of 10 plausible molecular models (X = A, Cr(I)þ; X = B, Cr(I)þ/ DME; X = C, Cr(I)OR; X = D, Cr(I)OR/DME; X = E, Cr(I)R; X = F, Cr(I)R/DME; X = G, Cr(II)þOR; X = H, Cr(II)þOR/DME; X = I, Cr(II)þR; X = J, Cr(II)þR/DME; R refers to the isobutyl ligand derived from the PIBAO cocatalyst) including four neutral and six cationic active sites for the active species of the Cr(EH)3/PIBAO catalyst system with or without DME coordination. In general, the starting structure compound 1X possesses two ethylene molecules coordinating to the chromium center. The five-membered
1592
Organometallics, Vol. 29, No. 7, 2010
Qi et al.
Figure 3. Gibbs free energy surface of model Cr(I)þ at the B3LYP level at 298.15 K. The solid line in black shows the metallocycle growth pathway; the dotted line in gray shows the 1-butene elimination pathway; the solid line in gray shows the β-hydrogen transfer pathway to give chromium hydride species 7A; the dotted line in black shows the β-agostic hydrogen shift pathway affording 1-hexene. Energy differences (kcal/mol) are expressed with respect to 1A corrected for the corresponding number of ethylene molecules. Energy barriers are indicated in italics.
ring 2X is formed from 1X through an oxidative coupling reaction. Then ring-opening of 2X can occur to give 4X with a coordinated 1-butene through two routes: (1) Two-step path: β-hydrogen transfer in 2X gives rise to a hydride compound 3X, and then 1-butene is yielded via reductive elimination to give 4X; (2) One-step path: 1-butene is liberated directly via an agostic-assisted β-hydride shift to give 4X. Alternatively, a third ethylene can coordinate to the chromium center to yield 5X, from which chromium sevenmembered ring 6X is formed through subsequent insertion of the third ethylene. Similar to the five-membered ring, 6X can give rise to 8X with a coordinated 1-hexene by either the twostep path with a hydride compound 7X as an intermediate or the one-step path of agostic-assisted β-hydrogen transfer. Otherwise the reaction can proceed by coordination (from 6X to 9X) and insertion of a fourth ethylene molecule, yielding chromium nine-membered ring 10X. In this work, the formation of nine-membered ring 10X represents a metallocycle growth pathway leading to polymerization, because metallocycle expansion can theoretically continue likewise until a β-hydrogen atom is transferred from one of the two β-methylene groups to the opposite R0 -carbon, giving a polymer chain.32 So the two competing reactions ethylene polymerization and ethylene trimerization are represented by the formation of nine-membered ring 10X and 1-hexene liberation through 8X, respectively. In addition, the CosseeArlman mechanism might also be operative to yield polymers. Chromium hydrides 3X and 7X can be Cosseetype active species responsible for polymerization, because the ethylene molecule can insert into the Cr-H or Cr-C bond and chain propagation can take place. The possibility (32) (a) Groppo, E.; Lamberti, C.; Bordiga, S.; Spoto, G.; Zecchina, A. Chem. Rev. 2005, 105, 115. (b) Tomov, A. K.; Chirinos, J. J.; Jones, D. J.; Long, R. J.; Gibson, V. C. J. Am. Chem. Soc. 2005, 127, 10166.
of this type of reaction pathway for polymerization will be considered systematically in our future work. On the basis of the proposed catalytic cycle shown in Scheme 1, DFT calculations on each active site model (10 models from A to J) have been performed, and the results will be discussed in the following sections. It is worth mentioning that TS[2X-4X] are not shown on the free energy surfaces of each model because the transition states of 1-butene elimination via agostic-assisted β-hydrogen transfer could not be located, except that in model Cr(I)R/DME, with an energy barrier as high as 58.9 kcal/mol. Also for the second step in the two-step path of 1-hexene liberation TS[7X-8X] could only be located for model Cr(I)þ/DME and Cr(I)OR and are not shown on the free energy surfaces of the other cases. Model A (Cr(I)þ) and Model B (Cr(I)þ/DME). Calculation results of model A, a simple monovalent cationic species Cr(I)þ, are shown in Figure 3. Compound 1A was chosen as reference. Energy differences of the other compounds were related to 1A and corrected with the corresponding number of ethylene molecules. The five-membered ring 2A was yielded from the initial structure 1A, passing an energy barrier of 4.4 kcal/mol (TS[1A-2A]). β-Hydrogen transfer in 2A had an energy barrier of 14.7 kcal/mol (TS[2A-3A]). It is more reasonable to regard the energy barrier of ringopening of 2A as the difference between TS[2A-3A] and 5A, 31.9 kcal/mol, because 2A could readily fall to 5A through the exoergic coordination of a third ethylene. Therefore, instead of 1-butene elimination, the seven-membered ring 6A was formed with a lower energy barrier of 15.4 kcal/ mol (TS[5A-6A]). The coordination of a fourth ethylene occurred with an obvious exoergic effect of 12.2 kcal/mol and thus immediately led to the reaction from 6A to the formation of 9A. The chromium hydride 7A could be yielded by β-hydrogen transfer in the seven-membered ring 6A, and
Article
Organometallics, Vol. 29, No. 7, 2010
1593
Figure 4. Gibbs free energy surface of model Cr(I)þ/DME at the B3LYP level at 298.15 K. The solid line in black shows the metallocycle growth pathway; the dotted line in gray shows the 1-butene elimination pathway; the solid line in gray shows the 1-hexene elimination pathway via a two-step route; the dotted line in black shows the 1-hexene elimination pathway via β-agostic hydrogen shift. Energy differences (kcal/mol) are expressed with respect to 1B corrected for the corresponding number of ethylene molecules. Energy barriers are indicated in italics, and heat absorption is shown in parentheses.
the energy barrier should be 23.9 kcal/mol, which was the difference between TS[6A-7A] and 9A, rather than 11.7 kcal/mol (TS[6A-7A]). The chromium seven-membered ring 6A could undergo agostic-assisted β-hydrogen shift (TS[6A-8A]) to give 8A with a coordinated 1-hexene. Similarly, the energy barrier of 1-hexene liberation through 8A was the difference between TS[6A-8A] and 9A, 27.7 kcal/ mol, which was higher than that required to form a ninemembered ring 10A (16.4 kcal/mol, TS[9A-10A]). Then the reaction should proceed to metallocycle growth leading to polymerization. Therefore, site A Cr(I)þ without DME coordination preferred ethylene polymerization rather than ethylene trimerization. Calculation results of model B, Cr(I)þ/DME, are shown in Figure 4. The five-membered ring 2B was yielded from structure 1B overcoming an energy barrier of 10.3 kcal/mol (TS[1B-2B]). In preference to liberation of 1-butene from 2B via a two-step path with energy barriers of 14.3 kcal/mol (TS[2B-3B]) to give chromium hydride 3B and 12.3 kcal/ mol (TS[3B-4B]) to give 4B with a coordinated 1-butene in each step, a seven-membered ring 6B was formed by the coordination (2Bf5B) and subsequent insertion of a third ethylene with an energy barrier of 9.0 kcal/mol (TS[5B-6B]) and exoergic effect. 1-Hexene (8B) was liberated from 6B under two routes: (1) Two-step path: β-hydrogen transfer to give chromium hydride 7B with an energy barrier of 13.9 kcal/mol (TS[6B-7B]) followed by reductive elimination of 1-hexene with an energy barrier of 15.2 kcal/mol (TS[7B-8B]); or (2) one-step path: direct reductive liberation via agostic-assisted hydride shift with an energy barrier of 16.7 kcal/mol (TS[6B-8B]). Because the total energy barrier of 1-hexene elimination via the two-step mechanism should be regarded as 25.2 kcal/mol (energy difference from 6B to TS[7B-8B]), 1-hexene liberation should follow a one-step path of agostic-assisted β-hydrogen transfer rather than a two-step path. The coordination of another
ethylene increased the energy of 9B by 8.7 kcal/mol (shown in parentheses), and the nine-membered ring 10B was exoergically formed from 9B with a moderate energy barrier of 10.5 kcal/mol (TS[9B-10B]). In DFT studies on tantalum-,16 titanium-,17b,23 and chromium18c-catalyzed selective ethylene oligermerization the sum of the binding energy of ethylene coordination (with an energy increase after coordination) and energy barrier of the insertion step to give a large ring was regarded as the total energy barrier for metallocycle expansion. So the total energy barrier of ninemembered ring 10B formation should be 19.2 kcal/mol, which was the sum of 8.7 kcal/mol (6B f 9B) and 10.5 kcal/ mol (TS[9B-10B]). So metallocycle growth was disfavored and 1-hexene was liberated via agostic-assisted β-hydride shift. Therefore, site B Cr(I)þ/DME with DME coordination preferred ethylene trimerization rather than ethylene polymerization. The above calculation results on site A Cr(I)þ and site B Cr(I)þ/DME are consistent with experimental evidence in terms of active site transformation from ethylene polymerization to ethylene trimerization on the Cr(EH)3/PIBAO catalyst system by addition of DME. Therefore, the monovalent cationic species site A Cr(I)þ and site B Cr(I)þ/DME could be plausible active sites of the Cr(EH)3/PIBAO catalyst system with and without DME coordination. Model C (Cr(I)OR) and Model D (Cr(I)OR/DME). Calculation results of model C, Cr(I)OR (R refers to the isobutyl ligand derived from the PIBAO cocatalyst), are shown in Figure S.1 in the Supporting Information. The five-membered ring 2C was readily formed from structure 1C with a low energy barrier of 6.1 kcal/mol (TS[1C-2C]). 1-Butene liberation through 4C via a two-step mechanism was disfavored because of the high energy barriers of 22.6 kcal/mol (TS[2C-3C]) and 23.6 kcal/mol (TS[3C-4C]) in each step. Structure 5C was yielded from the five-membered ring 2C by the coordination of a third ethylene to the chromium center.
1594
Organometallics, Vol. 29, No. 7, 2010
Then the seven-membered ring 6C was formed and the energy barrier was 13.2 kcal/mol (TS[5C-6C]). β-Hydrogen transfer in 6C had an energy barrier of 19.3 kcal/mol (TS[6C-7C]) to give a chromium hydride 7C, and subsequent elimination of 1-hexene from 7C had a higher energy barrier of 26.2 kcal/mol (TS[7C-8C]), whereas 1-hexene liberation through 8C via agostic-assisted hydrogen transfer had a much lower energy barrier of 18.2 kcal/mol (TS[6C-8C]). As a competing reaction, the nine-membered ring 10C could be formed through the coordination and insertion of a fourth ethylene with a total energy barrier of 24.1 kcal/mol by adding 5.7 kcal/mol (6C f 9C, shown in parentheses) and 18.4 kcal/mol (TS[9C-10C]). 1-Hexene liberation via agostic-assisted β-hydrogen shift in sevenmembered ring 6C had the lowest energy barrier (18.2 kcal/ mol, TS[6C-8C]); therefore, site C Cr(I)OR without DME coordination preferred ethylene trimerization rather than ethylene polymerization. Calculation results of model D, Cr(I)OR/DME (R refers to the isobutyl ligand derived from the PIBAO cocatalyst), are shown in Figure S.2 in the Supporting Information. The chromium metallocyclopentane 2D was formed from structure 1D overcoming an energy barrier of 10.2 kcal/mol (TS[1D-2D]). Ring-opening reaction of 2D via β-hydrogen transfer had an energy barrier of 29.1 kcal/mol (TS[2D-3D]), and the following reductive elemination of 1-butene had an energy barrier of 28.7 kcal/mol (TS[3D-4D]). Compared to the results obtained from model C Cr(I)OR, the energy required for extra ethylene coordination to give structures 5D and 9D was significantly higher at 24.4 kcal/mol (2D f 5D) and 29.1 kcal/mol (6D f 9D, shown in parentheses), respectively, while the energy barriers of the two corresponding ethylene insertion steps dramatically decreased to 5.2 kcal/mol (TS[5D-6D]) and 7.4 kcal/mol (TS[9D-10D]), respectively. These changes indicated that metallocycle expansion on model Cr(I)OR/DME was more likely to be controlled by the ethylene coordination step rather than the ethylene insertion step. The total energy barrier for the formation of seven-membered ring 6D from 2D was 29.6 kcal/mol, which was a little bit higher than the energy barrier of β-hydrogen transfer in the metallocyclopentane 2D (29.1 kcal/mol, TS[2D-3D]) (ethylene dimerization seemed to occur), or if the total energy barrier of 1-butene elimination was regarded as 46.7 kcal/mol (from 2D to TS[3D-4D]), the reaction should continue with metallocycle growth into seven-membered ring 6D but with low selectivity. The formation of nine-membered ring 10D from 6D was disfavored because of a total energy barrier as high as 36.5 kcal/mol, which was the sum of 29.1 kcal/mol (6D f 9D, shown in parentheses) and 7.4 kcal/mol (TS[9D-10D]). Then ringopening of 6D could occur through two routes: (1) β-hydrogen shift in 6D giving chromium hydride 7D with an energy barrier of 26.3 kcal/mol (TS[6D-7D]) and an endoergic effect of 17.4 kcal/mol; (2) one-step path: agostic-assisted βhydride transfer to liberate 1-hexene through 8D overcoming a little higher energy barrier of 29.8 kcal/mol (TS[6D-8D]) but possessing an exoergic effect of 5 kcal/mol. So 1-hexene elimination occurred via agostic-assisted β-hydrogen transfer. Therefore, site D Cr(I)OR/DME with DME coordination preferred ethylene trimerization rather than ethylene polymerization. The above calculation results on site C Cr(I)OR and site D Cr(I)OR/DME are not consistent with experimental evidence in terms of site transformation from ethylene
Qi et al.
polymerization into ethylene trimerization on the Cr(EH)3/ PIBAO catalyst system by addition of DME. Therefore, the monovalent neutral species site C Cr(I)OR and site D Cr(I)OR/DME could not be active sites of the Cr(EH)3/PIBAO catalyst with and without DME coordination. Model E (Cr(I)R) and Model F (Cr(I)R/DME). Calculation results of model E, Cr(I)R (R refers to the isobutyl ligand derived from the PIBAO cocatalyst), are shown in Figure S.3 in the Supporting Information. The reaction started with 1E, and the five-membered ring 2E was readily formed with an energy barrier of 6.5 kcal/mol (TS[1E-2E]). Here ring-opening of 2E via the two-step route was disfavored because energy barriers were as high as 28.9 kcal/mol (TS[2E-3E]) and 30.4 kcal/mol (TS[3E-4E]) in each step. Therefore a third ethylene coordinated to the chromium center to give 5E, and the energy was increased by 4.3 kcal/ mol (shown in parentheses). Subsequently, the seven-membered ring 6E was exoergically yielded from 5E with an energy barrier of 20.0 kcal/mol (TS[5E-6E]). The formation of the nine-membered ring 10E was disfavored, because its total energy barrier (30.6 kcal/mol), which was the sum of 7.8 kcal/mol (6E f 9E, shown in parentheses) and 22.8 kcal/ mol (TS[9E-10E]), was higher than the energy barrier of ring-opening of 6E to give 8E with a coordinated 1-hexene via agostic-assisted β-hydrogen transfer (29.8 kcal/mol, TS[6E-8E]). An alternative path of ring-opening of 6E via β-hydrogen transfer had a little bit lower energy barrier of 28.4 kcal/mol (TS[6E-7E]) but presented an endoergic effect of 16 kcal/mol to give chromium hydride 7E. Therefore, 1-hexene liberation occurred via agostic-assisted β-hydrogen transfer, and site E Cr(I)R without DME coordination preferred ethylene trimerization rather than ethylene polymerization. Calculation results of model F, Cr(I)R/DME (R refers to the isobutyl ligand derived from the PIBAO cocatalyst), are shown in Figure S.4 in the Supporting Information. The starting structure 1F was transformed into the five-membered ring 2F overcoming an energy barrier of 17.7 kcal/mol (TS[1F-2F]). The energy barriers required for β-hydrogen transfer in 2F to give chromium hydride 3F and reductive elimination of 1-butene through 4F were 31.7 kcal/mol (TS[2F-3F]) and 28.8 kcal/mol (TS[3F-4F]), respectively. The total energy barrier for the formation of sevenmembered ring 6F was 33.5 kcal/mol by adding 29.4 kcal/ mol (2F f 5F, shown in parentheses) and 4.1 kcal/mol (TS[5F-6F]), which was higher than ring-opening of 2F to give chromium hydride 3F via β-hydrogen transfer (31.7 kcal/mol, TS[2F-3F]). Ethylene dimerization seemed to be favored over the formation of metallocycloheptane 6F. Whereas the total energy barrier height of 1-butene elimination via a two-step mechanism could be regarded as 45.8 kcal/mol (from 2F to TS[3F-4F]), metallocycle growth into the seven-membered ring 6F should take place with low selectivity. The formation of nine-membered ring 10F from 6F had a total energy barrier as high as 40.6 kcal/mol by adding 34.9 kcal/mol (6F f 9F, shown in parentheses) and 5.7 kcal/mol (TS[9F-10F]). Consequently, ring-opening of 6F occurred. The energy barrier of β-hydrogen transfer (26.9 kcal/mol, TS[6F-7F]) to give the chromium hydride 7F was lower than that of agostic-assisted β-hydrogen transfer (32.8 kcal/mol, TS[6F-8F]) to give 8F with a coordinated 1-hexene, but the formation of 7F had an endoergic effect of 17.5 kcal/mol, whereas the formation of 8F had an exoergic effect of 4.5 kcal/mol; thus 1-hexene was liberated via
Article
Organometallics, Vol. 29, No. 7, 2010
1595
Figure 5. Gibbs free energy surface of model Cr(II)þOR at the B3LYP level at 298.15 K. The solid line in black shows the metallocycle growth pathway; the dotted line in gray shows the β-hydrogen transfer pathway of 2G to give chromium hydride 3G; the solid line in gray shows β-hydrogen transfer of 6G to give chromium hydride 7G; the dotted line in black shows the β-agostic hydrogen shift pathway affording 1-hexene. Energy differences (kcal/mol) are expressed with respect to 1G corrected for the corresponding number of ethylene molecules. Energy barriers are indicated in italics.
agostic-assisted β-hydrogen transfer. Therefore, site F Cr(I)R/DME with DME coordination preferred ethylene trimerization rather than ethylene polymerization. The above calculation results on site E Cr(I)R and site F Cr(I)R/DME are not consistent with experimental evidence in terms of site transformation from ethylene polymerization into ethylene trimerization on the Cr(EH)3/PIBAO catalyst system by addition of DME. Therefore, the monovalent neutral species site E Cr(I)R and site F Cr(I)R/DME could not be active sites of the Cr(EH)3/PIBAO catalyst with and without DME coordination. Model G (Cr(II)þOR) and Model H (Cr(II)þOR/DME). Calculation results of model G, Cr(II)þOR (R refers to the isobutyl ligand derived from the PIBAO cocatalyst), are shown in Figure 5. The formation of five-membered ring 2G from structure 1G required an energy barrier of 15.0 kcal/ mol (TS[1G-2G]). 2G immediately accepted the coordination of a third ethylene to give a stable structure 5G rather than yielding chromium hydride 3G via β-hydrogen transfer with an energy barrier of 13.5 kcal/mol (TS[2G-3G]). The seven-membered ring 6G was formed from 5G with an energy barrier of 12.5 kcal/mol (TS[5G-6G]). Then 6G readily fell to 9G through the coordination of a fourth ethylene with an exoergic effect of 8.3 kcal/mol (6G f 9G). As mentioned before, the energy barrier of β-hydrogen transfer in 6G should be regarded as the difference between TS[6G-7G] and 9G, 19.1 kcal/mol, and the energy barrier of agostic-assisted β-hydrogen transfer in 6G to give 8G with a coordinated 1-hexene should be the difference between TS[6G-8G] and 9G, 18.9 kcal/mol. The insertion of the fourth ethylene to form nine-membered ring 10G required a lower energy barrier of 15.5 kcal/mol (TS[9G-10G]), so ringopening of the seven-membered ring 6G was disfavored and the reaction proceeded with metallocycle growth. Therefore, site G Cr(II)þOR without DME coordination preferred ethylene polymerization rather than ethylene trimerization.
Calculation results of model H, Cr(II)þOR/DME, are shown in Figure 6. From 1H the five-membered ring 2H was yielded with an energy barrier of 12.6 kcal/ mol (TS[1H-2H]). An energy barrier of 33.8 kcal/mol (TS[2H-3H]) was required to give chromium hydride 3H via β-hydrogen transfer in 2H, whereas the total energy barrier for the formation of the seven-membered ring 6H was 31.1 kcal/mol, which was obtained by adding the coordination energy of a third ethylene (20.9 kcal/mol, 2H f 5H, shown in parentheses) and insertion energy barrier (10.2 kcal/mol, TS[5H-6H]). So the formation of the sevenmembered ring 6H was favored over ring-opening of 2H. The coordination of a fourth ethylene required energy of 24.8 kcal/mol (6H f 9H, shown in parentheses), and the subsequent insertion step had an energy barrier of 11.6 kcal/ mol (TS[9H-10H]), so the total energy barrier required for the formation of the nine-membered ring 10H was 36.4 kcal/ mol, which was much higher than the energy barrier of 1-hexene liberation through agostic-assisted β-hydrogen shift (13.4 kcal/mol, TS[6H-8H]). Every attempt to locate the transition state of β-hydrogen transfer (TS[6H-7H]) in the seven-membered ring 6H failed. Site H Cr(II)þOR/DME with DME coordination preferred ethylene trimerization rather than ethylene polymerization. The above calculation results on site G Cr(II)þOR and site H Cr(II)þOR/DME are consistent with experimental evidence in terms of site transformation from ethylene polymerization into ethylene trimerization on the Cr(EH)3/PIBAO catalyst system by adding DME. Therefore, the divalent cationic species site G Cr(II)þOR and site H Cr(II)þOR/ DME could be plausible active sites of the Cr(EH)3/PIBAO catalyst system with and without DME coordination. Model I (Cr(II)þR) and Model J (Cr(II)þR/DME). Calculation results of model I, Cr(II)þR (R refers to the isobutyl ligand derived from the PIBAO cocatalyst), are shown in Figure 7. The starting structure 1I was transformed into the
1596
Organometallics, Vol. 29, No. 7, 2010
Qi et al.
Figure 6. Gibbs free energy surface of Cr(II)þOR/DME at the B3LYP level at 298.15 K. The solid line in black shows the metallocycle growth pathway; the dotted line in gray shows the 1-butene elimination pathway; the dotted line in black shows the β-agostic hydrogen shift pathway affording 1-hexene. Energy differences (kcal/mol) are expressed with respect to 1H corrected for the corresponding number of ethylene molecules. Energy barriers are indicated in italics, and heat absorption is shown in parentheses.
Figure 7. Gibbs free energy surface of model Cr(II)þR at the B3LYP level at 298.15 K. The solid line in black shows the metallocycle growth pathway; the dotted line in gray shows the β-hydrogen transfer pathway of 2I to give chromium hydride 3I; the solid line in gray shows the β-hydrogen transfer pathway of 6I to give chromium hydride 7I; the dotted line in black shows the β-agostic hydrogen shift pathway affording 1-hexene. Energy differences (kcal/mol) are expressed with respect to 1I corrected for the corresponding number of ethylene molecules. Energy barriers are indicated in italics.
five-membered ring 2I overcoming an energy barrier of 10.5 kcal/mol (TS[1I-2I]). The formation of a chromium hydride species 3I via β-hydrogen transfer in 2I had to conquer an energy barrier of 20.5 kcal/mol (TS[2I-3I]). Ring-opening of 2I was obviously disfavored, because a third ethylene could coordinate to the chromium center to give 5I with an exoergic effect of 5.5 kcal/mol, and the subsequent insertion of the coordinated ethylene to yield the seven-membered ring 6I had a lower energy barrier of
10.7 kcal/mol (TS[5I-6I]). The coordination of a fourth ethylene barely decreased the energy of 9I compared with 6I. Metallocycle expansion from 9I into the nine-membered ring 10I had an energy barrier of 13.8 kcal/mol (TS[9I-10I]). Ring-opening of 6I required a higher energy barrier of 18.1 kcal/mol (TS[6I-7I]) to give chromium hydride 7I via β-hydrogen transfer or 14.3 kcal/mol (TS[6I-8I]) to yield 8I with a coordinated 1-hexene via agostic-assisted β-hydrogen shift. So 1-hexene was liberated
Article
Organometallics, Vol. 29, No. 7, 2010
1597
Figure 8. Gibbs free energy surface of Cr(II)þR/DME at the B3LYP level at 298.15 K. The solid line in black shows the metallocycle growth pathway; the dotted line in gray shows the 1-butene elimination pathway; the dotted line in black shows the β-agostic hydrogen shift pathway affording 1-hexene. Energy differences (kcal/mol) are expressed with respect to 1J corrected for the corresponding number of ethylene molecules. Energy barriers are indicated in italics, and heat absorption is shown in parentheses.
via a one-step path overcoming a total energy barrier of 14.9 kcal/mol, which was the difference between TS[6I-8I] and 9I. Therefore, site I Cr(II)þR without DME coordination preferred ethylene polymerization rather than ethylene trimerization. Calculation results of model J, Cr(II)þR/DME, are shown in Figure 8. The starting structure 1J was transformed into the five-membered ring 2J with a moderate energy barrier of 11.6 kcal/mol (TS[1J-2J]). Ring-opening of 2J was again disfavored because of the high energy barrier of 33.1 kcal/ mol via β-hydrogen transfer (TS[2J-3J]). The coordination of a third ethylene yielded 5J with an endoergic effect of 17.2 kcal/mol (2J f 5J, shown in parentheses). Then the seven-membered ring 6J was formed from 5J with an energy barrier of 10.6 kcal/mol (TS[5J-6J]) and an exoergic effect of 23.3 kcal/mol. Metallocycle expansion into the ninemembered ring 10J from 6J had a total energy barrier of 32.7 kcal/mol obtained by adding heat absorption required for coordination of extra ethylene (20.9 kcal/mol, 6J f 9J, shown in parentheses) and the energy barrier of ethylene insertion (11.8 kcal/mol, TS[9J-10J]), whereas 1-hexene liberation via agostic-assisted β-hydrogen transfer in 6J had a lower energy barrier of 17.2 kcal/mol (TS[6J-8J]) and an exoergic effect of 18.0 kcal/mol. Every attempt to locate the transition state of -hydrogen transfer (TS[6J-7J]) in seven-membered ring 6J failed. So the reaction definitely led to ethylene trimerization through agostic-assisted β-hydrogen transfer. Therefore, site J Cr(II)þR/DME with DME coordination preferred ethylene trimerization rather than ethylene polymerization. The above calculation results on site I Cr(II)þR and site J Cr(II)þR/DME are consistent with experimental evidence in terms of site transformation from ethylene polymerization into ethylene trimerization on Cr(EH)3/PIBAO catalyst system by addition of DME. Therefore, the divalent cationic species site I Cr(II)þR and site J Cr(II)þR/DME could be plausible active sites of the Cr(EH)3/PIBAO catalyst system with and without DME coordination.
Figure 9. Illustration of prerequisite transition-state structures (left: TS[2X-4X]; right: TS[6X-8X]) during agostic-assisted βhydrogen transfer to give 1-butene (left) and 1-hexene (right), respectively.
Two-Step or One-Step Mechanism. From the obtained calculation results, it is found that 1-butene elimination tended to follow the two-step path with chromium hydride 3X (X = A-J) as an intermediate, whereas 1-hexene liberation preferred to adopt a one-step path via agostic-assisted βhydrogen transfer. This phenomenon is consistent with several previous DFT reports.16,17b As shown in Figure 9, the chromium atom, the shifted β-hydrogen, the β-carbon, and the opposite R0 -carbon need to be coplanar during agostic-assisted β-hydrogen transfer, and also the two carbon atoms involved need to be close enough to facilitate the transfer of the β-hydrogen atom. However, it is hard for fivemembered ring 2X (X = A-J) to form the prerequisite conformation due to ring strain in the transition structure TS[2X-4X]. Although the energy barrier of 1-butene elimination via agostic-assisted β-hydrogen transfer (TS[2F-4F]) in model Cr(I)R/DME was found to be remarkably as high as 58.9 kcal/mol (see Supporting Information), every attempt to locate transition states of TS[2X-4X] in all the other models was not successful. On the other hand, 1-hexene elimination from seven-membered ring 6X (X = A-J) preferred the onestep path of agostic-assisted β-hydrogen transfer rather than the two-step path. One possible explanation is discussed as follows. First, the formation of an intermediate chromium hydride 7X (X = A-G and I) via β-hydrogen transfer in a
1598
Organometallics, Vol. 29, No. 7, 2010
Figure 10. Mulliken charge on the chromium center (shown in parentheses) and transition-state structures (atomic distances in A˚) of 1-hexene liberation via agostic-assisted β-hydrogen transfer (TS[6X-8X]) and reductive elimination of 1-hexene from chromium hydride (TS[7X-8X]) for models Cr(I)þ/DME (X = B) and Cr(I)OR (X = C).
seven-membered ring 6X (X = A-G and I) was accompanied by a significant endoergic effect, so 7X was thermodynamically not stable in energy (see Figures 3-5, 7, and S.1 to S.4). For model H Cr(II)þOR/DME and J Cr(II)þR/DME, the transition-state structures of TS[6X-7X] (X = H, J) were not obtained by any means; second, the subsequent reductive elimination giving 8X with a coordinated 1-hexene from chromium hydride 7X (TS[7X-8X]) was considered to be difficult to take place.16 Only in the two models B Cr(I)þ/ DME and C Cr(I)OR were the transition states TS[7X-8X] (X = B, C) successfully located and the energy barriers were higher than that of 1-hexene liberation via agostic-assisted βhydrogen transfer, TS[6X-8X] (X = B, see Figure 4; X = C, see Figure S.1). The origin of the difference between the two specific transition states giving 1-hexene could be found through the analysis of electronic density at the chromium center. Figure 10 illustrated the Mulliken charge on the chromium center and the structures of TS[6X-8X] and TS[7X-8X] (X = B, Cr(I)þ/DME; X = C, Cr(I)OR). For model B Cr(I)þ/DME, the Mulliken charge on the Cr center was þ0.336 in TS[7B-8B], lower than the þ0.556 in TS[6B-8B], indicative of much higher electron density at the Cr center in TS[7B-8B]. The atomic distances between Cr and C of CdC bond were 2.378 and 2.473 A˚ in TS[7B-8B], which were longer than those of 2.091 and 2.246 A˚ in TS[6B-8B], indicating that the coordination between CdC and the Cr center is much weaker in TS[7B-8B], and thus TS[7B-8B] is not as stable as TS[6B-8B]. The Cr-H distance was 1.620 A˚ in TS[7B-8B] and is much shorter than that of 1.641 A˚ in TS[6B-8B], which indicates that much easier hydrogen transfer to give 1-hexene could be expected in TS[6B-8B]. A similar relation between Mulliken charge (electron density) and atomic distances (stability of the transition state) could also be found for model C Cr(I)OR as shown in Figure 10. Therefore, the higher energy barrier of 1-hexene reductive elimination from chromium hydride (TS[7X-8X]) could result from the higher electron density on the Cr center in TS[7X-8X] than that in TS[6X-8X] (X = B, Cr(I)þ/DME; X = C, Cr(I)OR). In summary, the instability of the intermediate 7X and the much higher energy barrier of the following reductive elimination of 1-hexene made the two-step path
Qi et al.
unfavorable. So 1-hexene liberation followed one-step path via agostic-assisted β-hydrogen transfer (TS[6X-8X]). Cationic versus Neutral Active Site Models. As reported in previous studies, ethylene polymerized into polyethylene in the absence of DME3 and trimerized into 1-hexene in the presence of DME4 on the Cr(EH)3/PIBAO catalyst system. By comparing the energy barriers for 1-hexene liberation via agostic-assisted β-hydrogen transfer and nine-memberedring formation, which were two key competing reactions after the seven-membered ring was formed, we found that cationic active site models Cr(I)þ, Cr(II)þOR, and Cr(II)þR provided results of metallocycle growth leading to polymerization before adding DME and trimerization into 1-hexene after adding DME with high activity and selectivity, which agreed well with the experimental results. However, calculation results of neutral active site models were not consistent with experimental evidence. First, the neutral models without DME coordination (model C Cr(I)OR and E Cr(I)R) could not conduct the expected ethylene polymerization through metallocycle growth but trimerization, giving 1hexene, and the total energy barriers for metallocycle expansion from smaller ring size into larger ring size via a metallocycle mechanism were lower in cationic active site models (see Figures 3, 5, and 7) than those in neutral active site models (see Figures S.1 and S.3). Second, for neutral active site models with DME coordination (model D Cr(I)OR/ DME and F Cr(I)R/DME), theoretical results showed that the energy barriers for ring-opening of metallocyclopentane via β-hydrogen transfer leading to 1-butene elimination were comparable to the total energy barriers for the formation of seven-membered rings. So neutral active site models Cr(I)OR/DME and Cr(I)R/DME failed to give results of selective ethylene trimerization as expected due to the considerably high energy barriers of metallocycle expansion from five-membered rings into seven-membered rings. Similarly, Agapie et al.9b proposed that further ethylene insertion could not occur after a neutral metallocyclopentane was formed. Bhaduri et al.18b reported similar findings by the DFT method applying both metallocycle and Cossee-type mechanisms in neutral and cationic models for chromiumbased ethylene trimerization catalyst and showed that the metallocycle mechanism was not feasible with neutral species. Agapie et al.9b also conducted reactions of ethylene and 2-butyne with the cationic chromium species generated from a well-defined chromium precursor and detected the formation of 1-hexene and hexamethylbenzene, respectively, but when using its neutral chromium species, the expected trimers were not detected. So cationic species rather than neutral species are very important for highly selective ethylene trimerization via the metallocycle mechanism.9b The Role of DME. 1. Electronic Effect of DME. On the basis of the particular atomic distances obtained for the five models with DME coordination, it was noted that the geometric structures of the coordinated DME were different between neutral and cationic models. Compared with neutral models, the DME ligand in cationic models presented lengthened C-O distances (see Table S.2 in the Supporting Information) and shortened C-C bond distances (Table S.3), and also DME in cationic models had a stronger interaction with the chromium center because the distances of chromium from O atoms in DME were shorter (Table S.4). These changes in structural features of DME were consistent with that of diimine (Ar-Nd CH-CHdN-Ar, Ar = 2,6-diisopropylphenyl) reported
Article
Organometallics, Vol. 29, No. 7, 2010
1599
Table 2. Differences in Mulliken Charge on Chromium Center before and after Adding DME in Each Model cationic models a
q(A-B, Cr) 5X 6X 9X 10X
-0.030 0.120 -0.103 -0.086
neutral models
q(G-H, Cr) q(I-J, Cr) q(C-D, Cr) q(E-F, Cr) 0.052 -0.083 0.058 -0.011
0.055 -0.191 -0.042 -0.149
0.096 0.011 0.128 0.006
0.181 0.003 0.213 0.011
Italics representing that the Mulliken charge on the chromium center was increased after adding DME. a q(A-B, Cr) = q(A, Cr) - q(B, Cr), difference in Mulliken charge on the chromium center between model A Crþ and model B Crþ/DME.
Table 3. Differences in Mulliken Charge on Other Ligands Ln before and after Adding DME in Each Model cationic models
neutral models
q(A-B, Ln) q(G-H, Ln) q(I-J, Ln) q(C-D, Ln) q(E-F, Ln)
Figure 11. Illustration of structural changes of DME ligands (taking 1D in model Cr(I)OR/DME and 1H in model Cr(II)þOR/DME as examples, distances in A˚) in comparison with Cr-based systems coordinated with diimine reported in the literature33 (Ln, Lx, Ly, Lz = other ligands coordinated to Cr center, Ar = 2,6-diisopropylphenyl).
by Kreisel et al.33 (shown in Figure 11). The authors prepared and characterized several chromium complexes coordinated by diimine and found that the oxidation state of the chromium center increased with the lengthening in C-N bond distances and shortening in C-C bond length and the distances of Cr from N atoms. According to the authors,33 the coordinated diimines could allow for redistribution of electron density between the Cr center and the diimine ligand depending on the nature of the ligand field and the total charge of the complex. With this idea in mind, a detailed analysis of Mulliken charge was provided as follows to explore the electronic effect of DME. In order to clearly compare the Mulliken charge on the chromium center and DME, the remaining part of each stationary point except the Cr atom and DME was denoted as Ln. So the intermediates could be represented as Cr-Ln in models without DME coordination and DME-Cr-Ln in models after adding DME, respectively. Tables S.5 to S.9 in the Supporting Information summarize the Mulliken charge on each part of the stationary points in each model, named q(X, Cr), q(X, Ln), and q(X, DME) (X = A, Cr(I)þ; B, Cr(I)þ/DME; C, Cr(I)OR; D, Cr(I)OR/DME; E, Cr(I)R; F, Cr(I)R/DME; G, Cr(II)þOR; H, Cr(II)þOR/DME; I, Cr(II)þR; J, Cr(II)þR/ DME). For all models, DME acted as electron donors because all q(X, DME) > 0. In neutral models with DME coordination, þ0.200 < q(X, DME) < þ0.260 (X = D, F); in cationic models with DME coordination, þ0.340 < q(X, DME) < þ0.380 (X = B, H, and J), so the ability of DME to donate electrons to the chromium center in neutral models was not as strong as that in cationic models. That is to say, DME acted as a weak electron donor for neutral models and as a strong electron donor for cationic models. As mentioned above, the total energy barriers of metallocycle expansion in cationic active site models were lower than those in neutral models. The better performance of cationic (33) Kreisel, K. A.; Yap, G. P. A.; Theopold, K. H. Inorg. Chem. 2008, 47, 5293.
5X 6X 9X 10X
0.367 0.235 0.434 0.431
0.298 0.425 0.277 0.361
0.270 0.531 0.357 0.492
0.150 0.191 0.112 0.209
0.076 0.214 0.039 0.197
active species following the metallocycle mechanism than neutral active species is believed to be related to the different electronic effect of DME, especially the effect on 5X, 6X, 9X, and 10X, which were four key intermediates during metallocycle growth. For neutral models, the Mulliken charge on the chromium center of the four intermediates all decreased after DME coordination, q(C-D, Cr) > 0 and q(E-F, Cr) > 0 (see Table 2), because of the weak electron-donating effect of DME. For cationic models, the change in Mulliken charge on the chromium center could be divided into two groups: (1) positive values: Mulliken charge on chromium center was decreased after adding DME but decreased less than that in neutral models, except q(6A-6B, Cr) = 0.120; (2) negative values (shown in italics): Mulliken charge on chromium center was even increased after adding DME. In other words, in cationic models the Mulliken charge on the chromium center for some intermediates was decreased less than that in neutral models by the addition of DME, and for the other intermediates in cationic models their Mulliken charge on the chromium center was even increased instead of decreased. The reason might be that the electron-donating effect of DME in cationic models was so strong that the donated electron was pushed further onto the other ligands (Ln) rather than the chromium center. This explanation could be testified by the change in Mulliken charge on Ln before and after adding DME. As shown in Table 3, the Mulliken charge on Ln for each intermediate in the cationic models was decreased more than that in neutral models by the addition of DME. So DME in neutral models was a weak electron donor and led to a further increase in electron density on the chromium center. However, after adding DME in the cationic models the Mulliken charge on the chromium center was decreased only a little and even increased for most intermediates due to its strong electron-donating effect. 2. Steric Effects of DME. Table 4 lists energy differences of ethylene coordination and insertion during metallocycle expansion and 1-hexene liberation via agostic-assisted β-hydrogen transfer (TS[6X-8X]) in each model. As shown in Table 4, compared to the extent of increase in energy barriers for 1-hexene liberation (TS[6X-8X]) brought by the addition of DME, DME had a much greater effect on
1600
Organometallics, Vol. 29, No. 7, 2010
Qi et al.
Table 4. Energy Differences (kcal/mol) of Crucial Steps in Catalytic Cycle of Each Model Ring5 f Ring7 model cationic
neutral
Crþ Crþ/DME CrþOR CrþOR/DME CrþR CrþR/DME CrOR CrOR/DME CrR CrR/DME
A B G H I J C D E F
Ring7 f Ring9
coorda
insertionb
coordc
insertiond
1-C6e
-17.2 1.2 -15.9 20.9 -5.5 17.2 5.1 24.4 4.3 29.4
15.4 9.0 12.5 10.2 10.7 10.6 13.2 5.2 20.0 4.1
-12.2 8.7 -8.3 24.8 -0.6 20.9 5.7 29.1 7.8 34.9
16.4 10.5 15.5 11.6 13.8 11.8 18.4 7.4 22.8 5.7
15.5 16.7 10.6 13.4 14.3 17.2 18.2 29.8 29.8 32.8
a Heat absorption required for the coordination of a third ethylene molecule, 2X f 5X. b Energy barrier required for the insertion of a third ethylene to form a chromium seven-membered ring, TS[5X-6X]. c Heat absorption required for the coordination of a fourth ethylene molecule, 6X f 9X. d Energy barrier required for the insertion of a fourth ethylene to form a chromium nine-membered ring, TS[9X-10X]. e Energy barrier of 1-hexene liberation via agostic-assisted β-hydrogen transfer, TS[6X-8X].
Table 5. Atomic Distances (A˚) in Structure 9X for Each Model Cr-R or Cr-ORa
model cationic A B G H I J neutral C D E F
Crþ Crþ/DME CrþOR CrþOR/DME CrþR CrþR/DME CrOR CrOR/DME CrR CrR/DME
d d
1.722 1.825 2.023 2.594 1.800 1.946 2.042 2.243
Cr-CRb 2.010 2.029 1.977 2.050 1.976 2.056 2.029 2.063 2.021 2.070
2.023 2.041 1.980 2.703 2.003 2.099 2.025 2.137 2.022 2.174
Cr-Cethylenec 2.385 2.555 2.463 2.555 2.490 2.609 2.588 2.530 2.787 2.435
2.498 2.719 2.512 2.606 2.504 2.728 2.853 2.542 3.137 2.455
a Bond length of chromium atom and R-carbon in R group or oxygen atom in OR group. b Bond length of chromium atom and R-carbon atoms of the ring part. c Distance of chromium atom from carbon atoms in the coordinated ethylene. d No relevant structure.
metallocycle expansion, Ring5 f Ring7 and Ring7 f Ring9. Before adding DME, the coordination of extra ethylene molecules occurred spontaneously with exoergic effect (cationic models: X = A, G, and I) or with a very low endoergic effect, which could be ignored (neutral models: X = C and E), so the formation of metallocycle was determined by the ethylene insertion step. After adding DME, the energy barriers of ethylene insertion (TS[5X-6X] and TS[9X-10X], X = B, D, F, H, and J) all decreased, while the energy required for ethylene coordination (heat absorption of 2X f 5X and 6X f 9X, X = B, D, F, H, and J) increased more than 18 kcal/mol, which was the main disadvantage during metallocycle expansion to larger ring size. Taking the formation of the nine-membered ring as an example, the remarkable changes in energy could be attributed to the structural changes of 9X after adding DME, which is a Πcomplex of metallocycloheptane with an ethylene molecule coordinating to the chromium center. As shown in Table 5, chemical bonds Cr-C or Cr-O, which linked the chromium atom to a R or OR group (except models A and B, see column “Cr-R or Cr-OR”) and the two Cr-CR bonds in the ring part of 9X (see column “Cr-CR”), were obviously elongated after adding DME. The addition of DME made 9X (X = B, D, F, H, and J) a less stable state, so the energy required for ethylene coordination giving 9X increased significantly (6X f 9X). Also the addition of DME loosened the ethylene Π-complex 9X to make room for the coordinated ethylene molecule and therefore greatly facilitated the insertion step to form the nine-membered ring (9X f 10X).
Table 5 also lists distances of the chromium atom from the two carbon atoms in the coordinated ethylene molecule in 9X (see column “Cr-Cethylene”). From these data it is found that after adding DME the coordinated ethylene molecules became closer to the chromium centers for neutral active site models (X = D, F) but became farther away from chromium centers for cationic active site models (X = B, H, and J). As shown in Table 2, after adding DME the Mulliken charge on the chromium center in 9X was increased in cationic models B Cr(I)þ/DME and J Cr(II)þR/DME and were decreased for the other three models with DME coordination (D Cr(I)OR/DME, F Cr(I)R/DME, and H Cr(II)þOR/DME), but were decreased more in the two neutral models. So the fourth ethylene should become closer to the chromium center in cationic models and farther away from the chromium center in neutral models due to the electronic effect of DME. However, the shorter distances of the Cr atom from O atoms of the coordinated DME in cationic models than that in neutral models (see Table S.4 in the Supporting Information) indicated that DME possessed stronger steric hindrance to ethylene coordination in cationic models than that in neutral models, which was considered as a more important effect than the electronic effect of DME on structure 9X. Therefore, the coordinated ethylene molecules in 9X became farther away from the chromium centers for cationic active site models (X = B, H, and J) but became closer to chromium centers for neutral active site models (X=D, F). Considering these changes in coordination distance of ethylene molecules from chromium centers together with the main effect of the loosened structure of 9X (X = B, D, F, H, and J) on ethylene insertion discussed above, after adding DME the energy barriers of the ethylene insertion (TS[9X-10X]) to form a nine-membered ring in each model all decreased, but lowered more in neutral active site models (X = D, F) than that in cationic active site models (X = B, H, and J) (see Table 4, column “Ring7 f Ring9, insertion”). In summary, DME in neutral models acted as weak electron donors and increased the electron density of chromium center; while in cationic models DME were strong electron donors, and thus for some cationic intermediates the electron density of chromium center was just increased a little and for the other cationic intermediates the electron density of chromium center was even decreased. The steric effect of DME was to make both metallocylce growth and 1-hexene liberation difficult to occur for all active site models. Considering both the electronic and steric effect of DME, the
Article
Organometallics, Vol. 29, No. 7, 2010
1601
Table 6. Apparent Energy Barriers (kcal/mol) of Nine-Membered Ring Formation (P) and 1-Hexene Liberation via AgosticAssisted β-H Transfer (T) and the Differences between the Two Barriers for Each Cationic Model without DME P þ
A Cr(I) G Cr(II)þOR I Cr(II)þR
a
16.4 15.5 13.8
T
b
27.7 18.9 14.9
with DME Δ(T-P) 11.3 3.4 1.1
P
Td
Δ(P-T)
19.2 36.4 32.7
16.7 13.4 17.2
2.5 23.0 15.5
c
þ
B Cr(I) /DME H Cr(II)þOR/DME J Cr(II)þR/DME
a Energy barrier of TS[9X-10X]. b Energy difference between TS[6X-8X] and 9X. c Energy difference between 6X and TS[9X-10X]. d Energy barrier of TS[6X-8X].
addition of DME led to different catalytic performance between neutral and cationic models. The neutral models without DME coordination were dominantly ethylene trimerization catalysts. After adding DME into the neutral models, the electronic effect of DME increased electron density of chromium center, which made ethylene coordination and insertion unfavorable. Meanwhile the steric effect of DME greatly increased the energy required for ethylene coordination to chromium centers. As a result, in neutral models the total energy barriers for metallocylce expansion of Ring5fRing7 were much increased by the addition of DME, and thus made β-hydrogen transfer in five-membered ring leading to ethylene dimerization comparable with the formation of seven-membered ring, so after adding DME, neutral models D Cr(I)OR/DME and F Cr(I)R/DME performed ethylene trimerization but with dimerization as a side reaction. Before adding DME the three cationic models A Cr(I)þ; B Cr(I)þ/DME and G: Cr(II)þOR proceeded metallocycle growth dominantly leading to ethylene polymerization. In cationic models the electronic effect and steric effect of DME on metallocycle expansion were contrary to each other, the former facilitated the metallocycle expansion process while the latter was unfavorable for metallocycle growth. For metallocylce expansion of Ring5fRing7 in cationic models with DME coordination, the electronic effect of DME was more important, so seven-membered ring was formed and ethylene dimerization was disfavored. For metallocylce expansion of Ring7fRing9 in cationic models with DME coordination, the electronic effect of DME tried to ensure the coordination and insertion of fourth ethylene by generally decreasing the electron density of chromium centers, but the steric effect of DME played a more important role, so the formation of nine-membered ring was disfavored and 1-hexene was liberated from seven-membered ring in cationic models after adding DME. Therefore, cationic chromium active site models (such as A: Cr(I)þ; B: Cr(I)þ/DME; G: Cr(II)þOR; H: Cr(II)þOR/DME; I: Cr(II)þR; J: Cr(II)þR/ DME) might be the most plausible active sites for the Cr(EH)3/ PIBAO catalyst system with or without DME coordination. The apparent energy barriers of nine-membered ring formation (representing ethylene polymerization) and 1-hexene liberation (ethylene trimerization) and the differences between the two barriers for each cationic model before and after addition of DME were summarized in Table 6, a transformation from ethylene polymerization to ethylene trimerization should occur after addition of DME on these cationic models. A preliminary evaluation of the catalytic selectivity and activity of each active site model could be obtained.
Conclusions In this work, the role of DME in the switching mechanism from ethylene polymerization to trimerization was investigated
by the DFT method based on the 10 most plausible molecular models (A, Cr(I)þ; B, Cr(I)þ/DME; C, Cr(I)OR; D:, Cr(I)OR/DME; E, Cr(I)R; F, Cr(I)R/DME; G, Cr(II)þOR; H, Cr(II)þOR/DME; I, Cr(II)þR; J, Cr(II)þR/ DME) for the active species of the Cr(EH)3/PIBAO catalyst system with or without DME coordination. It was found that 1-butene elimination tended to follow the two-step path with chromium hydride as an intermediate, whereas 1-hexene liberation preferred to adopt a one-step path via agosticassisted β-hydrogen transfer. The DME ligand presented a flexible electronic effect: for neutral models DME acted as a weak electron donor and increased the electron density of the chromium center, whereas for cationic models the electrondonating effect of DME was so strong that the electron was pushed further onto the other ligands rather than on the chromium atom; therefore the electron density of the chromium center was just increased a little for some cationic intermediates and was even decreased for the other cationic intermediates. For each active site model, the steric effect of DME increased the energy barrier of the metallocylce growth more than that of 1-hexene liberation. On the basis of the overall consideration of electronic and steric effects of DME and the calculation results of each active site model, it was found that (1) ethylene trimerized into 1-hexene on neutral models without DME coordination because ninemembered-ring formation could not occur through metallocycle growth. After adding DME to neutral models, ethylene trimerization occurred but with ethylene dimerization as a side reaction, because both the electronic effect and steric effect of DME increased the total energy barriers of Ring5 f Ring7 and Ring7 f Ring9, even caused β-hydrogen transfer in the five-membered ring, leading to ethylene dimerization comparable with the formation of the seven-membered ring. The high energy barriers required for metallocycle growth in neutral models with or without DME coordination indicated that the metallocycle mechanism was not feasible with neutral active sites. (2) Before adding DME, cationic models gave theoretical results of metallocycle growth leading to ethylene polymerization. After adding DME, the electronic effect of DME was to facilitate metallocycle expansion, while the steric effect of DME was unfavorable for metallocycle growth. The steric effect of DME played a more important role during metallocycle expansion from Ring7 f Ring9, so energy barriers for the formation of chromium-cyclononane from chromium-cycloheptane became much higher than those of 1-hexene liberation via agostic-assisted β-hydrogen transfer, and therefore ethylene polymerization was transformed into ethylene trimerization by the addition of DME. Calculation results of neutral active site models (C, Cr(I)OR; D, Cr(I)OR/DME; E, Cr(I)R; F, Cr(I)R/DME) were not consistent with experimental evidence and thus could be excluded, whereas all the cationic active site models, which gave modeling results consistent with experiments, might be
1602
Organometallics, Vol. 29, No. 7, 2010
the most plausible active sites for the Cr(EH)3/PIBAO catalyst system with or without DME coordination. These theoretical results provided much deeper insight into understanding the highly selective trimerization mechanism and for further development of new catalysts with high performance as well.
Acknowledgment. We thank Westlake Chemical Corporation for financial support and permission to publish this work. This work is also financially supported by the Research Program of the State Key Laboratory of Chemical Engineering, Shanghai Pujiang Talent Plan Project (08PJ14032), and the Program of Introducing Talents of Discipline to Universities (B08021). We thank
Qi et al.
Prof. Sebastian Kozuch from the Institute of Chemistry and the Lise Meitner-Minerva Center for Computational Quantum Chemistry (Israel) and Prof. Yi Luo from Dalian University of Technology (China) for valuable discussions and suggestions. Supporting Information Available: Gibbs free energy surfaces for neural models (Figures S.1 to S.4), relative energy obtained for spin-state confirmation (Table S.1), geometric parameters of the DME ligand (Tables S.2 to S.4), Mulliken charges for each stationary point in each model (Tables S.5 to S.9), and listings of the Cartesian coordinates of stationary points in each model and TS[2F-4F]. This material is available free of charge via the Internet at http://pubs.acs.org.
