Since the pioneering work of Ziegler and Natta [1, 2] in the 1950s, transition metal catalysts for olefin polymerization have attracted considerable interest, largely because of their various application in industrial processes. Among the transition metal catalysts, vanadium-based catalysts exhibit unique characteristics for the synthesis of high molecular mass polymers with uniform molecular mass distributions [3, 4, 5], and they are also efficient catalysts for synthesizing syndiotactic polypropylene [6, 7, 8] and ethylene/α-olefin copolymers with high α-olefin incorporation [9, 10, 11, 12]. Hence, numerous experimental investigations have been reported for the design of ligand systems with vanadium catalysts [13, 14, 15, 16, 17, 18, 19, 20, 21]. The main disadvantage of vanadium catalysts is deactivation associated with the reduction to low-valent, less active, or inactive species during catalysis [22, 23, 24]. Moreover, because of the unpaired electrons in V(II), V(III), and V(IV) complexes, it is difficult to experimentally determine the structures of the active species of vanadium catalysts. Investigation of olefin polymerization mechanisms by vanadium catalysts, especially the formation of the active species, is still a scientific challenge.
Recent progress in computational chemistry has shown that it is an efficient tool to investigate kinetic mechanisms and the thermodynamic properties of transition metal reactions [25, 26, 27, 28]. Many theoretical investigations of olefin polymerization mechanisms involving transition metals have been reported, using ab initio [29, 30], semiempirical [31, 32], density functional theory (DFT) [26, 33, 34], and molecular mechanics methods [29, 35]. Among these methods, DFT has become the preferred method for electronic structure theory of complicated reaction systems, such as organic reactions, in part because of its optimal balance of reasonable accuracy and modest computational cost. Some theoretical investigations of vanadium catalytic systems have been reported. For example, Bühl [36] carried out a DFT study of ethylene insertion into V-C bonds to help experimentally modify (arylimido)vanadium (V) species to obtain catalysts for ethylene polymerization. Zambelli and coworkers [16] investigated the syndiotactic-specific polymerization of propylene with single-site catalysts, and several proposed active species and an olefin insertion mechanism involving vanadium were investigated with DFT calculations.
Our group previously synthesized a series of vanadium(III) catalysts bearing salicylaldiminato ligands [RN=CH(ArO)]VCl2- (THF)2 (Fig. 1) that showed high catalytic activity not only for ethylene homopolymerization but also for the copolymerization of ethylene with α-olefin in the presence of AlEt2Cl [18, 20]. However, subsequent modification of this catalytic system and establishing the structure-reactivity relationship are difficult experimentally because of the lack of precise information about the nature of the catalytically active species and the elementary reactions. In this paper, we clarified in detail the mechanism of ethylene polymerization catalyzed by salicylaldiminato V(III) complexes based on DFT investigations. For the calculations, [C6H5N=CH(C6H4O)]VCl2(THF)2 (2a, 1) was chosen as the model compound to explore the active form. Optimized structure 1 agreed well with the molecular structure obtained via X-ray crystal analysis. Our research aimed to understand (1) possible models for the vanadium active species, (2) the involved olefin polymerization mechanism, and (3) the most favorable chain termination pathway.
DFT calculations were performed using the Amsterdam Density Functional program [37], which has been shown to be suitable for calculating structure and vibrational model for large systems, and especially for treating complexes containing transition metals. Unrestricted spin-treatment was used for all calculations. All of the structure calculations were based on the Becke-Perdew exchange-correlation functional [38, 39]. A triple Slater-type orbital (STO) basis set was used for V, while all other atoms were described by a double-ζ plus polarization STO basis. The 1s22s22p6 configuration on vanadium and chlorine, and the 1s2 configuration on carbon, nitrogen, and oxygen were assigned to the core and treated by the frozen-core approximation. The structure of transition states was obtained by full transition-state optimizations based on linear transit calculations. First, along the stepwise decreasing reaction coordinates, all variables except for the reaction coordinate bond of the structures were optimized. Then, the maximum of this linear transit was used as the starting structure in the full transition-state optimization without any geometry constraints. For the ethylene insertion reaction, the reaction coordinate was chosen as the formation of the new C-C bond between the ethylene monomer and the polymer chain on the catalyst vanadium center.
To obtain more accurate activation energy of propagation and chain transfer steps in this catalytic system, single-point calculations were carried out using the M06-L functional [40] based on structure optimized by BP86 functional calculations. Solvent effects were estimated based on the gas-phase geometries with the conductor-like screening model (COSMO). A dielectric constant of 2.38 was chosen to represent toluene as the solvent. The atomic radii used were 0.19, 0.23, 0.13, 0.14, 0.23, and 0.116 nm for V, Al, O, N, C, and H, respectively. For comparison, single-point calculations were also performed using the B3LYP-D3 functional [41]. The M06-L and B3LYP-D3 SCF energy were calculated with an all-electron TZP basis set for V and an all-electron DZP basis set for all other atoms.
To determine the ground states of the salicylaldiminato vanadium system, unrestricted calculations for the singlet and triplet states of catalysts 2a (1), 2b, 2e, and 2f ([RN=CH(ArO)]- VCl2(THF)2, Ar = C6H4, R = Ph, 2a; R = p-CF3Ph, 2b; R = 2,6- iPr2Ph, 2e; R = cyclohexyl, 2f) were performed. The results showed that the singlet state of 2a was 9.3 kcal/mol higher in energy than the triplet state, and changing the ligands does not change this trend (the energy differences of 2b, 2e, and 2f were 9.2, 9.8, and 9.5 kcal/mol, respectively). As a result, it can be concluded that the ground state of this series of salicylaldiminato vanadium catalysts is the triplet state.
Table 1 shows the calculated bond lengths (pm) and angles (°) around the vanadium center of 2a compared with the structure obtained by X-ray crystallographic analysis in a previous experimental investigation [18]. The optimized bond lengths agree well with the experimental structural information. The calculated bond angles are accurate for the framework of the catalyst, while the results for bond angles involving oxygen atoms of THF ligands (O(2) and O(3)) are less accurate. The calculation methods used in this paper provide a reasonably accurate description for this series of vanadium catalysts.
The generally accepted mechanism for homogeneous polymerization catalysis is the Cossée-Arlman mechanism [42]. Without considering the influence of the cocatalyst, the insertion of olefin into the metal-carbon bond of the active species is usually described as a two-step process: monomer uptake and insertion. The uptake involves π-complexation of the monomer to the metal center, and the insertion goes through a transition state with a four-membered ring structure. Additionally, during the polymerization catalysis, the cocatalyst usually has two roles in the activation step: to alkylate the metal complex and then to abstract one of the alkyl groups to form an active species [43, 44].
For the [C6H5N=CH(C6H4O)]VCl2(THF)2 complex, AlEt2Cl was required to achieve high catalytic activity. Other organoaluminum compounds, such as modified methylaluminoxane (MAO), dry MAO, AlMe3, and AlEt3, have been proven to be inefficient for ethylene polymerization [27]. Therefore, we carried out theoretical investigations to determine the role of the cocatalyst in ethylene polymerization on the vanadium complex, as shown in Scheme 1.
If the true active species is assumed to be a cationic species, our theoretical results showed that the activation steps using different organoaluminum compounds showed similar energy profiles, not only in the alkylation process but also in the alkyl abstract process. This inconsistency between the theoretical and experimental results led us to hypothesize that AlEt2Cl has another crucial effect on the generating progress of real active species: forming bridge-linked complexes. When AlEt2Cl approaches the metal center, one chlorine atom of the vanadium precursor can be protected by forming Al-Cl-V bridging bonds before the alkylation reaction. The calculations indicated that the formation of two Al-Cl-V bridging bonds in model A and one in model B (Scheme 1) are exothermic reactions (ca. 34 and 31 kcal/mol, respectively). As described in Scheme 1, the Al-Cl-V and Al-C-V bridging bonds in model B can also occur in the presence of AlMe3, AlEt3, AliBu3, or MAO. However, the role of AlEt2Cl in the formation of two Al-Cl-Vl bridging bonds in model A is crucial because of the lack of chlorine atoms in the other organoaluminum compound. From the point of view of consistency with experiments, model A is obviously more reasonable than model B. The real structure of the active species in model A used for the calculation of the ethylene insertion mechanism is called a1.
Despite the crucial role of AlEt2Cl, there are other possible models for the vanadium active species. Thus, we also investigated the formation of several other active species and their catalytic mechanisms for ethylene polymerization, and compared these with bimetallic species a1 to confirm the assumption that the active species is cationic. Scheme 2 shows the formation processes of active species a1 and other possible active species originating from vanadium complex 1 (LVCl2(THF)2, where L is the salicylaldiminato ligand). Precatalyst 1 contains four possible leaving groups: two chlorine atoms and two THF molecules. Linear transit simulations for the approach process of AlEt2Cl to 1 always led to the loss of one THF molecule to form complex LVCl(THF)2 (2) because the coordination space around the vanadium center of 1 is crowded. Evidently, complex 2 is electron deficient, so AlEt2Cl approaches the vanadium center from the coordination vacant site, forming bridge-linked complex [LVCl3(THF)]AlEt2 (3) or [LVCl2(THF)]AlEt2Cl (4), whose structure contains two Al-Cl-V bridging bonds or one Al-Cl-V and one Al-C-V bridging bonds, respectively. The formation process of complex 3 was calculated to be exothermic by 33.7 kcal/mol, while the formation process of complex 4 was exothermic by 30.6 kcal/mol. It is noteworthy that the Al-C-V bridging bond is much weaker than the Al-Cl-V bridging bond, because the Al-C distance of the AlEt2Cl moiety in 4 elongated by only 4.6 pm compared with the free AlEt2Cl compound, while the Al-Cl distance of the AlEt2Cl moiety in 3 elongated by 15.9 pm. Accordingly, complex 4 can also be considered to contain only one Al-Cl-V bridging bond, because the carbon bridging bond is easily broken to form one coordination vacant site.
The alkylation of the unbridged chlorine atom in complex 4 will give rise to active species a3, which can be seen as an exchange progress of the chlorine atom of 4 with the alkyl groups of AlEt2Cl. As shown in Scheme 3, in the first step, AlEt2Cl coordinates to the vanadium center of 4 and forms a stable bimetallic complex, which is 18.7 kcal/mol more stable than the reactants. The following AlEtCl2 removal process is endothermic by 30.1 kcal/mol. As a result, the overall alkylation reaction of 4 is endothermic by ca. 11.4 kcal/mol. In contrast, without the loss of the THF molecule, the direct alkylation process of 3 is very difficult because the vanadium center of 3 is a stable six- coordinate form and is hardly accessible. Thus, the corresponding active species model was not considered via the direct alkylation of unbridged chlorine atom in complex 3.
From the results presented in the previous section, the loss of the THF molecule has to occur before alkylation of the unbridged chlorine atom in complex 3 can occur. This loss of the THF molecule from 3 is endothermic by 30.3 kcal/mol, and gives rise to complex 5. Subsequently, alkylation of LVCl3(AlEt2) (5) will give bimetallic species LVCl2Et(AlEt2) (a1), and the vanadium active species model presented in the previous section. Additionally, if another AlEt2Cl molecule coordinates to 5 with two Al-Cl-V bridging bonds, complex LVCl4(AlEt2)2 (7) will form, which is inactive for olefin polymerization because of the lack of a vacant coordinate site and alkyl group. The formation of 7 may be a possible side reaction in this catalytic system and will decrease the catalytic activity of the vanadium catalyst.
Another possible active species LVClEt(AlEt2Cl) (a2) can form following the route of 4 → 6 → a2 in Scheme 2. Similar to active species [LVClEt(THF)]AlEt2Cl (a3), the carbon bridging bond in a2 is weak and easily broken. Therefore, a2 and a3 can also be described as active species containing only one chlorine bridging bond, or even as intimate ion pairs. Comparing the energy profiles, there is no significant difficulty in the formation processes of neutral active species a1-a3. Additionally, the formation of chlorine bridging bonds in the active species benefits the structural stability, which is indicated by the high exothermicity in the coordination process of AlEt2Cl to the vanadium center.
The possible transformation reactions of three active species should be considered. As outlined in Scheme 2, active species a3 might transform to a2 through the loss of a THF molecule. If the AlEt2Cl moiety in a2 rotates to break the carbon bridging bond and forms a much stronger chlorine bridging bond, active species a1 will form with an exothermic energy of 9.7 kcal/mol. The transformation process from a3 to a1 can also proceed via the intermediate [LVCl2Et(THF)]AlEt2, which is formed when the AlEt2Cl moiety in a3 rotates to form the structure containing two Al-Cl-V bridging bonds.
From an energetic point of view, the main difficulty in the formation of neutral active species a1-a3 is the dissociation processes of one or two THF molecules from the vanadium center of the catalyst. Nevertheless, this dissociation process of THF may be facilitated by the energy released by AlEt2Cl coordination to the vanadium center, especially considering that a high AlEt2Cl concentration is necessary to obtain high catalytic activity in the experimental investigation.
As mentioned above, animportant role of organoaluminum compounds is to alkylate the catalyst precursor and abstract one alkyl group to give the cationic active species (see Schemes 1 and 4). As shown in Scheme 4, the alkylation of two chlorine atoms in 2 will produce complex LVEt2(THF) (8). Additionally, the THF group of 8 can be easily lost to give complex LVEt2 (9). However, subsequent alkyl abstract reactions of 8 and 9 are strongly energetically unfavorable from our theoretical results, which may be the main difficulty that needs to be overcome to form cationic active species [LVEt(THF)]+ (a4) and (LVEt)+ (a5). The process of AlEt2Cl abstracting one alkyl group from complex 8 is strongly endothermic by 31.5 kcal/mol owing to the electron deficient nature of complex 8, and the alkyl abstraction process of complex 9 needs even more energy (54.8 kcal/mol). These theoretical results indicate that the formation of cationic active species a4 and a5 is much more difficult than the formation of neutral active species.
Complexes 8 and 9 both have alkyl groups and vacant sites, which are necessary for olefin insertion. Thus, they can also be seen as possible stable active species because the subsequent alkyl abstraction processes of complexes 8 and 9 are difficult. However, further calculations indicated that the cleavage of a THF molecule is always accompanied by the coordination of ethylene to the vanadium center of complex 8. Thus, complex 8 can be excluded as a possible stable active species. Accordingly, complex 9 is included as neutral active species a6 in the discussion about the mechanism of ethylene insertion.
In summary, we have investigated the formation process of possible vanadium active species for ethylene polymerization, and obtained six stable species: two cationic species (a4 and a5) and four neutral species (a1-a3 and a6). As discussed above, active species a2 and a3 can also be described as intimate ion pairs. The formation of cationic active species is much more difficult than the formation of neutral active species because of the very high energy required for the alkyl abstraction process. Accordingly, the relatively energetically favorable structure a4 was the only cationic active species that was further investigated for the ethylene insertion process, and it is compared with neutral active species a1. Optimized structures of all of the possible active species are shown in Fig. 2.
Based on complex a1, we investigated the ethylene insertion process in the polymerization by following the well-established Cossée-Arlman mechanism. Theoretically, the ethylene coordination and insertion reaction can take place from several different directions because of the asymmetric structure of a1. However, linear transit calculations of each possible reaction pathway indicated that the attack of the incoming ethylene molecule was easiest at the trans position of the nitrogen atom. Fig. 3 shows the optimized structures of the π-complex and the transition state, and the potential energy surface of the dominant reaction pathway. In the first step, the incoming ethylene molecule coordinates to the vacant site in a1 forming π-complex C1, and this process is exothermic by 18.1 kcal/mol. Subsequently, ethylene inserts into the vanadium-carbon bond via the four-membered cyclic transition state TS1, which has an activation energy barrier of 14.7 kcal/mol relative to the π-complex (the energy of TS1 is −3.4 kcal/mol relative to the isolated reactants). The Cα-H distance in TS1 elongated to 113.8 pm, indicating a strong α-agostic interaction. This α-agostic interaction seems to facilitate the ethylene insertion reaction. The overall ethylene insertion reaction is a highly exothermic process (ca. 27.1 kcal/mol), and insertion product P1 resembles the starting structure a1. Thus, the next ethylene coordinate and insertion reaction will take place to propagate the polymerization reaction.
To investigate the most likely active species for ethylene polymerization, we compared the different ethylene insertion pathways of the various active complexes. Table 2 shows the relative energy of each pathway. For cationic active species a4, the ethylene complexation process is less exothermic than that of a1, and the insertion barrier is higher than active species a1 by 5.7 kcal/mol. On the whole, cationic active species a4 is not favorable over active species a1. Therefore, considering the very difficult formation process of active species a4, cationic species a4 can be ruled out as the most favorable active species in this catalytic system. Neutral active species a3 shows a similar ethylene insertion barrier to a1, but the ethylene complexation process of a3 is more difficult (exothermic by 13.0 kcal/mol). Furthermore, the insertion barrier of a1 (ca. 14.7 kcal/mol) is about 7 kcal/mol lower than that of a2 and about 10 kcal/mol lower than that of a6. As a whole, the much more facile ethylene coordination process and the lowest insertion barrier suggest that a1 is the active species. As a result, bimetallic species a1 should show higher catalytic activity for ethylene polymerization than the other possible active species. Mulliken charge analysis at the M06-L functional level shows that a1 has lower positive charges on the vanadium centre than the other possible active species (Mulliken charges of the V centre of a1, a2, a3, a4, and a6 are 1.45, 1.65, 1.61, 1.77, and 1.76 in atomic units). A reasonable explanation is that the less electrophilic V center of a1, which is generated by bridging the AlEt2Cl moiety, is most stable for the active species and the transition state, giving rise to enhanced catalytic activity.
The more facile formation process and higher catalytic activity for ethylene insertion of bimetallic species a1 compared with the other active species supports our hypothesis. At the same time, the crucial role of cocatalyst AlEt2Cl for active species a1 offers a more reasonable explanation of the experimental observation.
Experimental investigation has shown that ligand modification has a significant effect on the catalytic behavior of this series of salicylaldiminato vanadium catalysts [26]. For example, introducing two ortho-isopropyl groups into the N-aryl moiety of the ligand to form complex 2e increased the catalytic activity by about 10%. In contrast, introduction of a nonconjugated substituent cyclohexyl into the N-moiety of the ligand decreased the catalytic activity of complex 2f by about 50% compared with complex 2a. Following the dominant reaction pathway of ethylene polymerization catalyzed by 2a, ethylene insertion reactions with 2e and 2f were investigated. Table 3 shows the relative energy of each pathway. The calculated ethylene insertion barriers were in the order 2e < 2a < 2f, which agrees with the order of the experimental catalytic activity of 2e > 2a > 2f. This result is strong support for a1 being the active species.
The catalytic activity of complexes can be affected by various factors, such as reaction conditions, the formation processes of active species, and possible inactivation of active species. Thus, the real effect of ligands on the polymerization will be far more complex than the above comparison of the ethylene insertion barriers of 2a, 2e, and 2f. Further studies, including a more detail investigation of the effect of ligand modification and its application to the synthesis of new vanadium catalysts, are underway.
Previous experimental investigations [18, 20] have shown that the molecular mass of the polymer obtained from ethylene polymerization gradually decreases with increasing AlEt2Cl dosage. More importantly, NMR analysis shows that the end group of the polymer is saturated. These facts indicate that the dominant chain transfer reaction in this catalytic system is chain transfer to aluminum. In this section, the chain transfer reactions in ethylene polymerization catalyzed by active species a1 were investigated using DFT calculations to clarify the main factors determining the dominant chain transfer pathway. For comparison, the chain transfer reactions involved with cationic active species a4 were also investigated. Scheme 5 shows three important chain transfer reactions: (1) β-hydride transfer to the monomer, (2) β-hydride elimination, and (3) chain transfer to aluminum.
As shown in Scheme 5, the β-hydride transfer to the monomer and β-hydride elimination pathways originate from the β-agostic polymer chain structure, while the chain transfer to aluminum pathway derives from the α-agostic polymer chain structure. Therefore, it is necessary to consider the favorable product structure of the ethylene insertion reaction. As shown in Fig. 3, the most favorable reaction pathway is ethylene approaching the vanadium center and inserting into the polymer chain from the trans position with respect to the N atom. The direct product of ethylene insertion can be assumed to be the γ-agostic structure P1-γ, which is unstable and will soon convert to P1-α or P1-β because of the crowded environment around the vanadium center. P1-β was calculated to be slightly more stable than P1-α (ca. 2.9 kcal/mol), but its formation process may be more difficult. As described in Scheme 6, the polymer chain of P1-γ needs to rotate almost 180° to form P1-β, which is obviously unfavorable because the polymer chain and the ligands are eclipsing each other. The linear transit calculation result showed that the rotation process of the polymer chain needs to overcome an energy barrier of 8.5 kcal/mol. In short, P1-β has a more stable structure but P1-α has a simpler formation process. Thus, it can be assumed that both α-agostic and β-agostic structure polymer chains are present during polymerization.
If the product of ethylene insertion is β-agostic structure P1-β, ethylene can approach the polymer chain and result in the formation of β-agostic π-complex Cex. The following process where the β-hydride of the polymer chain exchanges with the incoming ethylene is a chain termination reaction. After the β-hydride transfer to the monomer reaction, a vinyl-terminated polymer chain will be removed from the vanadium center and give rise to a new ethyl vanadium complex, which can potentially initiate another ethylene insertion reaction.
The intermediates and transition states of the β-hydride transfer to monomer reaction were obtained by optimization of the structures determined by linear transit calculations. The reaction coordinate of linear transit calculation was chosen as the distance between the β-hydride on the polymer chain and the carbon atom of the approaching ethylene monomer. As shown in Fig. 4, the vacant site of P1-β is hardly accessible for steric reasons. Consequently, the β-agostic π-complex Cex is only 0.9 kcal/mol more stable than the isolated molecules. The further approach of ethylene to Cex leads to the formation of transition state TSex with an activation barrier of 10.4 kcal/mol. The structure of TSex shows evidence of H-transfer from the polymer chain to ethylene: the olefin C-C bond (139.7 pm) is longer than that in the free ethylene molecule (133.2 pm) and the Cα-Cβ bond (140.2 pm) of the polymer chain is shorter than that of P1-β (149.2 pm). Furthermore, TSex exhibits a strong β-agostic interaction because of the short Hβ-V bond (185.0 pm) and long Cβ-H bond (148.9 pm). After β-hydride transfer to the monomer, a vinyl-terminated polymer chain will dissociate from the vanadium center, and then the remaining ethyl vanadium complex can serve as a new active species for ethylene polymerization. The final products (ethyl vanadium complex and vinyl-terminated polymer chain) are 10.5 kcal/mol higher in energy than the initial reactants before β-hydride exchange.
In contrast, similar formation of the β-agostic π-complex in the pathway with cationic active species a4 is much more facile, with an exothermic energy of 16.8 kcal/mol. The energy barrier of the β-hydride transfer to monomer reaction with a4 is 17.6 kcal/mol (the energy of transition state of a4 is only 0.8 kcal/mol higher than the isolated molecules and the total reaction is exothermic by about 2.1 kcal/mol). Obviously, the β-hydride transfer to monomer reaction pathway with active species a4 is much more favorable than that with active species a1. Therefore, it can be concluded that the unique bis(chlorine-bridged) structure in a1 may be the main reason for the difficulty of the β-hydride transfer to monomer reaction compared with active species a4.
Β-Hydride elimination can be considered as the extreme form of the β-agostic interaction in complex P1-β, which is the β-hydride of the polymer chain completely transferred to the vanadium center. A metal hydride Cel and a vinyl-terminated polymer chain will then be produced. The β-hydride elimination energy profile evaluated by the linear transit method is shown in Fig. 5, where the reaction coordinate used for the calculation is the distance between the β-carbon of the growing polymer chain and the β-hydride. No obvious transition state was found in the β-hydride elimination energy profile. When the reaction coordinate is elongated to 5.0 pm, the energy of P1-β is ca. 34.4 kcal/mol higher than the initial structure. This significantly high energy required for the reaction indicates that β-hydride elimination is not a feasible pathway. For the ethylene polymerization pathway with cationic active species a4, β-hydride elimination is also a difficult chain transfer route, and is endothermic by 27.5 kcal/mol.
Considering that excess AlEt2Cl, as a cocatalyst, was used in the experimental investigations to attain high catalytic activity, polymer chain transfer to aluminum is another possible chain transfer pathway. The chain transfer to aluminum is assumed to occur via exchange of an ethyl group in AlEt2Cl with the polymer chain in the vanadium complex.
As shown in the energy diagram in Fig. 6, it is possible to assume a two-step mechanism for chain transfer to aluminum. Firstly, AlEt2Cl approaches the vanadium center of α-agostic complex P1-α to give intermediate Cex-al. Compared with the initial reactant P1-β in β-hydride elimination and β-hydride transfer to the monomer, the α-agostic structure polymer chain in P1-α more easily moves around the vanadium center to make room for the incoming AlEt2Cl. As a result, in the chain transfer to aluminum reaction, AlEt2Cl will approach the vanadium center from the vertical position of the salicylaldiminato plane. A linear transit calculation showed that this approaching process of AlEt2Cl required an energy barrier of about 8 kcal/mol, because of large steric repulsions between AlEt2Cl and the polymer chain or ligands of P1-α. Complex Cex-al is about 25.7 kcal/mol more stable than the isolated molecules. The second step is the exchange of an ethyl group on AlEt2Cl with the polymer chain on the vanadium center of complex Cex-al. We modeled the subsequent dissociation process by a linear transit calculation, where the Al-C1 bond was chosen as the reaction coordinate. Transition state TS2ex-al was obtained by full transition state optimization of the structure obtained from this linear transit calculation. The calculated energy barrier for the polymer chain transfer from the vanadium center to AlEt2Cl process was 12.2 kcal/mol. After the polymer chain completely transferred to aluminum, AlREtCl (R represents the polymer chain) will dissociate from the vanadium center to give an ethyl vanadium complex, which can initiate further ethylene polymerization. The whole chain transfer to aluminum reaction is endothermic by 2.2 kcal/mol.
Comparing the chain propagation reactions, the chain transfer to aluminum reaction has a lower activation barrier than the ethylene insertion reaction. However, considering the whole reaction process, the chain transfer to aluminum reaction is endothermic, while the ethylene insertion reaction is highly exothermic (27.1 kcal/mol). This indicates that the chain transfer to aluminum is not competitive with the ethylene insertion step, and this catalyst system can give high molecular mass polyolefins.
For the chain transfer to aluminum reaction with cationic active species a4, the product Cex-al of the AlEt2Cl approach process is 21.1 kcal/mol more stable than the ethylene insertion α-agostic product P1-α. The subsequent chain transfer process needs to overcome an energy barrier of 16.9 kcal/mol, which is similar to the chain transfer to monomer reaction (17.6 kcal/mol). The total chain transfer to aluminum reaction is endothermic by 3.8 kcal/mol. Considering that chain transfer to the monomer is an exothermic reaction, the most favorable chain transfer pathway for ethylene polymerization by active species a4 is chain transfer to the monomer, which is inconsistent with the experimental results. This further excludes the possibility that cationic species a4 is the favorable active form in this catalytic system.
Comparing the three chain transfer pathways for ethylene polymerization by active species a1, the process of chain transfer to aluminum is the simplest, because it originates from P1-α, whose formation process does not involve costly rotation of the polymer chain. The activation energy barrier of chain transfer to aluminum is similar to that of β-hydride transfer to the monomer but much lower than that of the β-hydride elimination reaction. Furthermore, the formation process of the π-complex of the chain transfer to aluminum reaction is highly exothermic, and the endothermicity of the total reaction is significantly lower than that of the other two reactions. As a result, it is concluded that chain transfer to aluminum is the main chain transfer pathway for this catalytic system. This is in good agreement with experimental observations that the molecular mass of the resultant polymer decreases with increasing Al/V ratio and the end group of the polymer chain is saturated. This is also indirect evidence for our conclusion that bimetallic neutral species a1 is the most likely catalytically active species in this vanadium(III) catalytic system.
DFT calculations were performed for ethylene polymerization catalyzed by vanadium(III) complexes bearing salicylaldiminato ligands. These calculations focused on the most likely active species for the catalytic reaction, and involved ethylene insertion and chain transfer mechanisms. By comparing the results of theoretical calculations and previous experimental results, especially regarding the crucial role of the AlEt2Cl cocatalyst, we propose a neutral bimetallic active species a1 for this vanadium catalytic system. The active species a1 contains two V-Cl-Al bridging bonds, which provides a reasonable explanation for the important effect of AlEt2Cl on ethylene polymerization. Then, this assumption was verified for both the formation process of the active species and the catalytic mechanism of ethylene polymerization.
Bridging an AlEt2Cl moiety by two Al-Cl-V bridging bonds helps to maintain the octahedral configuration of active species a1, and decreases the electron deficiency and coordinative unsaturation of the vanadium center, thereby stabilizing the active species. As a result, the formation of neutral bimetallic species a1 is more energetically favorable than the other possible active species. Additionally, bridging the AlEt2Cl moiety also facilitates ethylene insertion into the polymer chain. Consequently, the energy barrier of the ethylene insertion process with a1 is found to be significantly lower than the energy barriers with other possible neutral species. Moreover, the formation processes of neutral active species are energetically more favorable than those of cationic active species. The energetic superiority of both the formation and ethylene insertion processes supports our assumption of bimetallic active species a1. In addition, we performed comparative calculations of the chain transfer mechanisms involved with a1 and a4. The dominant chain transfer reaction with neutral active species a1 is chain transfer to aluminum, which is in good agreement with experimental observations. In contrast, the theoretical result with cationic active species a4 is inconsistent with previous experimental results. This supports that the active species is bimetallic species a1.
This investigation has provided insight into the structure of the vanadium active species for ethylene polymerization, the main steps of the olefin polymerization mechanism, and the role of cocatalysts in the polymerization. Moreover, our active species model helps to rationalize previous experimental findings and provides important information for further investigations of the design of efficient vanadium catalysts.
We are grateful to Computing Center of Jilin Province for essential support.
2015, Vol. 36 Issue (4): 657-666
PDF (1073 KB)
2015, Vol. 36 Issue (4): 657-666
PDF (1073 KB)