Active Sites in a Heterogeneous Organometallic Catalyst for the Polymerization of Ethylene.

Heterogeneous derivatives of catalysts discovered by Ziegler and Natta are important for the industrial production of polyolefin plastics. However, the interaction between precatalysts, alkylaluminum activators, and oxide supports to form catalytically active materials is poorly understood. This is in contrast to homogeneous or model heterogeneous catalysts that contain resolved molecular structures that relate to activity and selectivity in polymerization reactions. This study describes the reactivity of triisobutylaluminum with high surface area aluminum oxide and a zirconocene precatalyst. Triisobutylaluminum reacts with the zirconocene precatalyst to form hydrides and passivates -OH sites on the alumina surface. The combination of passivated alumina and zirconium hydrides formed in this mixture generates ion pairs that polymerize ethylene.

Introduction

Ziegler and co-workers discovered that mixtures of triethylaluminum and zirconium acetylacetonate polymerize ethylene to high-density polyethylene under mild conditions in 1953, and two years later Natta reported that TiCl4 and Et2AlCl mixtures polymerize propylene to stereoregular products (Figure a).[1,2] Derivatives from these initial discoveries evolved to heterogeneous catalysts used industrially that account for a majority of the polypropylene (PP, ∼50 millions tons) and polyethylene (PE, ∼100 million tons) produced per year. A key question related to the initial Ziegler–Natta solution catalysts was how the metal and the activator interact to form active organometallic species for polymerization reactions. This question becomes more difficult to address considering that most Ziegler–Natta catalysts are significantly more active when supported on MgCl2.[3] Reactions of Cp2TiCl2 (Cp = cyclopentadienyl) with Et2AlCl provided preliminary evidence for the formation of ionized organometallic active species in polymerization reactions.[4] Cp2TiCl2/Et2AlCl mixtures are not particularly active in polymerization, but the serendipitous discovery of methaluminoxane (MAO) activators resulted in soluble metallocene catalysts that have activities approaching those of heterogeneous Ziegler–Natta catalysts.[5] The isolation of reactive Cp2ZrMe(THF)+ established that cationic organometallic zirconium species are active in polymerization reactions,[6] and the design of efficient activators to form cationic organometallics led to general strategies that allowed for explicit molecular design of the active site in polymerization reactions (Figure b).[7,8] These activators play important roles in generating catalysts that regulate molecular weight properties of the polymer and in copolymerization reactions in solution.[9−11]

Figure 1

Evolution in the understanding of the Ziegler–Natta catalyst for olefin polymerization, showing key discoveries for homogeneous (top) and heterogeneous (bottom) catalysts (a). Current strategy to activate metallocenes in solution (b). Current strategy used industrially to form activated metallocenes on surfaces (c). Formation of well-defined sites on oxides with preformed organometallics, and the objective of this study to determine the active site structure in a model industrial catalyst for polymerization of ethylene (d).

Strategies to form cationic organometallic species on heterogeneous supports, the more important industrial class of catalysts for polymerization reactions, usually involve formulations containing a high surface area oxide, an excess of alkylaluminum (or MAO), and a metallocene precatalyst (Figure c).[12,13] Complications arising from the low quantity of active sites present in these catalysts prevent a detailed structural understanding of the active site. However, complementary studies of organometallics supported on oxides, which are likely important in these heterogeneous catalysts, arrived at similar conclusions as studies in solution. Tetraalkyl zirconium complexes supported on silica have low activity in polymerization reactions, but alumina supports provide much higher activities.[14,15] The origin of this support effect was not clear until solid-state NMR studies showed that Cp*2ThMe2 (Cp* = pentamethylcyclopentadienyl) reacts with Al2O3 to form [Cp*2ThMe][Me-AlO] ion pairs,[16,17] which also occurs in reactions of organozirconium complexes supported on Al2O3 or SiO2/Al2O3.[18] This model suggests that preformed organometallics interact with an appropriate oxide to form electrophilic ion pairs that are active in polymerization reactions, a strategy employed by several groups to understand these catalysts (Figure d).[19−23] Though compelling, these model systems differ significantly from heterogeneous catalysts used for most industrial applications because they are derived from precatalysts containing preformed M–R groups and do not contain a large excess of alkylaluminum required in commercial polymerization reactions with metallocene chloride precatalysts.

This study describes the generation and characterization of the catalytically active sites in a ternary 1,1′-dibutylzirconocene dichloride (Cpb2ZrCl2, 1)/triisobutylaluminum (AliBu3)/Al2O3 catalyst for the polymerization of ethylene (Figure d).[24,25] This mixture is complex and results in a network of reactions in solution and on the surface of Al2O3 to ultimately form catalytically active [Cpb2Zr-H][H-AlO] ion pairs on the AliBu3-passivated Al2O3 surface. The formation of ion pairs relates this catalytic mixture to the solution organometallic catalysts and well-defined heterogeneous catalysts shown in Figure .

Results and Discussion

A mixture of 1, AliBu3, and Al2O3 at a Zr/Al molar ratio of 1:12 ([Zr] = 150 μmol gAl–1) is very active in ethylene polymerization (8.4 × 107 gPE molZr–1 h–1) and produces a modestly narrow distribution of high molecular weight PE (Mn = 90.8 kg mol–1; Đ = Mw/Mn = 4.25). ICP-OES analysis of the isolated solid catalyst after washing shows that only 0.65 μmol of Zr gcat–1 is present, indicating that most of the metallocene does not adsorb to the alumina surface. Omitting 1, AliBu3, or Al2O3 from the reaction mixture results in negligible polymerization activity (see the Supporting Information).

AliBu3 and Al2O3 are expected to form a complex mixture of hydrolyzed alkylaluminum species bound to the Al2O3 surface,[26] some of which may activate 1 similar to MAO in solution. The reaction of Al2O3 calcined at 600 °C (∼3 −OH nm–2, 0.93 mmol −OH gAl–1) with excess AliBu3 in pentane forms 0.86 mmol of isobutane gAl–1 indicating that most of the −OH groups on alumina react with AliBu3. Isobutene (0.19 mmol gAl–1) and HAliBu2 also form in this reaction.

The 13C cross-polarization magic angle spinning NMR (CPMAS) spectrum of AliBu3/Al2O3 contains signals at 26 and 18 ppm for the Al–iBu fragment (Figure S7). 1H–27Al dipolar recoupled insensitive nuclei enhancement polarization transfer (D-RINEPT) experiments recorded under fast MAS (νr = 50 kHz) show that 1H NMR signals from the Al–iBu fragment are near Al(IV) and Al(VI) sites on the Al2O3 surface (see the Supporting Information for details). This result is consistent with a high coverage of Al–iBu groups on the Al2O3 surface. DFT studies of a hydrated (110) Al2O3 surface containing 3 −OH nm–2 show exergonic adsorption and grafting of AliBu3 onto the surface to form tetrahedral (≡AlO)2AliBu(O(AlO)2) shown in Figure a (see Supporting Information for details). Though a distribution of tetrahedral (≡AlO)2AliBu(O(AlO)2) is likely present on the alumina surface, the structure of these sites has little influence on catalysis because 1 reacts with AliBu3/Al2O3 to form inactive polymerization catalysts, showing that MAO-type sites are not present on AliBu3/Al2O3.

Figure 2

Polymerization activity of (≡AlO)2AliBu(O(AlO)2), formed from the reaction of AliBu3 with Al2O3, with 1 or products of the reaction of 1 and excess AliBu3 (a). The aluminum originating from the AliBu3 is shown in red. Products formed in the reaction of 1 with excess AliBu3 and the independent synthesis of 2, the major product in this reaction mixture (b). Generation of [CpbZr-H][H-AlO/AliBu3] that is consistent with polymerization activity data (c).

AliBu3/Al2O3 is clearly not involved in the activation of 1 but is undoubtedly relevant to formation of active sites in this catalyst. Polymerization activity is recovered when AliBu3/Al2O3 is contacted with a mixture of 1 and AliBu3 (Zr/Al = 1:12). Removal of excess AliBu3 from the solid catalyst prior to polymerization results in a catalyst that produces narrow molecular weight distributions of polymer (Đ = 2.37; Figure a) close to the expected value characteristic of single-site behavior (Đ = 2).

Under typical polymerization conditions, AliBu3 is present at sufficient excess to fully saturate the Al2O3 surface and react with 1. Indeed, the reaction of 1 with 12 equiv of AliBu3 in deuterated methylcyclohexane (C7D14) at typical concentrations for polymerization reactions forms a mixture of isobutene, ClAliBu2, HAliBu2, Cpb2Zr(μ-H)3(AliBu2)(AliBu3) (2), and Cpb2Zr(μ-H)3(AliBu2)3(μ-Cl)2 (3, Figure b). The 1H NMR spectrum of this mixture at −40 °C (2:3 ≈ 4:1) contains Zr–H signals at −0.98, −1.32, and −1.72 ppm for 2 as well as the Zr–H signals for 3, which was previously reported.[27]2 can be independently generated by mixing [CpbZrH2]2 (4) with equimolar AliBu3 and HAliBu2.

The formation of 2 involves Zr–Cl for Al–iBu exchange to form ClAliBu2 and Zr–iBu intermediates that undergo β-H elimination to form Zr–H species and isobutene. Reactions of Zr–H with Al–Cl regenerate Zr–Cl and form HAliBu2 that is needed to form 2 and 3. The large excess of AliBu3 facilitates exhaustive exchange with the metallocene to ultimately form Cpb2ZrH2, which is trapped by HAliBu3 and AliBu3 to form 2.

Figure a summarizes the polymerization activity of 2, 3, or 4 in the presence of AliBu3/Al2O3. 2 reacts with AliBu3/Al2O3 to form active polymerization catalysts with similar activities and polymer properties as in situ catalysts, but 3 does not form active polymerization catalysts when contacted with AliBu3/Al2O3, showing that the alkylaluminum activator can dramatically affect polymerization productivities. 4 also reacts with AliBu3/Al2O3 to form an active polymerization catalyst (1.2 × 107 g PE molZr–1 h–1; Đ = 2.75). The slightly lower activity of 4/AliBu3/Al2O3 is probably related to the higher Zr loading in this material (7.6 μmol Zr gcat–1), which is beneficial for mechanistic studies. This collection of data indicates that AliBu3 reacts with 1 to form 2, which is activated by AliBu3/Al2O3 to form the ionized [Cpb2Zr-H][H-AlO/AliBu3] shown in Figure C.

[Cpb2Zr-H]+ sites in 4/AliBu3/Al2O3 are expected to insert vinyl halides and undergo fast β-halide elimination to form unreactive [Cpb2Zr-X]+.[28,29] Quantification of the products in this reaction correlates with the amount of zirconium sites capable of olefin insertion. The reaction of 4-d2/AliBu3/Al2O3 (62% Zr–D) with excess cis-dichloroethylene forms cis/trans-vinyl chloride-d1, vinyl chloride, isobutene, and a small amount of ethylene (Figure a). An excerpt of the 1H NMR spectrum of this reaction mixture is shown in Figure b. On the basis of the 1H NMR peak integrals, 1.8 μmol g–1 of vinyl chloride-d1 form in this reaction, indicating that 23% of Zr-D+ present in 4-d2/AliBu3/Al2O3 are active in olefin insertion reactions; this value is higher than suspected for heterogeneous polymerization catalysts formed in the presence of alkylaluminum activators but significantly lower than the active site counts for cationic metallocenes in solution.[30]

Figure 3

Reaction of 4-d2/AliBu3/Al2O3 with cis-dichloroethylene to form reaction products (a). Excerpt of the 1H NMR spectrum from 4.6–5.4 ppm (b). The symbols above each signal in (b) correspond to ∼ = 13C satellite from cis-dichloroethylene; * = ethylene; $ = vinyl chloride; + = trans-vinyl chloride-d1; # = cis-vinyl chloride-d1. Proposed mechanism that accounts for formation of vinyl chloride-d1 and vinyl chloride-d0 (c).

The unlabeled products probably form by the successive reactions of Zr–D+ with cis-dichloroethylene shown in Figure c. Following β-chloride elimination, the surface-bound Zr–Cl+ (∼0.02 nm–2) is alkylated by a nearby Al–iBu (∼3 nm–2) that regenerates a Zr–H+ and forms isobutene. Subsequent reaction of Zr–H+ and cis-dichloroethylene results in the formation of vinyl chloride and Zr–Cl+. This scenario is consistent with the 1:1 ratio of isobutene: vinyl chloride-d0 obtained from the 1H NMR spectrum in Figure b.

Deuterium is an NMR-active quadrupolar isotope (spin I = 1). Solid-state 2H NMR spectra show characteristic broad powder patterns that are a result of interactions between the nuclear electric quadrupole moment, eQ, and the electric field gradient (EFG) tensor , eq . The line shape of a 2H MAS NMR spectrum at the slow exchange limit is described by the quadrupolar coupling constant (CQ, eq ) and the asymmetry parameter (η, eq ). Terminal M–D are expected to have η = 0, bridging M–D–M that deviate from linearity is expected to have η ≠ 0, and CQ is expected to increase as the effective nuclear charge increases.[31] Thus, 2H MAS NMR is capable of distinguishing between a variety of possible Zr–D structures in 4-d2/AliBu3/Al2O3.

Figure shows 2H MAS NMR spectra for 4-d2, monomeric (C5Me5)2ZrD2,[32−34] [(C5Me5)2ZrD][DB(C6F5)3],[35] and 4-d2/AliBu3/Al2O3. The CQ and η values extracted from this data are consistent with the expectations mentioned above. The 2H MAS NMR spectrum of 4-d2 is shown in Figure a and contains two sets of peaks assigned to the terminal Zr–D at 5.3 ppm with a CQ of 50 kHz and η = 0, and the bridging Zr–D–Zr at −3.3 ppm with a CQ of 44 kHz and η = 0.3, close to values reported for [Cp2ZrD2]2.[36] The magnitude of CQ for the Zr–D in (C5Me5)2ZrD2 (CQ = 44; η = 0, Figure b) is similar to 4-d2, indicating that neutral Zr–D are characterized by small CQ values. The 2H MAS NMR spectrum of [(C5Me5)2Zr-D][DB(C6F5)3], shown in Figure c, contains a signal for the Zr–D+ at 9.3 ppm with a CQ of 111 kHz (η = 0) and a signal at 0.7 ppm (CQ = 105; η = 0) for the D–B(C6F5)3. Both (C5Me5)2ZrD2 and [(C5Me5)2ZrD][DB(C6F5)3] also contain a sharp signal with a narrow CQ ≈ 20 kHz for sp3 C–D bonds that are under fast rotational exchange on the 2H NMR time scale, indicating that some deuterium is incorporated into the C5Me5 ligand.[37]

Figure 4

2H MAS NMR spectrum of dimeric [Cpb2ZrD2]2 (a), monomeric Cp*2ZrD2 (b), [Cp*ZrD][DB(C6F5)3] (c), and 4-d/AliBu3/Al2O3 recorded at −20 °C (d). Expansion of the 2H MAS NMR spectrum from 45 to −40 ppm of 4-d2/AliBu3/Al2O3 recorded at −20 °C (top) and −100 °C (e). Experimental spectra are shown in black, and simulations are shown in red, blue, or orange. Zr–H/H–Al exchange consistent with the 2H MAS NMR data (f).

The 2H MAS NMR spectrum of 4-d2/AliBu3/Al2O3 obtained at 18.8 T at 15 kHz spinning and −20 °C is shown in Figure d. This spectrum contains signals at 2.0 and 7.5 ppm. The signal at 2.0 ppm (CQ = 32 kHz, η= 0.2) is also present in AliBu3/Al2O3 and is assigned to the natural abundance 2H signal from AliBu3/Al2O3, but could also be a result of H/D exchange between 4 and Al–iBu groups that occurs in the synthesis of 4-d/AliBu3/Al2O3. The signal at 7.5 ppm has CQ of 129 kHz and η of 0.35 is suggestive of a bridging Zr–D+ site and supports the formation of [Cpb2Zr–D][D–AlO] as the active species in 4-d2/AliBu3/Al2O3. However, the signal for the [D–AlO] site is not present in the spectrum in Figure d. An expansion of the 2H MAS spectrum recorded at −20 °C and −100 °C is shown in Figure e. The spectrum at −100 °C contains a signal at 1.9 ppm for the surface AliBu3/Al2O3 (CQ = 30 kHz, η = 0.3), which is slightly broader than the signal recorded at −20 °C. This spectrum also contains signals at 9.3 ppm (CQ = 150 kHz, η = 0), similar to the chemical shift of the Zr–D+ in [(C5Me5)2Zr–D]+ and assigned to the terminal Zr–D+ of the cationic [Cpb2Zr–D]+ fragment in 4-d/AliBu3/Al2O3, and 5.3 ppm (CQ = 100 kHz, η = 0.5) assigned to the anionic [D–AlO] fragment in 4-d/AliBu3/Al2O3.

These results are consistent with the exchange process shown in Figure f. At −20 °C, the 2H NMR signals for [Cpb2Zr–D][D–AlO] undergo site exchange that results in average chemical shifts, reduced CQ, and perturbed η values that depend on the motion these two sites, which accounts for the observation of only one 2H NMR signal in 4-d2/AliBu3/Al2O3 at −20 °C. Similar behavior was encountered in metallocenium [MeB(C6F5)3] ion pairs,[38] suggesting that the [D–AlO]− anions are weakly coordinated to the zirconium deuteride cation in 4-d/AliBu3/Al2O3. At −100 °C, this exchange process is slow on the 2H NMR time scale, and individual signals for [Cpb2Zr–D][D–AlOx] in 4-d2/AliBu3/Al2O3 are obtained. −100 °C is cold enough to slow the exchange between the active sites in 4-d/AliBu3/Al2O3 but not cold enough to slow rotation in the sp3 C–D bonds in AliBu3/Al2O3 (CQ ≈ 170 kHz). The CQ and η values for the [Cpb2Zr–D]+ fragment in 4-d2/AliBu3/Al2O3 are in agreement with the trends observed in the representative molecular zirconium deuterides shown in Figure .

The bridging Zr–D–Al in 4-d2/AliBu3/Al2O3 is similar to other cationic zirconium hydrides containing bridging Zr–H–E (E = B(C6F5)3, HAlR2) in solution.[39−41] In many cases, displacement of the bridging hydride by ethylene is slow relative to chain growth in olefin polymerization reactions in solution.[42−45] DFT studies of [Cp2ZrMe][MeAlO], formed from the reaction of Cp2ZrMe2 with fully dehydroxylated alumina, showed that the metallocenium fragment is more weakly coordinated to certain sites on the alumina surface than a typical [MeB(C6F5)3] weakly coordinating anion.[46] This study, and the dynamics of 4-d/AliBu3/Al2O3 from the 2H MAS NMR data reported here, suggests that [D-AlO]− is also bound more weakly to the [Cpb2Zr-D]+ fragment than typical bridging hydrides in solution and is consistent with the high polymerization activity of 4/AliBu3/Al2O3.

Conclusion

The combination of 1, AliBu3, and Al2O3 results in active catalysts for the polymerization of ethylene that approach single-site behavior under appropriate conditions. Excess AliBu3 is essential in this mixture to rapidly react with the −OH sites on Al2O3 and to activate 1 to form 2.[47] Both of these reactions result in unexpected reaction products that play critical interconnected roles that lead to the formation of active sites in this catalyst. The distribution of (≡AlO)2AliBu(O(AlO)2 present in AliBu3/Al2O3 are not capable of reacting with 1 to form active sites. This result is surprising given the well-known ability of partially hydrolyzed alkylaluminums to activate metallocene precatalysts in solution.[5] However, the Al–iBu groups in AliBu3/Al2O3 are critical because they prevent the reaction of −OH on Al2O3 with the zirconium hydrides formed by the reaction of AliBu3 and 1. Passivation of −OH groups on Al2O3 with AliBu3 allows 2 to react with Lewis sites still present on the passivated Al2O3 surface[48,49] and is similar to the reactions of Cp*2ThMe2 with fully dehydroxylated alumina reported over 35 years ago.[17] The data presented here connects a typical ternary heterogeneous catalyst formulation relevant to industry to well-defined organometallics supported on oxides and homogeneous metallocene catalysts. This understanding gives a simple model to guide catalyst formulations that may result in heterogeneous catalysts for the synthesis of advanced polyolefin materials using a more rational structure–property optimization strategy.