Preparation of High-Purity Ammonium Tetrakis(pentafluorophenyl)borate for the Activation of Olefin Polymerization Catalysts

Homogeneous olefin polymerization catalysts are activated in situ with a co-catalyst ([PhN(Me)2-H]+[B(C6F5)4]− or [Ph3C]+[B(C6F5)4]−) in bulk polymerization media. These co-catalysts are insoluble in hydrocarbon solvents, requiring excess co-catalyst (>3 eq.). Feeding the activated species as a solution in an aliphatic hydrocarbon solvent may be advantageous over the in situ activation method. In this study, highly pure and soluble ammonium tetrakis(pentafluorophenyl)borates ([Me(C18H37)2N-H]+[B(C6F5)4]− and [(C18H37)2NH2]+[B(C6F5)4]−) containing neither water nor Cl− salt impurities were prepared easily via the acid–base reaction of [PhN(Me)2-H]+[B(C6F5)4]− and the corresponding amine. Using the prepared ammonium salts, the activation reactions of commercial-process-relevant metallocene (rac-[ethylenebis(tetrahydroindenyl)]Zr(Me)2 (1-ZrMe2), [Ph2C(Cp)(3,6-tBu2Flu)]Hf(Me)2 (3-HfMe2), [Ph2C(Cp)(2,7-tBu2Flu)]Hf(Me)2 (4-HfMe2)) and half-metallocene complexes ([(η5-Me4C5)Si(Me)2(κ-NtBu)]Ti(Me)2 (5-TiMe2), [(η5-Me4C5)(C9H9(κ-N))]Ti(Me)2 (6-TiMe2), and [(η5-Me3C7H1S)(C10H11(κ-N))]Ti(Me)2 (7-TiMe2)) were monitored in C6D12 with 1H NMR spectroscopy. Stable [L-M(Me)(NMe(C18H37)2)]+[B(C6F5)4]− species were cleanly generated from 1-ZrMe2, 3-HfMe2, and 4-HfMe2, while the species types generated from 5-TiMe2, 6-TiMe2, and 7-TiMe2 were unstable for subsequent transformation to other species (presumably, [L-Ti(CH2N(C18H37)2)]+[B(C6F5)4]−-type species). [L-TiCl(N(H)(C18H37)2)]+[B(C6F5)4]−-type species were also prepared from 5-TiCl(Me) and 6-TiCl(Me), which were newly prepared in this study. The prepared [L-M(Me)(NMe(C18H37)2)]+[B(C6F5)4]−-, [L-Ti(CH2N(C18H37)2)]+[B(C6F5)4]−-, and [L-TiCl(N(H)(C18H37)2)]+[B(C6F5)4]−-type species, which are soluble and stable in aliphatic hydrocarbon solvents, were highly active in ethylene/1-octene copolymerization performed in aliphatic hydrocarbon solvents.


Introduction
Since the serendipitous discovery of methylaluminoxane (MAO) by Kaminsky, much effort has been devoted to the synthesis of organometallic complexes to find efficient single-site homogeneous olefin polymerization catalysts, some of which are currently used in the bulk polyolefin industry [1]. The initial metallocene complexes (i.e., L-MCl 2 -type complexes, where L and M are cyclopentadienyl-type ligands and group-four metals, respectively) have been followed by the development of half-metallocene complexes, in which one of the two cyclopentadienyl-type ligands in metallocenes is replaced with an amido or aryloxo ligand, and further by post-metallocene complexes that do not contain any cyclopentadienyl-type ligands [2,3]. The central group-four transition metals have also been expanded to Ni, Pd, Fe, Co, Cr, and V, especially in post-metallocene complexes [4][5][6][7]. Metallocene, half-metallocene, and post-metallocene complexes are not active, and must be activated with a co-catalyst. MAO is a typical example of a co-catalyst used in activation reactions. The structure of MAO (obtained through partial hydrolysis of Me 3 Al) is ill-defined, but may be oligomeric Me 2 Al-[Al(Me)O] n-AlMe 2 , containing some free Me 3 Al, and the mechanism of the activation reaction with MAO is still elusive [8][9][10]. In industry, MAOactivated catalysts are used mainly in a slurry or gas phase process after anchoring on the surface of silica particles, where morphological control of the generated polymer particles is crucial [11,12]. MAO is expensive and should be fed in excess relative to organometallic complexes (Al/M > 100). In a slurry or gas phase process, the catalyst retention time is long (>1 h) and the productivity of the catalyst can be maximized, mitigating the burden of the catalyst and co-catalyst cost.
Another type of co-catalyst, [PhN(Me) 2 -H] + [B(C 6 F 5 ) 4 ] − or [Ph 3 C] + [B(C 6 F 5 ) 4 ] − , was developed for the activation of organometallic complexes, which is used in a solution process in which the catalyst retention time is relatively short (<20 min) [13][14][15]. Replacement of excess MAO with stoichiometric amounts of [B(C 6 F 5 ) 4 ] − -based co-catalyst has also been a crucial issue in the development of Cr-based ethylene tetramerization catalysts [16][17][18]. In the activation reaction, L-M(Me) 2 -type complexes were converted to an ion-pair complex [L-M(Me)] + [B(C 6 F 5 ) 4 ] − by the action of the co-catalyst (Scheme 1a-c) [19][20][21]. [PhN(Me) 2 4 ] − are insoluble in aliphatic hydrocarbon solvents (e.g., hexane, cyclohexane, or methylcyclohexane), in which the commercial solution process of olefin polymerization is performed. The poor solubility of these co-catalysts causes a burden in some cases, and synthesis of the soluble form has been pursued [22][23][24]. Typically, in order to achieve optimal productivity, an excess amount of [B(C 6 F 5 ) 4 ] − -based co-catalyst (>3 equiv.) should be fed as a slurry phase directly into a bulk polymerization reactor in which both the activation reaction and polymerization reactions take place [25]. After activation in a more polar aromatic hydrocarbon solvent (e.g., toluene), the activated complex may be fed into a reactor filled with aliphatic hydrocarbon solvent. However, [PhN(Me) 2 -H] + [B(C 6 F 5 ) 4 ] − and [Ph 3 C] + [B(C 6 F 5 ) 4 ] − are either completely insoluble or sparingly soluble, respectively, in toluene [26], which still requires an excess amount (e.g., 4 equiv.) to achieve optimal productivity [27][28][29]. For a typical post-metallocene pyridylamido-Hf complex, the activation reaction was thoroughly investigated, and was more complicated than previously thought [30][31][32]. Trialkylaluminum has been additionally fed into the activation reaction to scrub the impurities and to convert L-MCl 2 -type complexes into L-MR 2 , but [Ph 3 C] + [B(C 6 F 5 ) 4 ] − is destroyed by the action of trialkylaluminum, making the activation reaction more tricky [33].
The best strategy may be the generation of the activated complex in an aliphatic hydrocarbon solvent and feeding of the solution containing the activated species to a bulk aliphatic hydrocarbon polymerization medium. With this aim, a [B(C 6 F 5 ) 4 ] − -based cocatalyst containing long alkyl chains [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − , which is soluble in cyclohexane or methylcyclohexane, was introduced more than two decades ago by Dow [34]. Activated complexes generated with [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − may be soluble and stable through the coordination of Me(C 18 H 37 ) 2 N (a byproduct generated in the activation reaction) with the metal center (i.e., formation of [L-M(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] −type species; Scheme 1c) [35,36]. However, [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − is highly soluble, cannot be purified by recrystallization, and is contaminated with either water or Cl − salt impurities, which interfere with the activation reaction [37]. In this work, we report on a method to prepare high-purity trialkylammonium tetrakis(pentafluorophenyl)borate in an aliphatic hydrocarbon solvent and the activation reactions of some typical metallocene and half-metallocene complexes.   4 ] − , was isolated by collecting the toluene phase while discarding the water phase containing the byproduct, LiCl. However, the yield was unsatisfactory (71%). Some acidic impurities were concomitantly generated, which were washed out by treatment with Na 2 CO 3 . The NCH 2 protons were observed as two broad signals at 1.93 and 1.77 ppm in the 1 H NMR spectrum ( Figure 1a); the two geminal protons attached to each α-methylene carbon (i.e., NCH 2 ) are diastereotopic. The N-H and NCH 3 signals were also broad, not showing a splitting pattern, at 3.4-3.1 ppm and 1.5 ppm, respectively. Water impurities might interact with N-H and α-protons on the ammonium unit (NCH 2 and NCH 3 ) through hydrogen bonds, causing broadening of the signals. It was impossible to completely remove water impurities; the NCH 2 signals were persistently broad even after prolonged evacuation, treatment with desiccants such as molecular sieves, or refluxing toluene solution over the Dean-Stark apparatus. A set of signals assigned to ortho-, para-, and meta-fluorine in the [B(C 6 F 5 ) 4 ] − ion was clearly observed at −132.22, −161.72, and −165.95 ppm, respectively, in the 19 F NMR spectrum. No Cl − anions were detected in the product; treatment of [(CH 3 CN) 4 Ag] + [B(C 6 F 5 ) 4 ] − with a diethyl ether solution containing the product did not generate any AgCl precipitates, which indicated that LiCl was thoroughly removed by washing with water (Supplementary Materials Figure S1b).   The byproduct LiCl can also be removed by filtration [38]. The product [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − is soluble in aliphatic hydrocarbon solvents, such as cyclohexane or methylcyclohexane, while LiCl is insoluble, enabling the removal of LiCl by filtration after performing a metathesis reaction in methylcyclohexane. The yield was satisfactorily high (94%), but the filtration procedure was tedious; the LiCl particles generated were so fine they clogged and even penetrated the Celite pad. Filtration was repeated several times, using a thick Celite pad to obtain a clear solution. In the 1 H NMR spectrum, the NCH 2 signals were fairly sharp, showing a splitting pattern, but the N-H signal was absent, possibly due to broadening (Figure 1b). Treatment of [(CH 3 CN) 4 Ag] + [B(C 6 F 5 ) 4 ] − with a solution containing the product changed the clear solution turbid due to the formation of AgCl particles (Supplementary Materials Figure S1c), which indicated that the product was contaminated with Cl − ions.

Results and Discussion
Attempts to prepare [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − by using another commercial source of borate salt, K + [B(C 6 F 5 ) 4 ] − , gave similar results. In the salt metathesis reaction performed in two water/toluene phases, the yield was low (74%) and the product was contaminated with water, although the Ag + test indicated the absence of Cl − anions (Supplementary Materials Figure S1d). Owing to the presence of water impurities, the two diastereotopic NCH 2 signals collapsed into a single broad signal ( Figure 1c). In the salt metathesis reaction performed in anhydrous methylcyclohexane, the filtration procedure was also tedious, requiring several rounds of filtration to obtain a clear solution, and the product was contaminated with Cl − anions (Supplementary Materials Figure S1e). Two signals were observed for the diastereotopic NCH 2 protons, but they were broad and did not show any splitting pattern, which was indicative of water contamination in the resulting product (Figure 1d). The commercial source of K + [B(C 6 F 5 ) 4 ] − contained some amount of water impurities, which eventually caused water contamination in the product The byproduct in this reaction was a neutral compound, PhN(Me) 2 (boiling point, 194 • C), which could be simply and completely removed by evacuation at 50-60 • C. In this method, it was impossible for the product to be contaminated with Cl − and water once anhydrous reagents and solvents were used. The yield was quantitative, and no Cl − ions were detected in the Ag + test (Supplementary Materials Figure S1a). In the 1 H NMR spectrum, the NCH 2 and NCH 3 signals were observed to be well split (Figure 1e). When water was deliberately added, the well-split NCH 2 signals were broadened as observed in Figure 1d, inferring that water contamination was a cause of signal broadening (Supplementary Materials Figure S2). Employing the same synthetic protocol, [(C 12 H 25 Figure S3). The former was soluble in methylcyclohexane, whereas the latter was not. The secondary amine-derived [(C 18 H 37 ) 2 NH 2 ] + [B(C 6 F 5 ) 4 ] − , which is soluble in methylcyclohexane, was also obtained by using the same synthetic protocol (Supplementary Materials Figure S4), whereas primary amines such as (C 17 H 35 ) 2 C(H)NH 2 were not effective in the synthetic protocol.

Preparation of L-M(Me) 2 and L-MCl(Me)-Type Complexes for Activation Studies
In most cases in industry, as well as in the laboratory, olefin polymerizations have been performed with in situ-generated activated species with neither isolation of the activated species nor monitoring of the activation reaction, because the activated species (i.e., ion pair complexes) are, in most cases, unstable or insoluble in hydrocarbon solvents. In this work, NMR spectroscopy was used to monitor the activation reactions of commercial process-relevant metallocene and half-metallocene complexes (Scheme 3) with the prepared high-purity [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − , which did not contain any water or Cl − impurities. Complexes rac-[ethylenebis(tetrahydroindenyl)]Zr(Me) 2 (1-ZrMe 2 ) and rac-[ethylenebis(indenyl)]Zr(Me) 2 (2-ZrMe 2 ) are a representative among metallocene complexes. Hafnocene complexes [Ph 2 C(Cp)(3,6-t Bu 2 Flu)]Hf(Me) 2 (3-HfMe 2 ) and [Ph 2 C(Cp)(2,7-t Bu 2 Flu)]Hf(Me) 2 (4-HfMe 2 ) were prepared in this work. L-MCl 2 -type complexes are the most common source of olefin polymerization catalyst precursors, which have been conventionally converted to L-M(Me) 2 -type complexes by the action of MeLi or MeMgCl. Although 3-HfMe 2 and 4-HfMe 2 could be prepared via the conventional method (i.e., reacting L-MCl 2 -type complexes with MeMgCl), they were prepared in a more facile and direct manner from ligand precursors 3-Li 2 and 4-Li 2 in one pot; they gave a moderate yield (41%) when 3-Li 2 and 4-Li 2 were treated with HfCl 4 in the presence of two equivalents of MeMgBr [40]. The structure of 4-HfMe 2 was confirmed by X-ray crystallography, as shown in Figure 2.

Activation Reactions
In the activation reaction of 1-ZrMe 2 with [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − , performed with the aromatic hydrocarbon solvent C 6 D 6 in a sealed NMR tube, the 1 H NMR spectral signals assigned to 1-ZrMe 2 completely disappeared with the generation of methane, but the generated signals were too complicated to be assigned to a single species. However, when the reaction was performed in the aliphatic hydrocarbon solvent C 6 D 12 , a set of signals that could be assigned to the desired activated ion pair complex, [1-Zr(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − , was cleanly generated with a methane signal ( Figure 4). The byproduct amine Me(C 18 H 37 ) 2 N might loosely coordinate to the cationic Zr center, making the signals relatively broad. By coordination with Me(C 18 H 37 ) 2 N, the two indenyl moieties are inequivalent, leading to the observation of four indenyl protons and carbons separately at 5.97, 5.91, 5.68, and 5.55 ppm as a broad singlet in the 1 H NMR spectrum (Figure 4b and Supplementary Materials Figure S7) and 115. 8, 114.8, 113.0, and 109.2 ppm in 13 C NMR spectrum (Supplementary Materials Figures S8 and S9). Signal for Zr-CH 3 was observed at 0.73 ppm as a singlet in the 1 H NMR spectrum and at 45.7 ppm in the 13 C NMR spectrum. In the 19 F NMR spectrum, a set of clean signals assignable to ortho-, para-, and meta-fluorine of -C 6 F 5 were observed at −132.0, −163.6, and −167.2 ppm (Supplementary Materials Figure S10), indicating that the [B(C 6 F 5 ) 4 ] − anion was not destroyed during the activation reaction. The activated complex was stable in C 6 D 12 ; the signals assigned to [1-Zr(Me)(N(Me)(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − were persistently observed with no generation of other signals, even after a week. In contrast, many unidentified solids were deposited in the reaction of 2-ZrMe 2 . In the C 6 D 12 solution phase, the signals that were assigned to the desired [ were observed at the initial stage, but disappeared overnight along with the complicated signals that could not be interpreted.  Figure S11). The activated complex was stable as a solution in C 6 D 12 ; the 1 H NMR spectrum was unaltered when recorded after several weeks. Analysis of the 1 H NMR spectrum indicated that the desired ion-pair complex  Figure S12). However, in this case, some insoluble fractions were concomitantly generated. We suspected that the insoluble fraction was a dinuclear adduct of  The activation reaction of half-metallocene titanium complexes is complicated. When 5-TiMe 2 was reacted with an equivalent [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − in C 6 D 12 , the reactant signals in the 1 H NMR spectrum immediately disappeared with the generation of the CH 4 signal, but the signals were too complicated to be interpreted. However, they converged to a set of assignable signals overnight (Supplementary Materials Figure S13). The generated species was not the desired ion-pair complex [5-Ti(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − ; the Ti-CH 3 signal was not observed, but, instead, two broad signals were observed at 0.45 and −0.71 ppm, whose intensity was 1/3 relative to that of the C 5 CH 3 or SiCH 3 signal. We tentatively assigned the generated species to    [37]. In contrast, stability of [L-Zr(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − type ion pair complexes generated from Zr-based metallocenes (e.g., 1-ZrMe 2 , 2-ZrMe 2 , Cp 2 ZrMe 2 ) depended on the ligand structure while those species generated from Ti-based half-metallocene complexes were unstable in all cases (see also below). These stability difference might be attributed to M-C bond strength, which increases by moving from Ti to Zr and further to Hf.  Figure 7); a signal corresponding to Ti-CH 3 in the reactant disappeared completely and the CH 4 signal appeared along with a shift of C 5 CH 3 signals. NH and NCH 2 signals were observed at 2.75-2.51. A set of signals that could be assigned to 5-TiCl 2 was also observed at an amount of approximately 10 mol%, as was observed in the activation reaction of 6-TiCl(Me).

Polymerization Studies
The performance of the activated species was tested in an olefin polymerization performed in a small-sized (75 mL) bomb reactor under 20 bar of ethylene gas by using hexane solvent (15.5 g) and 1-octene comonomer (5.0 g) with a small amount of trioctylaluminum as a scavenger (Al/L-M(Me) 2 = 33). The temperature could not be controlled; it increased from 65 • C up to 164 • C within several minutes due to the heat generated by the exothermic reaction (Table 1). Instead, temperature was recorded in an isothermal 110 • C bath, which enabled us to qualitatively monitor the performance of the catalyst. When [1-Zr(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − (1.0 µmol) was fed into the reactor at a temperature of 65 • C, the temperature increased immediately, reaching 116 • C in 2 min, and then gradually decreased, reaching 94 • C after 15 min, when the polymerization was quenched. A large amount of polymer was formed as a form of slurry (9.3 g; entry 1), which indicated that 1-octene incorporation was not significant. In contrast, when [3-Hf(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − was fed, the temperature slowly increased for 10 min until reaching 82 • C, after which, the temperature increased dramatically, reaching 164 • C in 2 min, after which the temperature gradually decreased to 91 • C during the rest 13 min polymerization time (entry 2). Amine NMe(C 18 H 37 ) 2 should be detached from the Hf center to initiate polymerization, which might require a temperature of at least 80 • C. Even with the much higher temperature rise (164 vs. 116 • C) relative to [1-Zr(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − , the amount of generated polymer was less (5.2 g vs. 9.3 g). In the case of [3-Hf(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − , the generated polymer was not a slurry but a form dissolved in hexane, indicating that 1-octene incorporation was significant. In fact, the 1-octene content (F C8 ) in the generated copolymer was significantly higher than that in the copolymer generated with  4 ] − (F C8 , 34-35 vs. 33 mol%), and the F C8 values (33-35 mol%) measured for the copolymers generated with the hafnium species were comparable to or even higher than those measured for those generated with half-metallocene titanium species (24-35 mol%), which are known to be excellent for the incorporation of α-olefin. All the half-metallocene Ti species tested afforded polymers in a form dissolved in hexane.    4 ] − as soon as it was prepared prior to transformation to [7-Ti(CH 2 N(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − , yield was improved (5.6 g vs. 4.1 g; entry 13 vs. 14). It is worth mentioning that less amounts of polymers were gained under the identical polymerization conditions when 7-TiMe 2 was activated with [Me(C 18 H 37 ) 2 N-H] + [B(C 6 F 5 ) 4 ] − that was prepared by the conventional salt metathesis method (5.6 g vs. 5.2 g, 5.1 g, 5.1 g, and 4.5 g for the ones of which 1 H NMR spectra are shown in Figure 1a-d, respectively).
A low-molecular-weight polymer was generated with [1-Zr(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − (M w , 32 kDa), and bimodal-molecular-weight distribution was observed with high dispersity value (M w /M n , 13) (entry 1), which inferred that structure of the active species was not persistent during the polymerization especially due to the severe temperature rising. Much-higher-molecular-weight polymers were generated from 3-and 4-Hf species (M w , 700-800 kDa) with much broader multimodal-molecular-weight distributions (M w /M n , 40-80), which also might be due to temperature rising (entries 2-4). From a typical CGC  7), which might be attributed to relatively lower polymerization temperature (90-115 • C); it has been known that the typical CGC 5-Ti species did not exhibit thermal stability above 110 • C [27]. In contrast, half-metallocene 6-Ti and 7-Ti species generated polymers with unimodal-molecular-weight distributions in all cases even when polymerization temperature exceeded 120 • C (M w , 210-340 kDa; M w /M n , 2.3-3.5) (entries 8-15).

Preparation of 5-TiCl(Me) and 6-TiCl(Me)
ZnCl 2 (20.8 mg, 0.153 mmol) was added to 5-TiMe 2 (100 mg, 0.305 mmol) in anhydrous toluene (2 mL). Insoluble ZnCl 2 gradually disappeared, and an almost-clear solution was obtained after stirring for several hours at room temperature. After removing cloudy, insoluble fractions via filtration, the solvent was removed, using a vacuum line, to obtain a yellow solid. An analytically pure compound was obtained by recrystallization in hexane at −30 • C. Yellow cubic-shaped crystals suitable for X-ray crystallography were isolated (80 mg, 76%). 1   After sealing the tube, the activation reaction was monitored, using 1 H NMR spectroscopy. After finishing the 1 H NMR studies, the solution was completely transferred to a vial and diluted with cyclohexane to obtain 5.29 µmol-Hf/g stock solution, which was used in the polymerization studies.

A Representative Polymerization Procedure (Entry 4)
In a glove box, a dried bomb reactor (75 mL) was charged with hexane (15.5 g), 1octene (5.00 g), and trioctylaluminum (12.1 g, 33.0 µmol). The reactor was then assembled and removed from the glove box. The bomb reactor was immersed in a bath at 110 • C. When the temperature inside the bomb reactor reached 65 • C, a catalyst stock solution containing [4-Hf(Me)(NMe(C 18 H 37 ) 2 )] + [B(C 6 F 5 ) 4 ] − (1.00 µmol) was injected. After catalyst feeding, ethylene gas was immediately charged under a pressure of 20 bar. The temperature immediately started to rise, reaching 159 • C in 3 min. The polymerization was performed for 15 min under a constant ethylene pressure of 20 bar while monitoring the temperature in an isothermal bath at 110 • C. In some cases (entries 2-3, 7-10, and 14), polymerization was not initiated until temperature reaching some threshold, in which cases polymerization reaction was performed for 15 min counted from the initiation point. The reactor was cooled with an ice bath and the ethylene gas was vented off. The generated polymer was isolated by removing the solvent under vacuum at 130 • C for 2 h (5.41 g).

X-ray Crystallography
Specimens of suitable quality and size were selected, mounted, and centered in the X-ray beam, using a video camera. Reflection data were collected at 100 K on an APEX II CCD area diffractometer (Bruker), using graphite-monochromated Mo Kα radiation (λ = 0.7107Ǻ). The hemisphere of the reflection data was collected as ϕ and ω scan frames at 0.5 • per frame and an exposure time of 10 s per frame. The cell parameters were determined and refined by using the SMART program. Data reduction was performed by using SAINT software. The data were corrected for Lorentz and polarization effects. Empirical absorption correction was applied by using the SADABS program. The structure was solved by using direct methods and refined with the full matrix least-squares method, using the SHELXTL package and the olex2 program with anisotropic thermal parameters for all non-hydrogen atoms.
The crystallographic data for 4-HfMe 2 ·0.4(toluene) (CCDC# 2070013) that were used in all calculations were as follows: C 87. 6 4 ] − -type species, which are soluble and stable in aliphatic hydrocarbon solvents (e.g., methylcyclohexane and C 6 D 12 ), were highly active in ethylene/1-octene copolymerization performed in aliphatic hydrocarbon solvents (e.g., hexane). Feeding the activated species to a polymerization reactor (instead of in situ generation of the activated species in a bulk reactor) might be advantageous in running a commercial process.