Propylene Polymerization and Deactivation Processes with Isoselective {Cp/Flu} Zirconocene Catalysts

Abstract

Industrially relevant single-site precatalysts used to produce isotactic polypropylene (iPP) include C₂-symmetric {SBI} and C₁-symmetric {Cp/Flu} complexes of group 4 metals. While the latter can produce iPPs with a higher degree of isotacticity, they also suffer from poor productivity compared to their {SBI} counterparts. Several causes for this trend have been suggested—2,1-Regioinsertions are frequently pointed out, as they are suspected to drive the catalyst into a dormant state. While this event does not seem to significantly impact the productivity of {SBI} systems, the influence of these regioerror is poorly documented for isoselective {Cp/Flu} precatalysts. To address this issue, new Ph₂X(Cp)(Flu) (Ph₂X = Ph₂C, FluC, Ph₂Si) proligands (2a–k) and some of the corresponding dichlorozirconocenes (3a–h,k) were synthesized. These new compounds were characterized and tested in homogeneous propylene polymerization at 60 °C and the amounts of regioerrors in the resulting polymers were examined by ¹³C NMR spectroscopy. A possible correlation between poor productivity and a high number of regioerrors was investigated and is discussed. Furthermore, a C-H activation process in the bulky nBu₃C substituent upon activation of 4c (the dimethylated analog of 3c) by B(C₆F₅)₃ has been evidenced by NMR; DFT calculations support this C-H activation as a deactivation mechanism.

1. Introduction

Since the development of isoselective polymerization of propylene with C₁-symmetric cyclopentadienyl-fluorenyl {Cp/Flu}-metallocene catalysts [1,2,3,4], many academic groups have set out to find “the” ideal structure that would produce polypropylene with the highest degree of isotacticity and regiocontrol, well-controlled molecular weight characteristics, and high productivity. This performance race has resulted in the synthesis of a vast number of structures with various substitution patterns [5,6,7,8,9], for the purpose of studying the influence of ligand architecture on the above parameters [10,11,12].

On the other hand, for the series of C₂-symmetric silicon-bridged ansa-bis(indenyl) or {SBI}-metallocene catalysts, significant breakthroughs have been achieved in the last decade, and a number of new structures have emerged that exhibit significantly improved performances (Scheme 1). Among them, of note are metallocene catalysts A [13,14,15,16] and B [17] that bear bulky aromatic 4-indenyl substituents and afford astonishing productivities, with highly crystalline iPP having remarkably low amounts of stereo- and regioerrors. Very recently, zirconocene catalyst C, which incorporates conformationally rigid tripticenyl substituents, has been disclosed to be unprecedentedly efficient in terms of a range of relevant parameters: catalytic performance at high polymerization temperatures (up to 150 °C), stereocontrol (1 propylene misinsertion per 40,000 insertions at room temperature), and molecular weights (>10⁶ g·mol⁻¹) [18].

In a previous contribution [19], we reported the possible origin of the differing productivity and behavior when comparing isoselective {SBI}- [20] and {Cp/Flu}-type [21,22,23,24,25,26,27,28,29] propylene polymerization precatalysts (Scheme 2). Indeed, the former catalysts are typically one order of magnitude more productive than the latter systems (15·10⁴ kgPP·mol⁻¹·h⁻¹ with {SBI}-1 vs. 14·10³ kgPP·mol⁻¹·h⁻¹ with {Cp/Flu}-1 at 60 °C and 5 bar), both under homogeneous and heterogeneous (supported) conditions. Yet, the kinetic data [19] unequivocally demonstrated that {Cp/Flu}-based systems (e.g., {Cp/Flu}-2) are intrinsically more active (in terms of propagation rates) and less prone to secondary (2,1-) propylene misinsertions than the {SBI}-systems. However, the very few resulting regiodefects appeared to be much more deleterious than for their {SBI} analogs.

In the present contribution, we aimed to extend the understanding of the impact of structural and/or electronic parameters on the regioselectivity of {Cp/Flu}-type isoselective catalytic systems, and decifer possible relationships between regiodefects and catalyst productivity. Thus, we considered a new series of modifications in order to: (1) vary the bulkiness of the 3-R-Cp substituents; (2) assess the impact of the nature and steric hindrance of the substituents in the 2,7- vs. 3,6-positions in the Flu platform; and (3) compare the constrained Ph₂C- vs. Ph₂Si-bridged {Cp/Flu} ligand platforms. This study also reports on the syntheses of these new precatalysts and their structural characterization both in solution and in the solid state. The performances of the new complexes were explored, after activation with MAO, in homogeneous propylene polymerization.

2. Results and Discussion

Syntheses of Proligands. In our previous studies [18,19,20], we used an efficient and scalable procedure for the synthesis of proligands {R¹R²C-(Flu)(Cp)}H₂ via nucleophilic addition of substituted fluorenyl anions onto diversely multisubstituted fulvenes. Following this protocol, an extended series of new constrained (aryl)₂C-bridged proligands (2a–j) was here prepared in good yields (Scheme 3) and isolated as air-stable solids. The Ph₂Si-bridged 2k was synthesized using a modified procedure reported for {R₂Si(Flu)(Cp)}H₂, where R = Me or Ph [30,31,32,33,34]. After optimization of the conditions and purification by column chromatography, proligand 2k was isolated with an acceptable yield (47%) as a crystalline solid. Proligands 2a–k were characterized by ¹H and ¹³C NMR spectroscopy, ASAP mass-spectrometry, elemental analysis (see Protocol section in the Supporting Information), and X-ray crystallography (see the Supporting Information Figures S77–S87; Figure 1 for 2i).

Synthesis and Structure of Zirconocene Complexes. In order to prepare zirconocene dichlorides, regular salt metathesis reactions between anhydrous ZrCl₄ and ligands dianions, generated in situ in Et₂O, were undertaken (Scheme 4). Thus, analytically pure zirconium complexes 3a–h were isolated in reasonable yields after multiple crystallization attempts from heptane or heptane/CH₂Cl₂ mixtures as characteristically pink, microcrystalline materials. Following the same procedure, 3k was isolated as a yellow microcrystalline powder. The low isolated yields found for some of the metallocene products resulted from multiple crystallization attempts undertaken to achieve the satisfactory purity needed for characterization and polymerization experiments.

Despite numerous attempts, the respective zirconium complexes derived from proligands 2i and 2j could not be isolated using the standard conditions shown in Scheme 4. Even upon using the conditions reported for the synthesis of complexes {FluC(3-tBu-Cp)₂}ZrCl₂ and {FluC(Cp)₂}ZrCl₂ incorporating the same FluC bridge (deprotonation of the proligands with nBuLi at −40 °C, followed by the salt-metathesis reaction with ZrCl₄ at −70 °C in CH₂Cl₂ and then under stirring for 2 h at −20 °C, see [35]), these experiments did not result in the isolation of the desired complexes, and complex mixtures of unidentified products were systematically obtained.

The solution structures of 3a–h and 3k were studied using ¹H and ¹³C NMR spectroscopy and were found to be consistent with the C₁-symmetry of these species. Analysis of the ¹³C NMR data (see Figures S45, S47, S49, S51, S53, S55, S58, S60, S62 and S64 in the Supporting Information) revealed particularly upfield C9-fluorenyl carbon signals (δ_C 74.5–80.8 ppm), which is indicative of a reduced coordination hapticity (η⁵→η³) of the fluorenyl ligands maintained in solution [36,37]. The iASAP-MS spectra for most of the complexes are also available in the Supporting Information (Figures S72–S79).

Single crystals of 3b–e,g and h suitable for X-ray diffraction studies (Figure 2, Figure 3, Figure 4, Figure 5, Figure 6 and Figure 7) were grown from CH₂Cl₂/n-hexane (1:1 v/v) solutions at room temperature. In the solid state, the molecules exhibited geometrical parameters (

Table 1. Selected bond lengths (Å) and angles (°) for metallocene complexes 3b–e, g and h.

	{Cp/Flu}5H	3b	3c	3d	3e	3g	3h
Zr–C(1)	2.442 (4)	2.4450 (17)	2.4391 (14)	2.451 (4)	2.443 (8)	2.440 (5)	2.442 (6)
Zr–C(2)	2.492 (3)	2.4745 (18)	2.4720 (15)	2.484 (4)	2.489 (8)	2.459 (5)	2.468 (6)
Zr–C(3)	2.630 (4)	2.6079 (18)	2.6431 (15)	2.645 (4)	2.594 (8)	2.582 (5)	2.587 (6)
Zr–C(4)	2.547 (4)	2.5284 (18)	2.5706 (15)	2.558 (4)	2.514 (8)	2.554 (6)	2.525 (6)
Zr–C(5)	2.457 (4)	2.4384 (18)	2.4408 (15)	2.433 (4)	2.429 (8)	2.463 (5)	2.459 (6)
Zr–Cpcent	2.204	2.242	2.204	2.209	2.180	2.190	2.181
Zr–C(9)	2.426 (3)	2.4108 (18)	2.4229 (15)	2.445 (4)	2.403 (8)	2.407 (5)	2.420 (6)
Zr–C(10)	2.497 (4)	2.5128 (18)	2.5020 (15)	2.510 (4)	2.500 (8)	2.505 (5)	2.488 (6)
Zr–C(11)	2.666 (4)	2.6740 (18)	2.6621 (15)	2.683 (4)	2.686 (8)	2.681 (5)	2.695 (6)
Zr–C(12)	2.712 (4)	2.6633 (18)	2.6732 (15)	2.707 (4)	2.670 (8)	2.688 (5)	2.660 (6)
Zr–C(13)	2.588 (4)	2.5083 (18)	2.5263 (15)	2.552 (4)	2.514 (8)	2.524 (4)	2.563 (6)
Zr–Flucent	2.269	2.188	2.240	2.272	2.245	2.250	2.256
Cpcent–Zr–Flucent	118.03	118.15	117.80	116.85	117.93	118.45	118.61
C(1)–C(6)–C(9)	98.60 (3)	99.13 (14)	98.73 (11)	99.20 (3)	99.10 (6)	99.20 (6)	100.10 (5)
C(7)–C(6)–C(8)	104.00 (4)	104.33 (14)	104.82 (12)	105.60 (3)	104.30 (7)	104.50 (4)	101.10 (5)

) essentially similar to those observed in a series of similar zirconocene dichlorides incorporating R₂C-bridged {Cp/Flu} ligands described by our group (e.g., {Ph₂C(3,6-tBu₂-Flu)(3-tBu-Cp)}ZrCl₂ ({Cp/Flu}^5H)) [12,25,26,27] and several others. [2,7,8,9,38,39,40] The coordination of the central five-membered ring of the fluorenyl ligand in most complexes deviated slightly from η⁵ towards η³, as revealed by the differences in the Zr–C(ring) distances (ca. 0.3 Å between the shortest and the longest bond lengths); this is consistent with the above comment in solution. The Cp_cent–Zr–Flu_cent bite angles were found in a rather narrow range of values (116.85–118.61°) and were similar to the corresponding values in the Ph₂C-bridged metallocene {Cp/Flu}^5H (118.03°) [25].

Propylene polymerization. Ansa-metallocenes 3a–h,k, in combination with MAO, were evaluated in the propylene polymerization (toluene solution, 5 bar constant pressure, T_pol = 60–100 °C). Each polymerization experiment was independently repeated three times under the same conditions, revealing good reproducibility in terms of the activity (gas uptake), productivity (polymer yield), and physicochemical properties (M_w, M_n, T_m, isotacticity) of the isolated polymer. For benchmarking purposes, the performance of the reference zirconocene {Cp/Flu}-2 (Scheme 2) was explored under identical conditions. Due to a high exothermicity at [Zr]₀ = 10.0 µM, the polymerization experiments with {Cp/Flu}-2 were conducted at lower concentrations (5.0 µM). Selected polymerization results are summarized in

Table 2. Propylene polymerization experiments ^a.

Entry	Precat.	Tpolymb	Prod (kgPP·mol·h−1)	Tm (°C) c	Tcrystc (°C)	Mnd (×103)	Mw/Mnd	[m]4 e (%)	1,2-ins e (%)	2,1-ins e (%)	1,3-ins e (%)	nBu termini e (%)
1	{Cp/Flu}-2	60 (68)	50,700 f	131.1	93.6	50.8	2.5	82.1	99.8	<0.02	0.1	0.1
2	3a	60 (66)	35,500	135.8	102.6	11.4	2.1	86.5	99.3	<0.1	0.3	0.2
3	3b	60 (65)	5300	141.1	106.2	15.4	2.1	88.6	99.2	0.2	0.4	0.3
4	3c	60 (65)	241	137.1	106.2	-	-	-	-	-	-	-
5	3d	60 (64)	0	-	-	-	-	-	-	-	-	-
6	3e	60 (76)	33,600	132.9	100.3	13.7	2.1	81.6	99.3	<0.02	0.3	0.04
7		60 (67)	16,700 g	139.8	104.5	21.7	2.1	86.2	99.5	<0.1	0.3	≈0.1
8	3f	60 (66)	13,300	148.4	112.6	17.6	2.3	91.7	99.6	<0.02	0.1	< 0.1
9	3g	60 (66)	12 700	136.1	102.3	21.0	2.3	82.9	99.4	0.1	0.3	0.1
10	3h	60 (68)	17,200	n.o.	n.o.	24.9	2.1	54.0	99.7	<0.02	0.1	<0.02
11	3k	60 (64)	1870	131.7	119.8	13.4	2.2	31.9	99.6	<0.02	<0.01	<0.02
12		100 (102)	640	n.o.	n.o.	3.4	1.4	-	-	-	-	-

^a Polymerization conditions, otherwise stated: 300 mL high-pressure glass reactor; solvent: toluene, 150 mL; P (propylene) = 5 bar; Tpolym = 60 °C, [Zr]0 = 10.0 µM, [Al]/[Zr] = 5000; n.o. = not observed. ^b Data in brackets refer to the maximum temperature reached in the reactor. ^c Determined by DSC. ^d Determined by GPC/SEC. ^e Determined by ¹³C NMR spectroscopy. ^f [Zr]0 = 5.0 µM, [Al]/[Zr] = 10,000. ^g [Zr]0 = 5.0 µM, [Al]/[Zr] = 5000.

(and reported in Table S3).

Several trends could be specifically drawn. However, these activity data should be considered with care, since the polymerization reactions carried out with highly active systems, such as that based on {Cp/Flu}-2, were quite exothermic, even with low precatalyst loadings. As a result, the temperature of the reaction mixtures could hardly be controlled, often outreaching configured values after a few minutes (see Table 2). The trends evidenced the flexibility and limits of the polymerization processes with these catalytic systems:

-

The most active precatalyst within the given series of propylene polymerization precatalysts appeared to be the reference metallocene {Cp/Flu}-2. For instance, in a typical experiment performed at 60 °C (entry 1), the productivity of the latter catalyst was found to be 1.5–10 times higher (50,700 kg PP·mol⁻¹·h⁻¹) than those observed for all other precursors (up to 35,500 kg PP·mol⁻¹·h⁻¹).

-

The second by productivity (35,500 kg PP·mol⁻¹·h⁻¹) appeared to be the less sterically constrained system incorporating 3a, which is analogous to {Cp/Flu}-2 but has no Me substituent at the position 5 of the Cp ligand.

-

Increasing the bulkiness of the 3-R substituent at the Cp ligand (R = tBuMe₂C in 3b and R = 1-methylcyclohexyl in 3e) resulted in a significant drop in productivity (5300 kg PP·mol⁻¹·h⁻¹ and 16,700 kg PP·mol⁻¹·h⁻¹; entries 3 and 7, respectively). At the same time, metallocene systems 3c and 3d, both incorporating the bulky aliphatic 3-nBu₃C substituent at the Cp ligand, were unexpectedly found to be very poorly or completely inactive (entries 4 and 5, respectively). The latter result can stem from deactivation involving a nBu group of the 3-nBu₃C substituent (vide infra).

-

Keeping the 3-tBu substituent at the Cp ligand while altering the nature of the 2,7-/3,6- substituents also resulted in catalysts featuring inferior productivities. Thus, 3f–3h afforded comparable productivities in the range of 12,700–17,200 kg PP·mol⁻¹·h⁻¹ (entries 8–10).

-

Regarding the molecular weights of the PPs produced, all active systems yielded similar products, with M_n lying in the range of 11.4–24.9 kg·mol⁻¹, showing no specific trend and being lower than that of the PP obtained with the reference metallocene catalyst {Cp/Flu}-2 (50.7 kg·mol⁻¹). The polydispersity values (PDI = 2.1–2.5) were all consistent with the single-site behavior of these catalyst systems.

-

The stereoregularities of the PPs obtained from {Cp/Flu}-based catalysts can vary over a broad range, depending on the substituents both on the Cp and Flu moieties [10,11,18,19,20]. The new metallocenes with bulky 3-R substituents on the Cp ligand (3a,b,e) afforded polypropylenes featuring pentad [m]⁴ values (86.2–88.6%) higher than those observed with {Cp/Flu}-2 (82.1%). These stereoregularity levels are also reflected in the corresponding T_m values determined for these polymers (139.8–141.1 vs. 131.1, respectively). The highest stereoselectivity ([m]⁴ = 91.2%) within the series was achieved with metallocene catalyst 3f with the 2,3,6,7-tetra-substituted version of fluorenyl ligand (Oct), which afforded PP with T_m of 148.4.

-

Being the 2,7-Mes₂-Flu-substituted analogue of {Cp/Flu}-2, system 3h quite unexpectedly afforded PP with a low isotacticity index ([m]⁴ = 54.0%; see the Supporting Information Table S4). This result cannot be explained by the higher bulkiness of the mesityl substituents.

-

The incorporation of a Si-bridge (entry 11) appeared to confer a negative effect on all parameters (productivity, molecular weight, stereoregularity (Table S4), and T_m), as evidenced upon comparing the results obtained with 3k and its Ph₂C-bridged analogue {Cp/Flu}-2. In order to improve the productivity, a polymerization experiment was conducted at 100 °C (entry 12), but it resulted in an almost total loss of activity; this possibly reflects significant deactivation or even degradation of the catalyst at this temperature. The viscous material recovered from these attempts was found to be a mixture of oligomers exhibiting no melting transition. In a previous contribution, Chen and Rausch et al. [30] explained that the loss of syndioselectivity and poorer productivity of {Me₂Si(Flu)(Cp)}ZrCl₂ with respect to the Me₂C-bridged congener were due to steric hindrance of metal center. Thus, the larger Cp_cent-Si-Flu_cent bite angle, which is ca. 10° greater than that in the carbon-bridged counterpart, makes the coordination sphere both less accessible and less selective towards the coordination of monomer.

Influence of metallocene structure on regioselectivity. The deactivation of MgCl₂-supported Ziegler–Natta catalyst systems used for propylene polymerization via 2,1-misinsertions and reactivation of their dormant state by the deliberate introduction of H₂ as a chain-transfer agent (CTA) have been reported by Cipullo et al. [41]. In our previous contribution [19], we established a marked impact of regioirregular (2,1-, secondary) insertions on the stability and activity of propagating species in the polymerization of propylene with {Cp/Flu}-based systems. In the latter case, the formation of “dormant” Zr–sec-alkyl species by 2,1-misinsertion of the α-olefin (Scheme 5), reluctant to further insertions, has been also shown to have a critical role for deactivation process. The existence of Zr–sec-alkyl species can be indirectly evidenced from the ¹³C NMR spectroscopic data with the presence of head-head and tail-tail sequences in the resulting polymer. The presence of nBu chain-ends in the polymers, resulting from protonolysis (transfer to H₂ or acidic hydrolysis) of “dormant” Zr–sec-alkyl species, can be hence considered as a fingerprint of massive deactivation of the catalyst. The reorganization of “dormant” Zr–sec-alkyl species though a multistep process (1,3-insertion) [42,43] comprising a β-H elimination reaction from the methyl group of the ultimate unit and then 1,2-reinsertion of the σ-olefin obtained in the previous step, could constitute a re-activation process resulting in restitution of catalytic activity (Scheme 5).

On attempting to find out the connection between the lower productivities of some {Cp/Flu}-based systems and their propensity towards affording regioirregular insertions, a closer inspection of the ¹³C{¹H} NMR spectra of the PPs produced with 3a–c,e–h and the reference system {Cp/Flu}-2 was carried out (Table 2). For a more complete set of data, a complementary analysis was conducted with the polymers obtained with some previously reported analogous {Cp/Flu}-based systems (Scheme 6;

Table 3. Propylene polymerization data obtained with precatalysts from Scheme 6 ^a.

Entry	Precat.	[Zr]0 (µmol.L−1)	[Al/[Zr]	Prod (kgPP·mol·h−1)	[m]4 b (%)	1,2-ins b (%)	2,1-ins b (%)	1,3-ins b (%)	nBu Termini b (%)
1	{Cp/Flu}-1	10.0	5000	1100	90.5	99.8	<0.02	0.1	<0.02
2	{Cp/Flu}-3Me	10.0	5000	15,470	88.1	99.8	<0.02	<0.1	<0.02
3	{Cp/Flu}-3Et	10.0	5000	9730	89.5	99.8	<0.02	<0.1	<0.02
4	{Cp/Flu}-3Ph	20.0	5000	2000	53.8	99.5	<0.02	<0.01	<0.02
5	{Cp/Flu}-4Me	10.0	5000	29,600	90.7	99.8	<0.02	0.02	<0.02
6	{Cp/Flu}-4Et	10.0	5000	26,260	90.8	99.7	<0.02	0.02	<0.02
7	{Cp/Flu}-5	10.0	5000	1550	80.9	99.6	<0.02	0.09	0.2
8	{SBI}-1	2.0	25,000	62,700	93.6	98.6	0.26	0.5	0.05
9	{SBI}-2	2.0	25,000	27,900	83.4	99.2	0.45	0.05	<0.1

^a Polymerization conditions, otherwise stated: 300 mL high-pressure glass reactor; solvent: toluene, 150 mL; P (propylene) = 5 bar; T_polym = 60 °C. ^b Determined by ¹³C NMR spectroscopy.

, entries 1–7) [11,25,26,27]. The respective resonances derived from the 2,1- and 1,3-insertions and those from the terminal nBu chain-ends were identified, quantified, and used to plot the combined bar diagram (Figure 8). Yet, regrettably, no correlation could be identified between the productivity and the content of the titled sequences from regioirregular insertions or nBu chain-ends. Nevertheless, it appears that minor changes (e.g., 3b or 3a vs. {Cp/Flu}-2) in the {Cp/Flu}-metallocene structure can induce dramatic effects on regio- and stereo-control as well as on productivity.

In striking contrast with the {Cp/Flu}-systems, the {SBI}-based counterparts feature a much higher propensity towards affording 2,1- and 1,3-insertions, thus resulting in much larger contents of the corresponding regioerrors (Table 3, entries 8 and 9), which are apparently not deleterious for the productivity of these systems. In fact, in that case, the M–sec-alkyl species are still reactive and capable of either undergoing further regular 1,2-insertions of propylene or regenerating an active hydrido species upon β-H elimination [19].

Deactivation of the catalytic system based on 3c. Complexes {Ph₂C(2,7-tBu₂-Flu)(3-nBu₃C-Cp)}ZrCl₂ (3c) and {Ph₂C(3,6-tBu₂-Flu)(3-nBu₃C-Cp)}ZrCl₂ (3d), in combination with MAO, showed particularly low productivity in the polymerization of propylene that could not be rationalized on the basis of steric factors. To address this phenomenon, we aimed at understanding the role of activation conditions as well as of ion-pairing effects. The introduction of excess MAO necessary for the activation of 3c is likely to generate a complex NMR spectrum that would not allow for observing all the phenomena involved, especially in the aliphatic region. Therefore, the generation of the corresponding ion pair was carried out upon using molecular activators such as B(C₆F₅)₃ [27].

The dimethyl-zirconocene 4c was selectively prepared from the parent 3c by a reaction with 2 equiv. of Grignard’s MeMgBr (see SI). Further treatment of 4c with 1 equiv. of B(C₆F₅)₃, carried out on an NMR scale in toluene-d₈ at −50 °C, resulted in quantitative and selective generation of the corresponding ion-pair 4c-MeB(C₆F₅)₃ (Scheme 7). All attempts to grow crystals of 4c-MeB(C₆F₅)₃ failed thus far. The ¹H NMR spectrum recorded at this temperature (Figure 9a) showed the complete disappearance of the Zr–Me signals (δ_H −1.09 and −0.99 ppm) and the appearance of two equal intensity resonances at δ_H −0.38 and −0.22 ppm corresponding to the Zr–Me and B–Me groups, respectively. The ion pair 4c-MeB(C₆F₅)₃ was found stable in the temperature range of −50–+10 °C and was characterized by multinuclear NMR spectroscopy. For example, the ¹H and ¹³C{¹H} NMR spectra of 4c-MeB(C₆F₅)₃, recorded at −50 °C (Figures S65 and S66, respectively), each contained a single set of signals. Furthermore, the difference in the chemical shifts of the meta- and para-F resonances, |Δδ (m,p-F)| = 4.8 ppm in the ¹⁹F{¹H} NMR spectrum of 4c-MeB(C₆F₅)₃, is in agreement with the inner-sphere ion-pairing (ISIP) nature of that species. Typically, the corresponding |Δδ (m,p-F)| values for ISIPs are greater than 3.5 ppm, while those for OSIPs are smaller than 3.0 ppm; see: [48]. This observation is consistent with the fact that this ion pair exists as a single isomer, in which the coordinated MeB(C₆F₅)₃⁻ anion occupies the less hindered lateral coordination site [28,29]. This is in contrast with less bulky ion-pair analogues derived from similar C₁-symmetric metallocenes, which consist of mixtures of two isomers that feature different coordination of the MeB(C₆F₅)₃⁻ anion (in the more open and the less hindered lateral coordination site, respectively).

Above 15 °C, a new series of signals appeared in the ¹H NMR spectrum (Figure 9b), evidencing a new species whose formation was accompanied by concomitant formation of methane (δ_H 0.19 ppm). The reaction was completed after 24 h at room temperature, and the recorded ¹H NMR spectrum contained only one new series of resonances (Figure 9c). Close inspection of the aliphatic region of this ¹H NMR spectrum allowed for the identification of the nature of the product—the new cationic Zr–alkyl complex 5c-MeB(C₆F₅)₃ formed by an intramolecular C–H activation reaction (Scheme 7) between one nBu group and Zr–Me, eventually releasing 1 equiv. of methane. Among the three possible products A–C arising from a C–H activation site in the alkyl chain of the nBu group, the selective formation of regioisomer C took place as judged from the NMR data. Additional 2D (COSY, HSQC) NMR spectroscopic techniques allowed for a proper assignment of all signals (Figures S70 and S71). The Zr–(2-(2-ethyl)ethyl) fragment appears in the ¹H NMR spectrum (Figure 9c) as a set of characteristic signals: a multiplet (δ_H −1.62 ppm) for the methine hydrogen H^c (Zr–CH), two multiplets (δ_H 0.01 and 1.41 ppm) for the diastereotopic hydrogens H^b and H^b′ (CHHCH₃), a triplet (δ_H 0.36 ppm; ³J_H-H = 7.0 Hz) for the hydrogens H^a (CH₃), and a triplet (δ_H 1.74 ppm; ³J_H-H = 14.0 Hz) and a doublet of doublets (δ_H 2.45 ppm; ²J_H-H = 6.5 Hz, ³J_H-H = 14.0 Hz) for the diastereotopic hydrogens H^d and H^d′ (−CCHHCH–Zr). In the ¹³C{¹H} NMR spectrum, four distinct resonances are observed, including a distinctive one for the metallated carbon (Figure S70) [49]: δ_C 80.6 (Zr–CH), 32.0 (CH₂CH₃), 17.1 (CH₂CH₃), and 59.5 (−CCH₂CH–Zr) ppm. Furthermore, in the ¹⁹F{¹H} NMR spectrum of 5c-MeB(C₆F₅)₃ (Figure S69), the difference in the chemical shifts of the meta- and para-F resonances, |Δδ (m,p-F)| = 5.0 ppm, is in line with the ISIP nature of this species [48].

Based on the above results, we surmised that a similar rapid deactivation process is operational with 3c and 3d in the presence of MAO, thus affording at some point very stable sterically congested ion pairs with [Me-MAO]⁻ anions, which are further reluctant towards reactions with propylene.

Theoretical studies of a possible deactivation pathway. To gain further insight into this deactivation phenomenon, DFT calculations were conducted using as model the putative cationic methyl-zirconocene complex [4c-Me]⁺ (Scheme 8) obtained upon activation with MAO or B(C₆F₅)₃, namely [{Ph₂C(2,7-tBu₂Flu)(3-nBu₃-C₅H₂)}Zr–Me]⁺ (B3PW91 level, see the Supporting Information for details). The objectives of these non-exhaustive computations were to assess the total electronic energy profiles for two possible concurrent processes, that is: (a) regular coordination/insertion of propylene into the Zr–Me bond of [4c-Me]⁺ to give the respective insertion species [4c-iBu]⁺, vs. (b) intramolecular C-H activation in the pendant nBu₃C group by the cationic metal center generating metallacyclic species (most possible isomers B and C, Scheme 7), and consecutive reactions of the latter species with propylene. Despite metallocene species with fully substituted Cp and Flu rings were considered in the computations, some simplifications/assumptions were made to reduce the calculation costs: (i) the influence of the counter anion was disregarded, and only “naked” cationic species were considered; and (ii) the methyl or iso-butyl group (representing the polymeryl chain) resides in the opposite direction to the bulky nBu₃C substituent. In addition, when reactions with propylene were conceived, it was assumed that (iii) the insertion of propylene proceeds in a primary (“pr”) fashion, and (iv) finally, as usually considered for isospecific polymerization of propylene mediated by C₁-symmetric {Cp/Flu}-metallocene catalysts, the molecule of monomer coordinates to the cationic zirconium center with its methyl group positioned “head down” into the free space in the central region of the fluorenyl ligand, while the methylene group is directed toward the less-congested quadrant (opposite to the methyl or iso-butyl groups/polymeryl chain).

The regular propagation pathway was assessed at the first insertion step, that is, conversion of the π-complex [4c-Me]⁺ (Scheme 8) to isobutyl species [4c-iBu]⁺. As expected, this process was found to be favorable on both kinetic and thermodynamic grounds, showing the reasonable calculated activation barrier ΔE^≠_{TS1 prsi} = 12.1 kcal·mol⁻¹ for the primary insertion of propylene with si face, which should result in isotactic polymer. Barriers in the range 7.1–8.6 and 10.0–15.0 kcal·mol⁻¹ for the first and second primary insertions of propylene into the Zr–C bond, respectively, were calculated for the cationic species [{R¹₂C(3,6-tBu₂Flu)(3-tBu-5-R²-C₅H₂)}Zr–(polymeryl)]⁺ (R¹ = H, Ph; R² = Me, Ph); see reference [10].

The metallation (C–H activation) of the nBu group in the nBu₃C substituent, requiring a proximal Zr–R function and accompanied by a concomitant alkane (methane or iso-butane) elimination, was computed to occur through somewhat higher but still accessible transition states both for R = Me in [4c-Me]⁺ (ΔE^≠_TS2 = 19.7 and 20.2 kcal⋅mol⁻¹) and for R = iBu in [4c-iBu]⁺ species (ΔE^≠_TS2′ = 18.4 and 19.4 kcal⋅mol⁻¹). While the C–H activation reaction from [4c-iBu]⁺ species was computed to be essentially endothermic (8.7 and 6.5 kcal⋅mol⁻¹), the same process from [4c-Me]⁺ species appeared to be thermodynamically favored (−3.9 and −1.7 kcal⋅mol⁻¹). This suggests that the C–H activation phenomenon leading to isomers B and C can preferably occur for the cationic complex initially formed at the precatalyst activation stage and, especially, in the absence of propylene. In line with the experimental results, the formation of isomer C was found computationally to be more favored both kinetically (ΔΔE^≠ = 0.5–1.0 kcal⋅mol⁻¹) and thermodynamically (ΔΔE = 2.2 kcal⋅mol⁻¹), regardless the nature of [4c-R]⁺ precursor. Further propylene insertions in the metallacyclic isomers B and C (to give the corresponding products P) were computed to be favored thermodynamically (ΔE = −11.4–−5.5 kcal⋅mol⁻¹), and connected through accessible transition states (ΔE^≠ = 12.9–17.3 kcal⋅mol⁻¹).

These results indicate that the formation of metallacyclic species by intramolecular C–H activation is a smooth process and likely to take place during the very first steps of polymerization. However, this process seems to be not deleterious for the posterior reactivity with propylene and could lead to the formation of larger metallacyclic propagating species. Apparently, deactivation takes effect on later steps through a series of intramolecular C–H activation processes successively involving all the nBu groups of the nBu₃C substituent and eventually resulting in a steric obstruction of the active site.

3. Conclusions

New se C₁-symmetric ansa-zirconocenes with multisubstituted {R₂E-(Flu)(Cp)}²⁻ ligands were prepared. In most cases, the activation of these complexes with MAO gave highly active and isoselective propylene polymerization catalytic systems. The highest productivity, comparable to that of the {Cp/Flu}-2 reference system, was observed with 3a and 3e, with both incorporating bulky 3-R-Cp substituents (tBuCMe₂C and MeCy, respectively). The complete inactivity of the 3-nBu₃C-Cp substituted 3c and 3d in propylene polymerization was rationalized from NMR spectroscopic studies on a model system incorporating the [4c]⁺[MeB(C₆F₅)₃] ion pair and from DFT calculations. Both indicated a concurrent decomposition route involving intramolecular C-H activation in the nBu₃C-groups. The most isoselective system within the whole series appeared to be 3f, bearing an octamethyloctahydrodibenzofluorenyl-based ligand. Quite unexpectedly, the 2,7-Mes₂-Flu substitution in 3h appeared to be detrimental for stereocontrol and resulted in much poorer isotacticity. The impact of different substitution patterns on the occurrence of regioerrors involving both 2,1- and 1,3-insertions was sought in this study. However, no distinct correlation could be identified thus far between them nor between the observed productivities and the content of the regioirregular sequences in the obtained polymers. Further investigations aimed at clarifying the origin of the different catalytic behaviors among this series of metallocene catalysts are underway in our laboratories.