Fusing Carbocycles of Inequivalent Ring Size to a Bis(imino)pyridine-Iron Ethylene Polymerization Catalyst: Distinctive Effects on Activity, PE Molecular Weight, and Dispersity

The 4,6-bis(arylimino)-1,2,3,7,8,9,10-heptahydrocyclohepta[b]quinoline-iron(II) chlorides (aryl = 2,6-Me2C6H3Fe1; 2,6-Et2C6H3Fe2; 2,6-i-Pr2C6H3Fe3; 2,4,6-Me3C6H2Fe4; and 2,6-Et2-4-Me2C6H2Fe5) have been prepared in good yield by a straightforward one-pot reaction of 2,3,7,8,9,10-hexahydro-1H-cyclohepta[b]quinoline-4,6-dione, FeCl2·4H2O, and the appropriate aniline in acetic acid. All ferrous complexes have been characterized by elemental analysis and FT-IR spectroscopy. In addition, the structure of Fe3 has been determined by single crystal X-ray diffraction, which showed the iron center to adopt a distorted square pyramidal geometry with the saturated sections of the fused six- and seven-membered carbocycles to be cis-configured. In combination with either MAO or MMAO, Fe1–Fe5 exhibited exceptionally high activities for ethylene polymerization (up to 15.86 × 106 g(PE) mol−1 (Fe) h−1 at 40°C (MMAO) and 9.60 × 106 g(PE) mol−1 (Fe) h−1 at 60°C (MAO)) and produced highly linear polyethylene (HLPE, Tm ≥ 128°C) with a wide range in molecular weights; in general, the MMAO-promoted polymerizations were more active. Irrespective of the cocatalyst employed, the 2,6-Me2-substituted Fe1 and Fe4 proved the most active while the more sterically hindered 2,6-i-Pr2Fe3 the least but afforded the highest molecular weight polyethylene (Mw: 65.6–72.6 kg mol−1). Multinuclear NMR spectroscopic analysis of the polymer formed using Fe4/MMAO at 40°C showed a preference for fully saturated chain ends with a broad bimodal distribution a feature of the GPC trace (Mw/Mn = 13.4). By contrast, using Fe4/MAO at 60°C a vinyl-terminated polymer of lower molecular weight (Mw = 14.2 kg mol−1) was identified that exhibited a unimodal distribution (Mw/Mn = 3.8). Moreover, the amount of aluminoxane cocatalyst employed, temperature, and run time were also found to be influential on the modality of the polymer.

As an alternative strategy in bis(imino)pyridine ligand design, our group has recently explored the fusion of carbocycles to the central pyridine unit in (Figure 1) as a means to form well-defined Fe-/Co-based complexes bearing singly [22][23][24][25][26][27][28][29][30] and doubly fused derivatives (e.g., B, C, D, and E in Figure 1) [31][32][33][34][35][36][37][38]. Indeed, the fused ring size has been shown to be pivotal to the catalytic activity and polymeric properties [6,[36][37][38]. Of particular note, iron-containing B [31], C [32], and D [36] can produce strictly linear polyethylenes with a wide range of molecular weights, end-group types, and activities (up to 10 7 g(PE) mol -1 (Fe) h -1 ) [6,36]. For example, six-membered B produced the lowest molecular weight polymer (8 kg mol -1 ) [31], while seven-and eight-membered C and D showed a predilection towards higher molecular weight vinyl-polyethylenes (up to 188 kg mol -1 ) [32,36]. In light of these performance differences, it is probable that the ring size impacts on properties such as the ring flexibility/tension and the overall steric properties of the chelating ligand which in turn influences the propagation and chain transfer steps of polymerization [6]. As a more recent development, we have demonstrated that cobalt-containing E (Figure 1), incorporating both six-and seven-membered carbocycles, not only showed the highest catalytic activity of the cobalt-containing A-E series but also generated valuable vinyl-terminated PE waxes with narrow molecular weight distributions [38]. Significantly, such low molecular weight polymers provide promising raw materials for the production of functional polymers, copolymers, and coating materials [32,38].

Results and Discussion
The FT-IR spectra of Fe1-Fe5 revealed stretching frequencies in the range 1604-1614 cm -1 that are quite typical of bound N-imine groups; no absorptions corresponding to the free diketone were detectable [36][37][38][39]. Further support for the structural identity of Fe1-Fe5 was provided by the microanalytical data which were in complete agreement with elemental compositions for complexes of general formula (N,N,N)FeCl 2 .

Ethylene Polymerization.
To investigate the aptitude of Fe1-Fe5 to act as precatalysts for ethylene polymerization, MAO and MMAO were selected as cocatalysts to allow two parallel studies to be performed. Both these types of aluminoxane have been previously shown as among the most effective for precatalyst activation in iron-based polymerization catalysis [4,6,11,12,26,27,31,32,36].

Catalytic Evaluation of Fe1-Fe5
Using MMAO. To establish the optimum reaction conditions for these MMAO-activated polymerizations, Fe4 was chosen as the precatalyst for initial assessment and the amount of aluminoxane cocatalyst (Al : Fe molar ratio), temperature, and run time systematically varied with the ethylene pressure set at either 1, 5, or 10 atm.
Firstly, the runs were performed at 1 atm C 2 H 4 and the effect of Al : Fe molar ratio on the performance of Fe4/MMAO examined (runs 1-5, Table 2). With the reaction temperature kept at 20°C, the tests were undertaken using different Al : Fe ratios of 1000, 1500, 2000, 2500, and 3000; the optimum activity of 1:82 × 10 6 gðPEÞ mol -1 ðFeÞ h -1 was observed at a value of 2500 (run 4, Table 3). The polyethylenes for all five runs were of low molecular weight falling in the range 5.8-15.7 kg mol -1 ; there was no evidence for any short chain oligomers (e.g., from C 4 to C 32 ). It was also noted that the lowest molecular weight polymer within this range corresponded to the polymerizations conducted with higher Al : Fe molar ratios, a finding that can be linked to the higher rate of chain transfer from the iron active species to aluminum on increasing the amount of alkyl aluminum reagent [26, 27, 32, 36-38, 40, 41]. The influence of temperature was then investigated with the Al : Fe molar ratio retained at 2500. By varying the test temperature from 10 to 60°C (runs 4 and 6-10, 1:82 × 10 6 gðPEÞ mol -1 ðFeÞ h -1 was observed at 20°C. It was also evident that the molecular weight of the polyethylene gradually decreased from 9.5 to 1.2 kg mol -1 as the temperature was raised which can be credited to increased chain transfer at higher temperature [4,22,26,27,[32][33][34][35][36][37][38]. Meanwhile, the molecular weight distribution narrowed as the temperature was raised (M w /M n : from 13.2 to 2.2), an observation also noted at higher ethylene pressure (vide infra) and elsewhere [4,32,36]. Secondly, Fe4/MMAO was also screened at 10 atm C 2 H 4 ; the experimental findings are compiled in Table 3. Bearing in mind the temperature/activity correlations seen at 1 atm C 2 H 4 , a similar study was performed at 10 atm with the Al : Fe molar ratio kept at 2500. On increasing the temperature from 30 to 80°C (runs 1-6, Table 3), the optimum activity of 15:15 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 was observed at 40°C. Notably, only modest reductions were evident at either 30 or 50°C, with all values falling at the 10 7 g(PE) mol -1 (Fe) h -1 level. Indeed, only with the temperature above 60°C did the activity start to significantly drop, with a relatively low value of 4:49 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 observable at 80°C (run 6, Table 3). As shown in Figure 3, the molecular weights of the polyethylenes decreased from 59.9 to 1.9 kg mol -1 as the temperature was increased from 30 to 80°C, while the molecular weight distributions ranged from bimodal (≤40°C), with two M pk peaks (peak 1 and peak 2) clearly visible in their GPC traces ( Figure 3), to unimodal (≥50°C) [36]. Moreover, the molecular weight distributions progressively narrowed as the temperature was raised (M w /M n : from 16.0 to 1.5). To account for the modality variations, it would seem likely that two different chain transfer pathways were occurring at temperatures of ≤40°C (e.g., β-H elimination and transfer to aluminum) while at ≥50°C one type of chain transfer was prevalent [11,13,32,36,37].
With the temperature held at 40°C, the polymerization tests were then executed using five different Al : Fe ratios (2000, 2250, 2500, 2750, 3000, and 3250). Inspection of the data indicates that there were only modest effects on the activity across this range in Al : Fe ratios (runs 2 and 7-11, Table 3). Nevertheless, the highest value of 15:15 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 was observed at a ratio of 2500 (run 2, Table 3). On the other hand, the GPC traces indicated that the molecular weight of the polymers gradually decreased from 35.5 to 16.7 kg mol -1 on increasing the ratio from 2000 to 3250 ( Figure 4). As noted at 1 atm C 2 H 4 , this finding can be ascribed to chain transfer from the active species to aluminum on increasing the amount of MMAO resulting in faster chain termination and lower molecular weight polymers [4,26,27,32,36,40,41]. A similar trend has been observed for their symmetrical comparators B, C, and D ( Figure 1) [26,27,32,36]. Notably, bimodal distributions (M w /M n : from 11.0 to 7.7) were again a characteristic of all these runs with two M pk peaks (peaks 1 and 2) viewable in their GPC traces ( Figure 4) with the higher molecular weight fraction progressively becoming the minor component with larger amounts of MMAO [26,27,36,37].

Bond lengths (Å)
Bond angles (°) To investigate the effect of the reaction time on the polymerization, the tests were conducted at run times of between 5 and 60 minutes (runs 2 and 12-15, Table 3) with the Al : Fe ratio fixed at 2500 and the temperature at 40°C. A maximum level of activity of 37:36 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 was attained after 5 minutes (run 12, Table 3), which by the 60-minute mark had noticeably lessened to 8:55 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 (run 15, Table 3). This would imply that the active species was rapidly produced following MMAO addition and then suffered gradual deactivation as the time elapsed [22,24,26,32,36,40,41]. In addition, the molecular weight of the polymers steadily increased over time (M w : from 5.8 to 72.9 kg mol -1 ) with broad bimodal distributions becoming a key feature of the GPC traces over longer test times; notably, the higher molecular weight fraction became the more significant one with more extended run times ( Figure S1). On reduction of the ethylene pressure from 10 to 5 atm, the activity (8:52 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 ) dropped by nearly a half (run 16 vs. 3, Table 3). By comparison, at 1 atm C 2 H 4 ,  the lowest activity (1:82 × 10 6 gðPEÞ mol − ðFeÞ h −1 , Table 2) was observed which can be credited to the lower ethylene concentration at lower pressure [22,23,26,27,32,[36][37][38].
Finally, to allow a comparison of the current catalysts (E/MMAO) with previously reported iron systems (A-D, Figure 1), selected catalytic and polymer parameters for  Table 3).     Figure 6 (see SI, Tables S1-S5). To maintain consistent conditions for the five catalysts, the polymerization tests for A Me2Ph /MMAO and B Me2Ph /MMAO had to be reperformed at 10 atm C 2 H 4 (see SI, Tables S1 and S2) as the original reports used either lower pressure (1.3 bar) or MAO as cocatalyst [4,31]. All the MMAO-activated systems exhibited very high activities when the polymerization runs were performed at either 40°C or 50°C (up to 10 7 g(PE) mol -1 (Fe) h -1 ), with their relative values at 40°C following in decreasing order E Me2Ph > D Me2Ph > C Me2Ph > B Me2Ph~AMe2Ph . They also produced a range of different types of polymers from polyethylene waxes to high molecular weight polyethylene. Indeed, their molecular weights, as a function of the iron precatalyst, were found to decrease in the order Strikingly, E Me2Ph /MMAO was the most active catalyst (15:86 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 at 40°C, Table S5) and formed polymer with a molecular weight in the midrange of the values. This trend in molecular weight indicates that the smaller the ring size of the carbocycle, the lower the molecular weight of the resultant polyethylene. As would be anticipated, the molecular weight of the polymer produced using the 6/7-membered E Me2Ph lies in between that seen for its symmetrical fused-ring comparators B Me2Ph (6/6) and C Me2Ph (7/7). While steric effects imparted by the fused carbocycle are undoubtedly influential on the molecular weight, it would seem likely that other factors such as ring flexibility and chelation properties also play a role on affecting activity and molecular weight [6,36,38].
Pertaining to the dispersity of the polymers, the relatively inflexible ( Table 4 (runs 1-5, Table 5). The best activity (6:53 × 10 5 gðPEÞ mol −1 ðFeÞ h −1 ) was observed at an Al : Fe ratio of 2000. The molecular weights of the resultant polymers were found to decrease gradually from 85.4 to 27.6 kg mol -1 on raising the Al : Fe ratios from 1000 to 3000. As with the lower pressure runs undertaken using Fe4/MMAO, the polymers generated using Fe4/MAO at 1 atm C 2 H 4 also displayed broad molecular weight distributions (M w /M n = 13:7 -20:6) over the range in molar ratios. Meanwhile, an investigation of the reaction temperature was conducted at 1 atm C 2 H 4 with the Al : Fe ratio at 2000. By raising the temperature from 10 to 60°C (runs 3 and 6-10, Table 5), the topmost activity of 12:13 × 10 5 gðPEÞ mol −1 ðFeÞ h −1 was observed at 30°C, while at 60°C only trace amounts of polymer were detected. Subsequently, the performance of Fe4/MAO at higher ethylene pressure was carried out; the results are tabulated in Table 4. The influence of reaction temperature was firstly explored at 10 atm C 2 H 4 with an Al : Fe molar ratio of 2000. On increasing the reaction temperature from 30 to 90°C (runs 1-7, Table 4), the maximum activity of 9:28 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 was observed at 60°C which represents a higher optimum operating temperature to that seen with Fe4/MMAO; this finding highlights the improved thermal stability of the current catalyst. As borne out by their GPC traces (Figure 7), the molecular weights of the polyethylenes decreased from 64.2 to 7.5 kg mol -1 on elevating the temperature from 30 to 90°C, which as mentioned earlier can be ascribed to temperature-induced chain transfer [4,26,27,32,33,36,38,40,41]. As with Fe4/MMAO, the GPC traces obtained using Fe4/MAO over the 30-90°C range indicated the distributions to be bimodal-like at temperatures of ≤40°C, while at ≥50°C they become more unimodal (Figure 7) [36,37].
With the temperature at 60°C, the polymerization runs were carried out using different Al : Fe ratios of 1250, 1500, 1750, 2000, 2250, and 2500 (runs 4 and 8-12, Table 4). The results indicate little effect on the activity across this range in ratios with the highest value of 9:60 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 achievable with an Al : Fe ratio of 1500 (run 9, Table 4). On the other hand, the molecular weight of the polymeric materials was found to decrease gradually from 21.8 to 15.9 kg mol -1 on changing the ratio from 1250 to 2500 (Figure 8), on account of the more rapid chain transfer [4,26,27,36,37,[43][44][45][46]. In comparison with the Fe4/MMAO system, the polymers obtained using Fe4/MAO displayed a narrower molecular weight distribution over the range in molar ratios (M w /M n range: 3.1-5.1). To facilitate an investigation of the catalytic lifetime of Fe4/MAO, the runs were performed over different reaction times from 5 to 60 minutes at 60°C and at an Al : Fe ratio of 1500 (runs 9 and 13-16, Table 4). The highest activity of 21:44 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 was observed after 5 minutes (run 13, Table 4) as was the case with Fe4/MMAO. The activity then gradually decreased to 5:39 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 after 60 minutes (run 16, Table 4) in agreement with gradual deactivation of the active species [22, 26, 27, 32, 33, 36-38, 40, 41]. In addition, the molecular weight of the polymers progressively increased (M w : from 7.5 to 47.8 kg mol -1 ) over time with a modest broadening in the distributions evident (M w /M n : from 2.1 to 6.7) (Figure 9). Interestingly, by representing this GPC data as dN f /(d log M) vs. log M n plots ( Figure S2 in SI), where dN f stands for the number fraction of macromolecules having molecular weight M n , some evidence for two types of active sites was evident. In particular, close inspection of Figure S2 reveals a  low molecular weight polyethylene that disappeared on prolonged reaction time. This observation may suggest a minor contribution of a second type of active site that gradually reduced as the run proceeded. Nevertheless, a gradual increase of M n with polymerization time represents the major trend which is in line with a suppression of the chain transfer rate owing to a depletion of the aluminumalkyl chain transfer agent [13]. On lowering the ethylene pressure to 5 atm (run 17, Table 4), a decline in activity (4:17 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 ) was observed (run 17 vs. run 9, Table 4). Such pressure effects can be attributed to the lower solubility of ethylene in toluene at an ambient ethylene pressure as compared to that at higher pressure [11,32,33,36,43,44].
With reference to the polymers generated using Fe1-Fe5/MAO, the melting temperatures were all around 130°C (Table 4) in accord with a highly linear polymeric backbone. This assertion was supported by the 1 H and 13 C NMR spectra of a sample of polyethylene obtained using Fe4 at 60°C (run 9, Table 4) with a high-intensity single resonance centered around δ 30.0 for the methylene repeat unit in 13 C NMR spectrum along with the corresponding peak at δ 1.37 in the 1 H NMR spectrum (Figures 10 and 11) [22,26,27,32,33,[36][37][38]. Interestingly, the 13 C NMR spectrum also revealed weaker vinylic carbon resonances (-CH=CH 2 ) at   Table 4). around δ 114.36 and δ 139.49 along with more upfield n-propyl peaks (δ 14.22, 22.92, and 32.24). Support for the presence of a vinyl end-group was further provided by the appearance of downfield proton resonances at δ 5.01 and δ 5.90 in the 1 H NMR spectrum ( Figure 11). Based on these NMR observations, it would imply that the main termination pathway in this MAO-promoted polymerization involves β-hydride elimination [4,11,32,36,38,43,44].

Conclusions
In summary, a new family of iron(II) chloride complexes of type E (Fe1-Fe5) bound by an unsymmetrical chelating bis(imino)pyridine ligand fused with both six-and sevenmembered carbocyclic rings has been successfully synthesized and fully characterized. Comparison of the structural properties of Fe3 (6/7) with iron comparators containing the symmetrically fused N,N,N-ligands, B (6/6), C (7/7), and D (8/8), highlights the effects of variation in ring strain/flexibility as well as steric and chelation properties. Upon treatment with either MMAO or MAO, Fe1-Fe5 showed exceptionally high activities (15:86 × 10 6 gðPEÞ mol −1 ðFeÞ h −1 at 40°C) for ethylene polymerization forming strictly linear polyethylenes with a broad range of molecular weights. The steric properties of the precatalyst were shown to be influential with the least sterically hindered 2,6-Me 2 -containing Fe1 and Fe4 displaying higher activity than the more hindered analogues Fe2 (R 1 = Et), Fe3 (R 1 = i-Pr), and Fe5 (R 1 = Et) forming wax-like materials. By contrast, higher molecular weight polymer was obtained with the most sterically encumbered precatalyst Fe3. The polyethylenes were found to display distributions anywhere between narrow unimodal and broad bimodal that could be, to some degree, influenced by the nature/amount of the aluminoxane cocatalyst, temperature, and run time. Moreover, end-group analysis highlighted the role of both β-H elimination (vinyl chain ends) and chain transfer to aluminum (saturated chain ends) as termination pathways. Overall, these hybrid 6-/7-membered ring catalysts exhibit excellent performance characteristics in ethylene polymerization when compared with iron-based B (6/6), C (7/7), and D (8/8), that can, to some level, be explained in terms of the steric properties imparted by the fused carbocycles and the chelation properties of the N,N,N-pincer ligand.   Table 4).    High-purity ethylene was purchased from Beijing Yanshan Petrochemical Co. and used as received. Other reagents were purchased from Acros, Aldrich, or local suppliers. The properties of the resulting polymeric materials such as melting temperatures (T m ) were measured by differential scanning calorimetry (DSC), while the molecular weight (M w ) and dispersity (M w /M n ) were determined by gel permeation chromatography (GPC). In selected cases, high-temperature 1 H and 13 C NMR spectroscopy has also been undertaken to gain further information on the structural properties of the polyolefinic materials; gas chromatography (GC) has been employed in all cases to detect for any short-chain oligomeric fractions (e.g., C 4 -C 32

Polymerization Studies
4.3.1. Ethylene Polymerization at P C2H4 = 1 atm. A 100 mL Schlenk tube, equipped with a stirrer, was employed for the lower pressure polymerization runs. Under an atmosphere of C 2 H 4 , Fe4 (3.0 μmol) was added followed by toluene (30 mL) and then the required amount of cocatalyst (MAO, MMAO) introduced by using a syringe. The solution was then stirred at 1 atm C 2 H 4 with the temperature set at the required value. After 30 min, the pressure was released and the reaction mixture quenched with 30 mL of C 2 H 5 OH (10% HCl). The polymer was washed with C 2 H 5 OH, dried under reduced pressure at 50°C, and weighed.

Ethylene
Polymerization at P C2H4 = 5 or 10 atm. A 250 mL stainless steel autoclave, equipped with a mechanical stirrer, a temperature controller, and an ethylene pressure control system, was employed for the higher pressure polymerization runs (5 or 10 atm C 2 H 4 ). The autoclave was evacuated and refilled with ethylene three times. Firstly, when the required temperature was reached, the selected iron complex (3 μmol), dissolved in toluene (30 mL), was injected into the autoclave under an atmosphere of ethylene (ca. 1 atm), followed by the addition of more toluene (30 mL). Secondly, the required amount of cocatalyst (MAO and MMAO) and additional toluene were added successively by syringe taking the total volume of solvent to 100 mL. The autoclave was immediately pressurized with 5 or 10 atm ethylene, and the stirring commenced. After the required reaction time (5, 10, 30, 45, and 60 min), the reactor was cooled to room temperature with a water bath and the excess ethylene pressure vented. The reaction was quenched with 30 mL of C 2 H 5 OH (10% HCl). The polymer was collected and washed with C 2 H 5 OH and dried under reduced pressure at 50°C and weighed.
4.4. X-Ray Structure Determination. X-ray diffraction (XRD) was employed to determine the molecular structure of Fe3.
The XRD patterns were conducted on a Rigaku Sealed Tube CCD (Saturn 724+) diffractometer with graphitemonochromated Mo-Kα radiation (λ = 0:71073 Å) at 173(2) K, and the cell parameters were obtained by global refinement of the positions of all collected reflections. Intensities were corrected for Lorentz and polarization effects and empirical absorption. The structures were solved by direct methods and refined by full-matrix least-squares on F 2 . All nonhydrogen atoms were refined anisotropically, and all hydrogen atoms were placed in calculated positions. Structure solution and structure refinement were performed using SHELXT-2015 [47,48]. Crystal data and processing parameters for Fe3 are summarized in Table S6.

Data Availability
All data is available in the main text or the Supplementary Materials. runs 2 and 12-15, Table 3). Figure S2: GPC traces (dN f /(d log M n ) vs. log M n ) for the PEs generated using Fe4/MAO at different reaction times (10 atm C 2 H 4 , 60°C, and Al : Fe ratio = 1500; runs 13-16, Table 4). The corresponding GPC traces, represented using dW f /(d log M) vs. log M, are shown in Figure 9. (Supplementary Materials)