Ethylene and 1-butene Oligomerization with Benzimidazole Complexes of Nickel and Iron: A Case of Tandem Reaction

Abstract

The coordination chemistry of benzimidazole ligands combines σ donation and π backbonding. Owing to this electronic flexibility, benzimidazole ligands stabilize both electron deficient and electron-rich transition states in the catalytic cycle of Ziegler-Natta polymerizations. In this study, Fe(III) and Ni(II) complexes of 2-substituted-benzimidazoles were tested as catalysts for ethylene and 1-butene oligomerization. The tests realized in toluene yielded mainly butenes and minor amounts of hexenes. When dichloromethane was used as solvent, a tandem reaction took place and 1-butene produced by ethylene dimerization was further oligomerized, yielding octenes and dodecenes as main products. All tested catalysts exhibited moderate selectivity for 1-octene, indicating 1-ω enchainment in 1-butene dimerization. Beyond catalytic tests, a theoretical study of the ligand 2,2′-(furan-2,5-diyl)bis(1H-benzimidazole) confirmed the planar structure of this compound as evidenced by NMR spectroscopy.

1. Introduction

Ethylene oligomerization is a key process in the petrochemical industry and an intermediate step between oil refineries and diverse final products such as plastics, rubbers, lubricants, and surfactants [1,2,3]. The lighter α-oligomers 1-butene, 1-hexene, and 1-octene are widely used as comonomers for the production of linear low-density polyethylene (LLDPE), a commodity plastic used in packaging. The huge demand for these highly valued compounds has driven research and development of more selective and active catalysts [4,5].

Ethylene oligomers are produced by truncated Ziegler-Natta polymerization. In these catalytic systems, the active species comprise a growing alkyl chain bound to a coordinatively unsaturated metallic center, whose stability is enhanced by agostic interactions [6].

Two reactions may occur at the stabilized metallic center. One is the insertion of a new monomeric unit between the metal and alkyl groups, which leads to chain growth. The other is β-hydride transfer from alkyl to metal-free valence, splitting alkyl anions into hydrides and alkenes. Furthermore, a monomer molecule displaces the bound alkene, initiating a new catalytic cycle with the hydridocomplex [7].

Research on selective and active catalysts for ethylene oligomerization still depends on trial-and-error and metals of groups 4, 6, 8, 9, and 10 are common but not exclusive. Also, a plethora of ligands (Figure 1) have been used [8,9,10,11,12,13]. In a seminal work, Brookhart et al. [14] discovered that ligands with large hindrances near the metal lead to highly active nickel catalysts for polymerization, as bulk substituents hamper β-hydride transfer, favoring chain growth instead of termination [15,16]. For Ni catalysts, the production of heavier oligomers requires high ethylene pressure, as in the SHOP process [17]. Some years after Brookhart’s discovery, tridentate imines showed the same catalytic properties for iron and cobalt compounds [18].

By combining the features of bidentate ligands with the soft phosphorus and hard oxygen donors used in the SHOP process and those of Brookhart’s imines, Grubbs synthesized phenoxy-imine nickel catalysts by the end of the 1990s [19]. Both phenoxy and imine groups are good σ-donors and π-acceptors and are stabilized by electronic delocalization. The overall effect was named “electronic flexibility” by Fujita et al. [20,21], which is defined as the ability of a given ligand to donate more electronic density when required by a metal cation and to withdraw electronic density from the electron-rich intermediates. This effect enables a fine-tuning of Lewis acidity at the metallic center, resulting in smaller activation energies for the coordination, insertion, and olefin release steps, which correlate well with the activity, as hindrance is correlated with the selectivity to polymers.

Several classes of imines, including N-alkyl, N-aryl, and pyridines, have been studied for ethylene oligomerization. Benzimidazole-based catalysts are relatively scarce⁸, and owing to their similarity to purine bases, this heteroaromatic group is ubiquitous in medicinal chemistry [22,23,24,25]. To the best of our knowledge, Ni-containing 1-H-benzimidazoles-2-phenolates have only been studied twice for ethylene oligomerization, exhibiting moderate activity and good selectivity for butenes [12,26]. Also, no Fe(III) catalysts for ethylene and 1-butene oligomerization with this class of ligands have been reported. In this study, we synthesized two Fe(III) benzimidazole-phenolate catalysts and compared them with the well-known Ni(II) compound. In addition, Ni(II) and Fe(III) complexes of a 2,5-bis(benzimidazole)-furan ligand were prepared as in situ catalysts for ethylene oligomerization.

2. Materials and Methods

Analytical grade compounds, such as sodium metabisulfite (Merck, Rio de Janeiro, Brazil), salicylaldehyde (Aldrich, Cajamar, Brazil), o-phenylenediamine (Merck), hydroxymethylfurfural (Aldrich), acetylacetone (Merck), triphenylphosphine (Acros, Sao Paulo, Brazil), (n-C₄H₉)₃N (Tedia, Rio de Janeiro, Brazil), FeCl₃ (Aldrich), Ni(NO₃)₂.6H₂O (Aldrich), and KMnO₄ (Aldrich), were used as received. Authentic samples of cyclohexane (Merck), 1-octene (Aldrich), 1-hexene (Acros), 1-decene (Acros) and 1-dodecene (Aldrich) were used as standards for the chromatographic analysis. Ethylene (Air Products, Duque de Caxias, Brazil) and 1-butene (Messer, Rio de Janeiro, Brazil) were percolated in a silica column with Cu₂O prior to use. Toluene (Vetec) was dried over sodium/benzophenone and distilled under a N₂ atmosphere. Methylene chloride (Vetec) was dried over P₂O₅ (Merck) and distilled under N₂. Both solvents were stored in Schlenk flasks with molecular sieves. The precursor complexes [Fe(acac)₃] (Aldrich) and [Ni(acac)₂(H₂O)₂] (Aldrich) were used as received.

2.1. Synthesis of Benzimidazole-Phenol Ligands

The ligands 2-(1H-1,3-benzimidazol-2-yl)phenol (HL1) and 2-{[2-(2-hydroxyphenyl)-1H-1,3-benzimidazol-1-yl]methyl}phenol (HL2) were synthesized (Figure 2) according to the procedures described by Betti et al. [27].

In a round-bottom flask, 15 mmol of Na₂S₂O₅ was dissolved in 20 mL of water under N₂ atmosphere. Neat salicylaldehyde (30 mmol) was added dropwise and the mixture was stirred for 5 min. A fine white precipitate was observed. A solution of phenylenediamine (30 mmol in 20 mL of ethanol) was added and the mixture was warmed to 70 °C with vigorous stirring in air for 150 min. A condenser was used to reduce solvent loss. After cooling the mixture to room temperature, it was refrigerated overnight. The crude precipitate containing a mixture of HL1 and HL2 was filtered, sequentially washed with ethanol/water (1:1) and pure water, and then dried over silica.

2.1.1. Isolation of HL2

The dried mixture was dissolved in hot ethanol, and colorless crystals of pure HL2 (m.p. 219 °C) [28] were obtained after two days. The crystals were filtered and washed with cold ethanol (yield: 670 mg, 14.1%). Infrared data (ATR, cm⁻¹): 3234, 3043, 1591, 1452, 1256, 1238, 1100, 1136, 839, 737. ¹H NMR (dmso-D6, 300 MHz) 5.32 (s, 2H), 6.44 (dd, 1H), 6.61 (td, 1H), 6.84 (dd, 1H), 6.89 (td, 1H), 7.04 (dd, 1H) 7.04 (td, 1H) 7.24 (m, 2H) 7.37 (m, 2H), 7.45 (dd, 1H) 7.73 (dd, 1H). ¹³C NMR (dmso-D6, 75 MHz) 43.3, 110.9, 115.1, 116.2, 116.4, 118.7, 119.1 (2 coalescent signals) 122.1, 122.6, 122.7, 126.7, 128.4, 130.3, 131.4, 135.2, 141.9, 152.1, 154.4, 156.4.

2.1.2. Isolation of HL1

The filtrate obtained after the HL2 removal was poured into 100 mL of water, stirred, and filtered again. The solid fraction was discarded, and the remaining solution was stored in a freezer. Pure L1 (m.p. 242 °C) [29] was isolated as an off-white powder after a week (yield: 870 mg, 25.6%). IR (ATR, cm⁻¹) 3464, 3326, 3057, 1262, 725. ¹H NMR (dmso-D6, 300 MHz) 7.04 (m, 2H), 7.27 (d, broad, 2H), 7.37 (td, 1H), 7.66 (d, broad, 2H), 8.06 (dd. 1 H) 13.18 (s, 2H). ¹³C NMR (dmso-D6, 75 MHz) 112.0, 113.1, 117.7, 118.4, 119.6, 122.9 123.7, 126.7, 133.2, 133.7, 141.3, 152.2, 158.5.

2.2. Synthesis of L3

This compound L3 was synthesized by a two-step procedure (Figure 3). In a first step, the precursor 5-(hydroxymethyl)furfural (2.843 g, 20 mmol) was dissolved in 200 mL of 2.5 M NaOH. The solution was kept in an ice bath, and 7.370 g of KMnO₄ was added in small portions under vigorous stirring, and agitation continued for 10 min after addition. The manganese dioxide precipitated and was filtered. Then, concentrated HCl was added to the filtrate until the pH reached zero. The solution was kept in the refrigerator overnight. Pure furan-2,5-dicarboxylic acid (FDCA) was isolated as pale-yellow crystals (yield 2.528 g, 81%) with a high melting point (>300 °C) [29].

In the second step, FDCA is condensed with o-phenylenediamine to yield the desired benzimidazole [30]. In a round-bottom flask, a mixture containing 780 mg (5 mmol) of FDCA and 1.080 g (10 mmol) of o-phenylenediamine was dissolved in 10 mL of 85% phosphoric acid. The system was stirred at 150 °C, under a nitrogen atmosphere, for 20 h. During this period, a Dean-Stark apparatus was used for the continuous removal of water. After cooling the mixture to room temperature, 10 mL of distilled water was added dropwise to avoid overheating. Then, the diluted mixture was transferred to a beaker containing 100 mL of cold water. NaOH 2.5 M was slowly added under vigorous stirring until a neutral pH. The crude precipitate was filtered, washed with cold water, and dried over silica for 5 days. Pure L3 was isolated after washing the solid with 40 mL of NaOH 0.2 M. The purified light brown compound was vacuum filtered, washed with cold water, and dried over silica (737 mg, yield = 49%). IR data (KBr plates, cm⁻¹): 3435, 3063, 1632, 1418, 1280, 1236, 740. 1H NMR (dmso-D6, 300 MHz, 50 °C.) 7.22–7.28 (dd, 4H), 7.40 (s, 2H), 7.59–7.66 (m, broad, 4H), 13.03 (s, very broad, 2H). 13C NMR (dmso-D6, 75 MHz) 113.0, 119.0, 123.1, 135.2, 143.4, 144.3, 146.8.

2.3. Synthesis of Complexes C1–C3

2.3.1. [Ni(L1)₂] (C1)

In a round-bottom flask, 1 mmol of HL1 and 1.5 mmol of N(n-C₄H₉)₃ were dissolved in 10 mL of hot ethanol. Then, 5.0 mL of an ethanolic nickel nitrate solution (1 mol/L) was added. The mixture was warmed up to 60 °C and stirred for 30 min. A light-brown precipitate was observed. The mixture was then cooled and stored in a refrigerator for 20 h. The solid was vacuum-filtered, washed with cold ethanol, and dried over silica (yield 78 mg, 33%). IR (ATR, cm⁻¹) 3052, 2968, 2917, 1604, 1564, 1539, 1479, 1263, 1140, 724.

Elemental analysis. Calculated C = 65.45; H = 3.80; N = 11.74. Found: C = 65.19; H = 3.97; N = 11.35.

Mass spectroscopy (see Figure A1).

[M + H⁺] spectrum: [Ni(L1)₂]⁺, 477 Da, 4%; [H₂L1]⁺, 211 Da, 100%. [M − H⁺] spectrum: [H(L1)₂]⁻, 419 Da, 2%; [L1]⁻, 209 Da, 100%.

2.3.2. [Fe(L1)₂(acac)] (C2)

In a round-bottom flask, 0.33 mmol (2 eq.) of HL1 was dissolved in 10 mL of ethanol. Then, this solution was added to an ethanolic solution of [Fe(acac)₃] (1 eq. in 5 mL) and stirred at room temperature for 20 h. The reddish-brown precipitate was filtered, washed with ethanol, and dried over silica for three days (yield: 57 mg, 60%). IR (ATR, cm-1) 3446, 3052, 2964, 1599, 1581, 1557, 1519, 1474, 1441, 1361, Elemental analysis. Calculated: C = 64.82; H = 4.56; N = 9.75. Found: C = 63.44; H = 4.70; N = 9.91.

Mass spectroscopy (see Figure A2).

[M + H⁺] spectrum: [Fe(L1)₂]⁺, 474 Da, 10%; [H₂L1]⁺, 211 Da, 100%. [M − H⁺] spectrum: [Fe(L1 − H)₂]⁻, 472 Da, 4%; [L1]⁻, 209 Da, 100%.

2.3.3. [Fe(L2)₂(acac)] (C3)

A total of 100 micromols of L2 was solubilized in 45 mL of hot CH₂Cl₂. After the addition of a solution containing 50 µmol of [Fe(acac)₃] in 5 mL of CH₂Cl₂, a brownish-orange precipitate was observed. The mixture was then maintained at 30 °C and stirred for 90 min. After resting overnight at room temperature, the fine powder was decanted and filtered. The solid was washed with CH₂Cl₂ and dried under vacuum until constant weight (yield: 31 mg, 79%). IR (ATR, cm⁻¹), 3053, 1583, 1558, 1523, 1475, 1248, 739. Elemental analysis. Calculated: C = 68.80; H = 4.75; N = 7.13. Found: C = 67,02; H = 4.87; N = 7.27. The compound was poorly soluble in methanol and MeCN and was not analyzed by mass spectroscopy.

2.4. Preparation In Situ of the Pre-Catalysts C1’–C5’

The catalysts C1’–C4’ were prepared by mixing of solutions containing 5 μmols of ligand dissolved in 8 mL of dry CH₂CL₂ with 5 μmols of the respective acetylacetonate dissolved in 2 mL of the same solvent under N₂ atmosphere (Figure 4). Each mixture was stirred for 10 min and immediately used as catalyst.

The catalyst C5’ was prepared by dissolving 8 µmol of FeCl₃ in 3 mL of methanol. A solution of L3 (8 µmol) in methanol (5 mL) was added to this mixture. The system was stirred at 40 °C for 1 h, and one aliquot of 2 mL was transferred to a Schlenk flask. The solution was concentrated in a rotary evaporator and the residue was dried at 50 °C under vacuum for 2 h. The remaining solid was suspended in 10 mL of dichloromethane and stirred under N₂ for 20 h. This suspension was used as a heterogeneous catalyst following the general procedure for ethylene oligomerization.

2.5. General Procedure for Ethylene and 1-Butene Oligomerization

All manipulations involving compounds sensitive to air and moisture were performed under an inert N₂ atmosphere using glove bags or Schlenk techniques.

In a typical experiment, 5 μmol of the catalyst and 10 mL of solvent were added to a Schlenk flask. Air was then removed through repeated vacuum and dry N₂ cycles, and ethylene was bubbled into the solution for about 30 s. The reaction commenced with the addition of 200 equivalents of cocatalyst MAO under an ethylene flow. After 30 min of oligomerization at 25 °C, the monomer feed was stopped, and the alkene mixture was further oligomerized for an additional 5 min. To quench the reaction, the Schlenk flask was placed in an ice/salt bath, and 2 mL of water was added. After 5 min of vigorous stirring, an aliquot of the organic layer was dried over sodium sulfate and analyzed by gas chromatography (GC). Each sample was prepared by mixing 540 microliters of the dried reaction mixture with 40 microliters of 1-decene or cyclohexane as an internal standard. For some catalytic runs, a second sample without an internal standard was analyzed.

The catalytic tests with 1-butene followed this protocol, except that the solvent volume was reduced to 5 mL. The evaluation of catalyst activities was based on 1-butene conversion.

2.6. Methodology and Instrumentation

NMR spectra were acquired on a VARIAN Mercury 300 MHz unit using deuterated dmso-D6 as an internal reference. Infrared (IR) spectra were obtained using a Nicolet iS5 Thermo Scientific spectrophotometer. Quadrupole-Orbitrap Thermo QExactive (TOF-ESI) was used for mass spectroscopy analysis of the complexes. Both the positive [M + H+] and negative [M − H+] methods were applied. Elemental analyses were performed using a Unicube (Elementar Analysensysteme, Langenselbold, Germany) with sulfanilamide as standard. The UV-visible spectra were acquired using a Shimadzu UV-2600i spectrophotometer (Shimadzu, Kyoto, Japan). Theoretical calculations were performed using the GAMESS US package available online at www.chemcompute.com [31,32]. Geometries were optimized with the DFT B3LYP [33,34,35] functional in conjunction with the 6–31G* basis set. Further single-point energy calculations were performed by substituting 6–31G* with a more complete SPK-DZP basis set [36,37].

Qualitative analysis of the reaction products was performed using gas chromatography–mass spectrometry (GC-MS) on a Shimadzu GC-2030 system (Shimadzu, Kyoto, Japan) equipped with a DB-1MS column (30 m × 0.32 mm). Quantitative analysis was conducted with 1-decene as an internal standard using a Shimadzu GC-2030 gas chromatograph with a DB-1 column (100 m × 0.32 mm) and a flame ionization detector (FID), following a previously reported method [38]. The oven temperature was initially set at 32 °C, then increased at a rate of 2 °C/min to 48 °C, followed by a ramp of 7.5 °C/min to 100 °C, where it was held for 8 min. Subsequently, the temperature was increased at 15 °C/min to 300 °C and maintained for 13.7 min. The injector and FID temperatures were set at 50 °C and 310 °C, respectively.

3. Results and Discussion

3.1. Infrared of Ligands and Complexes

The IR data relative to the ligand HL1 and complex C1 were similar to those reported in the literature [12]. In both spectra a broad band centered near 3400 cm⁻¹, relative to N-H stretching, appears. However, in the ligand HL1 spectrum a sharp band at 3325 cm⁻¹ due to O-H stretching partially covers N-H band (Figure 5a). The N-H stretching band also appears as a broad band in the ligand L3 spectrum (Figure 5b).

The spectra of precursor [Fe(acac)₃], ligand HL2 and pre-catalysts C2 and C3 are presented in the Supplementary Information. All benzimidazole compounds exhibited a sharp peak near 840 cm⁻¹ relative to ortho substituted benzene ring. A duplet at 1515 cm⁻¹ and 1563 cm⁻¹ relative to conjugated enolato/ketone appears in [Fe(acac)₃] spectrum and these peaks are shifted to about 1520 cm⁻¹ and 1580 cm⁻¹ in complexes C2 and C3 [39].

3.2. NMR Study of Ligands

The NMR spectra of HL1 and HL2 (see Supplementary Information) were similar to those reported in the literature [12,28]. The tautomeric equilibrium [40,41], characteristic of 1-H-benzimidazole compounds, simplified the spectra of HL1 and L3 since positions 7 and 8 became equivalents as positions 6 and 9 of benzimidazole rings.

Three ¹H NMR spectra of L3 were recorded in dmso D-6 at various temperatures (Figure 6). The spectrum recorded at 25 °C exhibits a poorly resolved double duplet centered at 7.26 ppm (peak a), relative to hydrogen atoms at positions 7 and 8 of benzimidazole rings (Figure 6, at the bottom, in red). A sharp peak relative to CH groups of furan moiety (peak b) appears at 7.39 ppm) In addition, a broad signal at 7.63 ppm, assigned to CH at positions 6 and 9 (peak c) was observed. These peaks became more resolved at higher temperatures owing to the faster rotation of the single bonds. At 25 °C, the strong interactions with the viscous solvent dimethyl sulfoxide were sufficient to limit rotation, and the overall spectrum was a summation of different conformations, leading to unresolved peaks. The peaks relative to positions 6 and 9 are considerably more affected by temperature, which suggests that the distance to the NH groups is a key factor.

When the rotation became relatively fast compared to the acquisition time in the NMR experiment, a well-resolved spectrum of the “average conformation” was obtained [42]. Notably, when the temperature was raised from 25 °C to 50 °C and finally to 80 °C, the chemical shifts in furan CH groups were moved downward at a ratio of approximately 0.005 ppm per Celsius degree.

The ¹³C spectrum of L3 (Figure 7a) exhibits four well-defined peaks and three poorly resolved peaks. Two peaks at 113.0 (peak b) and 143.4 ppm (peak f) refer to CH and quaternary carbon of furan moiety, respectively. The CH signals of benzimidazole fragment appear at 123.1 ppm (peak a), referring to CH groups at positions 7 and 8 and at 119.0 ppm, which was assigned to CH groups at positions 6 and 9 (peak c). The signal at 146.8 ppm (peak g) is attributed to quaternary carbon at position 2 of benzimidazole ring, while the signals at 135.17 ppm (peak e) and 144.33 ppm (peak d) were assigned to quaternary carbons at positions 4 and 5 of benzimidazole ring, respectively [43]. Bidimensional HSQC spectrum of L3 (Figure 7b) shows one peak relative to equivalent positions 7 and 8 of benzimidazole fragment at (7.22 ppm; 122.3 ppm) and another at (7.39 ppm; 112.5 ppm) relative to positions 3 and 4 of furan ring. A broad proton signal near 7.6 ppm interacts with a carbon signal near 118 ppm, corresponding to equivalent CH groups at positions 6 and 9.

3.3. UV-Visible Study of C5’

Since Fe(III) complex with L3 ligand had not been isolated before catalytic tests and only a subtle color change was observed, we sought definitive evidence of metal coordination. We recorded UV-visible spectra of the ligand, precursor salt FeCl₃, and the mixture of precursor with ligand for 10⁻⁴ M solutions in methanol, as shown in Figure 8. The solutions were maintained at room temperature for 24 h before analysis. Moreover, we calculated the “Delta” spectrum, defined as the difference between the spectrum of a solution containing precursor salt and ligand mixture and the sum of precursor salt and free ligand spectra.

Intense absorptivity was observed near 210 nm for the FeCl₃ solution. This band was attributed to solvato-complexes, such as [Fe(methanol)_nCl_x]^(3−x)+ [44]. The free ligand L3 also absorbs in this region; however, fortunately, the “Delta” spectrum exhibits a negative peak, roughly symmetrical to the peak around 210 nm in FeCl₃ spectrum. Therefore, we assumed that the concentration of solvato-complexes vanished in the presence of L3.

In addition, the “Delta” spectrum shows the appearance of two new peaks centered at 370 nm and 390 nm. Noteworthily, the peak at 390 nm is localized in a region where the free ligand and solvato-complexes have near-zero absorptivity, confirming that L3 coordinates to Fe(III).

3.4. Molecular Modelling of Compound L3

After confirming that L3 complexes were formed in solution, we performed theoretical studies to better understand the nature of this ligand. DFT calculations were performed using the B3LYP functional [33,34,35] and low-cost 6–31G* basis set for structure optimization. Further single-point energy calculations were performed using the same functional and SPK-DZP basis sets [37].

The optimized ligand adopted a planar structure with a small energy gap between the HOMO (−0.203 a.u.) and LUMO (−0.053 a.u.). Both frontier orbitals (Figure 9) are delocalized over the entire molecule. Fujita [45] pointed out that the electronic flexibility owing to these electronic features favors high catalytic activity.

In the planar conformation, this ligand can coordinate using the ether moiety, as suggested by Vyas and Kaur [25], who synthesized nickel and silver adducts of L3. The planar geometry also enhances π-electron delocalization and consequent resonance stabilization.

Because the rotation of the connecting bonds between the rings eliminates this resonance effect, L3 molecules should be kept preferentially in the planar conformation at low temperatures. The ¹H NMR spectra at different temperatures (Figure 6) are in excellent agreement with the calculated electronic properties of L3.

3.5. Oligomerization of Ethylene and 1-Butene

In this study, five catalysts prepared in situ and two isolated compounds were tested for ethylene oligomerization (Figure 4). Once literature reports the usage of Ni(II) and Fe(II) complexes with L1 and similar ligands for ethylene oligomerization [11,12,26] in toluene, we initially performed our catalytic tests in this solvent. A comparison of our results with those reported by Shi and Tadjarodi groups is shown in

Table 1. Ethylene oligomerization in toluene.

Precatalyst (Activator) a	Activity b	P (atm)	Oligomer Distribution c
			C4 (α)	C6 (α)	C8 (α)	Ref.
C1 (MAO) d	73	1	69.4 (0)	30.6 (0)	-	This work
C2 (MAO) d	inactive	1	-	-	-	This work
C1 + PPh3 (MAO) d,e	109	1	84.7 (0)	15.3 (15)	-	This work
C1 (Et2AlCl) f	368	20	86.6 (89)	12.6 (57)	0.8 (100)	[12]
C1 + PPh3 (Et2AlCl) g	1100	1	93.3	6.7	-	[26]
C1 (Et2AlCl) h	301	1	92.4 (>15)	7.6 (>15)	-	[26]
Fe(L1)2 (Et2AlCl) i	344	20	93.1 (7)	6.9 (100)	-	[11]

^a Al/Ni or Al/Fe = 200. ^b In kg_ethylene/mol_catalyst∙hour. 1-decene was used as internal standard. ^c Based on GC analysis; (α) denotes the percentage of linear α-olefins. ^d Conditions: 5 μmol of catalyst, 10 mL of solvent, T = 25 °C, t = 30 min. ^e Two equivalents of PPh₃ were added to the system. ^f Conditions: 10 μmol of catalyst, 100 mL of solvent, T = 30 °C, t = 30 min. ^g Conditions: 5 μmol of catalyst, 30 mL of solvent, 10 eq. of PPh₃; T = 20 °C, t = 30 min. ^h Conditions: 5 μmol of catalyst, 30 mL of solvent, T = 20 °C, t = 30 min. ⁱ Conditions: 10 μmol of catalyst, 100 mL of solvent, T = 30 °C, t = 30 min. The pre-catalyst was isolated as K₂[Fe(L1)₂Cl₂]∙4 H₂O.

. These authors obtained better activities and selectivity for 1-butene using the same complex C1. Also, the protective effect of triphenylphosphine (PPh₃) was confirmed by our results.

The Fe(III) catalyst C2 was inactive in toluene. Once Fe(III) is very sensitive to water, we postulated that an oxo-bridged dinuclear complex was obtained, probably the same compound μ-O-[Fe(L1)₂]₂ reported by Wahlgren [46].

The absence of 1-butene and a higher percentage of hexenes in our results indicated that 1-butene was isomerized and co-dimerized by C1 as ethylene concentration vanished after closing the monomer feed.

Table 2. Ethylene oligomerization in CH₂Cl₂ ^a.

Entry	Catalyst b	Activity c	Oligomer Distribution d
			C4 (α)	C6 (α)	C8 (α)	C10	C12
1	C1	17	37.4 (0)	16.9 (11)	45.7 (16)	-	-
2	C2	4.4	10.0 (0)	0	90.0 (25)	-	-
3	C2’	4.8	10.7 (0)	0	89.3 (25)	-	-
4	C3’	6.9	12.8 (0)	0	87.2 (29)	-	-
5 e	C5’	9.0	20.1 (0)	0	79.9 (9)	-	-
6 f	C1’		55.0 (53)	14.5 (15)	20.8 (11)	0	9.7
7 f	C2		27.5 (0)	0	29.3 (21)	0	43.2
8 f	C2’		12.1 (0)	0	56.9 (21)	0	31.0
9 f	C3’		11.3 (21)	2.8 (0)	41.6 (26)	0	44.3
10 e,f	C5’		6.9 (23)	0	21.8 (8)	0	71.3
11 g	C1’	218	45.5 (45)	4.6 (66)	29.0 (12)	0	21.2
12 g	C2’	42	9.5 (0)	0	55.70(18)	0	34.8
13 g	C3’	46	6.4 (44)	0	54.2 (18)	0	39.3
14 g	C4’	151	35.4 (49)	16.2 (13)	29.3 (11)	0	19.0

^a Conditions: 5 μmol of catalyst, 200 eq of MAO, 10 mL CH₂Cl₂, T = 25 °C, t = 30 min, P = 1 atm. ^b The symbol ’ denotes catalysts prepared in situ. ^c In kg_ethylene/mol_catalyst∙h. ^d Based on GC analysis; C10 and C12 compounds were not analyzed in GC runs relatives to entries 1–5, (α) denotes the percentage of linear α-olefins. ^e 2 μmol of catalyst. ^f These CG runs were performed without internal standard. ^g Cyclohexane was used as internal standard.

contains the results of ethylene oligomerization in dichloromethane. Since the literature reports only butenes and hexenes as main products with octenes present, if ever, in trace quantities, we chose 1-decene as internal standard and excluded the peaks that appeared after it. This method worked quite well for the catalytic tests performed in toluene, as no tetramer or heavier oligomer of ethylene was observed. However, when dichloromethane was used as solvent, the GC runs showed the presence of octenes and dodecenes. Thus, after the samples containing 1-decene were analyzed (reported in entries 1–5), a second sample of each test relative to entries 2–5 was prepared without 1-decene, and their GC data are shown in entries 7–10. Thus, entries 2 and 7 refer to the same catalytic test as entries 3 and 8, and so on. An additional test with C1’ (entry 6) was performed. The in situ prepared catalyst C1’ was more selective to 1-butene than the pre-isolated catalyst C1 (entry 1).

A second set of catalytic tests, using catalysts C1’–C4’ prepared in situ, was performed, and their samples were analyzed with cyclohexane as internal standard. Because small amounts of 1-decene isomers were present (see Supplementary Information) in the internal standard, we decided to use cyclohexane to avoid misinterpretation of chromatographic data (entries 11–14).

In the solvent CH₂Cl_2, ethylene oligomerizations yielded octenes and dodecenes as main products, with hexenes being scarcely produced by iron catalysts, as shown in Table 2. This solvent effect leads to results quite different from the expected Schulz-Flory distribution and is correlated to a tandem reaction in which 1-butene is produced by ethylene dimerization and further oligomerized, yielding octenes and dodecenes (Figure 10). In our analysis, we considered that oligomerization of hindered 2-butenes cannot be competitive with ethylene oligomerization.

Noteworthily, the apparent low activities and high selectivities for octenes (as shown in entries 1–5) are probably due to 1-butene loss, because the results in entries 6–10 exhibit considerable amounts of 1-butene and to the premature end of GC analyses just after 1-decene elution, which excluded the dodecenes. Despite these observations, we decided to retain these data because they record an improvement in the methodology.

The more realistic oligomer distribution in entries 6–10 shows a moderate selectivity to α-olefins, with iron complexes C2, C2’, C3’ and C5’ exhibiting better yields for 1-butene oligomerization. Also, C₁₀H₂₀ compounds were not observed in entries 6–14, as shown in Figure A3, confirming that dodecenes were not obtained by successive ethylene insertions.

Nickel (II) catalysts C1, C1’ and C4’ exhibited higher activity than Fe(III) ones. Since the Fe(II) catalysts prepared by Tadjarodi with HL1 and similar ligands exhibited activities as high as Ni compounds [11,12] the oxidation state of pre-catalyst may limit activity.

The catalyst C5’ with a (bis-benzimidazole)-furan tridentate ligand exhibited higher selectivity to dodecenes (entry 10) and converted almost all butene produced into higher alkenes while C4’ produced the broader mixture of oligomers, indicating the presence of several active species. For all tested catalysts, the production of 1-octene indicates 1 − ω enchainment when 1-butene is dimerized, instead of ethylene tetramerization since hexenes were scarce in reactional media.

Olivier-Bourbigou and Casagrande Jr. tested several nickel catalysts that contained ether moieties [47,48]. In general, these catalysts produced mainly butene isomers and smaller amounts of hexenes with higher selectivity for alpha olefins when the MAO cocatalyst was used.

Differing from ethylene, the coordination and insertion of 1-butene require unhindered free valences at the catalyst. The coordinative properties of arenes are well-known [49,50,51,52] and toluene competes with ethylene and 1-butene for free sites at catalysts. Therefore, in toluene only the smaller and more reactive ethylene is oligomerized at a reasonable rate, and 1-butene accumulated in the reactor. On the other hand, in non-coordinating dichloromethane, 1-butene oligomerization becomes increasingly competitive as this compound is accumulated by ethylene oligomerization. Because our procedure included 5 min of ethylene depletion, 1-butene production ceased and the accumulated product was consumed, which explains the observed low selectivity for 1-butene.

Two tests were carried out with 1-butene as the monomer, in which the amount of remaining 1-butene was used to calculate the conversion to oligomers. The general procedure involved ethylene oligomerization, but the solvent volume was reduced to 5 mL. Under these conditions, catalyst C1 was much more active for the isomerization than for the oligomerization of 1-butene, as shown in

Table 3. Results of catalytic tests with 1-butene in dichloromethane ^a.

Catalyst	Conversion (%)	Selectivity (%)
		C4	C8 (α) b	C12
C1	98	96.8	1.6 (4)	1.6
C2	4.1	6.0	47.5 (17)	46.5

^a Conditions: 5 μmol of catalyst, 200 eq of MAO, 5 mL of solvent, T = 25 °C, t = 30 min, and P = 1 atm. ^b (α) denotes the percent of linear α olefin.

. In contrast, the less active C2 oligomerized 1-butene with only 6% isomerization. This is in good agreement with the results in Table 2, which show lower activities and higher selectivities for octenes and dodecenes observed for iron catalysts.

For both catalysts, the amounts of dimers and trimers produced were similar, but C2 was more selective to the high-value 1-octene, confirming its ability to perform 1 − ω enchainment of 1-butene units, as observed when ethylene was used as feedstock. In addition, C1 produced only one trimer, whereas C2 produced different C₁₂H₂₄ isomers, as shown in Figure S9 (see Supplementary Information).

4. Conclusions

All catalysts oligomerized ethylene, producing butenes in the first step. In toluene, ethylene oligomerization produced mainly butenes and minor amounts of hexenes. However, in dichloromethane, a tandem reaction produced octenes and dodecenes via the oligomerization of 1-butene. Owing to the scarcity of hexenes (C₂H₄ trimers), the presence of dodecenes suggests co-dimerization of butene and one terminal C₈H₁₆ isomer rather than hexene dimerization. Moreover, catalysts prepared in situ by the simple mixing of benzimidazoles and metal salts exhibited good activities and laborious isolation of the complexes is not required.

In addition, butene dimerization yielding 1-octene is related to chain walking with 1-ω enchainment, a promising synthetic route for this high-value compound. The catalysts studied in this work enable this reaction, although the optimization of 1-octene production is a challenging task.

The coordination chemistry of L3, a furan-benzimidazole compound known since the 1960s, is a promising field, and catalytic systems containing this ligand are active for ethylene oligomerization. Notably, the rotation of the single bonds of this polyaromatic compound has high energy barriers, as indicated by NMR spectra and molecular modelling.

The hypothesis of a tandem reaction to explain the production of octenes and dodecenes was confirmed by the oligomerization of 1-butene in CH₂Cl₂. As observed in the experiments, nickel catalyst C1 and iron catalyst C2 produced dimers and trimers of 1-butene. Catalyst C1 isomerized a major part of 1-butene, yielding inert internal alkenes. In contrast, the less active C2 produced a distribution of octenes and dodecenes, with moderate selectivity for 1-octene.