Preparation and Study of Reusable Polymerized Catalysts for Ester Hydrogenation

Abstract

A cross-linked catalyst organic framework was prepared by an alternating ring-opening olefin metathesis polymerization between dichloro{N,N′-bis({(2-diphenylphosphino)phenyl}methylidene)bicyclo[2.2.1]-hept-5-ene-2,3-diamine}ruthenium, 1,2-N-di(cis-5-norbornene-2,3-endo-dicarboximido)-ethane, and cis-cyclooctene catalyzed by RuCl₂(=CHPh)(PCy₃)₂ in the presence of a BaSO₄ support. The heterogenized catalyst hydrogenated methyl benzoate at a similar rate to the homogeneous catalyst (0.0025 mol % catalyst, 10 mol % KO^tBu, 80 °C, 50 atm, tetrahydrofuran, 21 h, ∼15 000 turnovers during the first 1 h). The catalyst was used five times for a total of 121 680 turnovers. A study on the reusability of this catalyst showed that ester hydrogenations with bifunctional catalysts slow as the reaction proceeds. This inhibition is removed by isolating and reusing the catalyst, suggesting that future catalyst design should emphasize avoiding product inhibition.

Introduction

We report a cross-linked, reusable polymeric catalyst for ester hydrogenations under mild conditions. Catalytic hydrogenation is an atom-economic alternative to the stoichiometric reduction of esters using main-group hydride reagents.^1,2 The first ester hydrogenation catalysts were heterogeneous copper chromite species³⁻⁵ that required harsh conditions (250–300 °C, 130–200 atm H₂) and high catalyst loadings (∼10 wt %).⁶ Current heterogeneous⁷ ester hydrogenation catalyst systems are more active and include Ru–ZrO₂,⁸ Ru–Sn,^9,10 Ru–Ge,¹¹ and Ag–Au¹² bimetallic systems (typical conditions: 30–50 atm H₂, 140–300 °C).¹² The first homogeneous catalysts were ruthenate complexes that hydrogenated activated esters under moderately forcing conditions (10–100% conv., 0.6 mol % Ru, 90 °C, 6.1 atm, toluene, 4–20 h).¹³ The Milstein group reported a breakthrough in 2006 with the ruthenium PNN pincer compound [(PNN)RuH(CO)] [PNN = 6-(di-t-butylphosphinomethylene)-2-(N,N-diethylaminomethyl)-1,6-dihydropyridine] that hydrogenated a number of esters with turnover numbers (TONs) ranging from 82 to 100 (5.3 atm H₂, 1 mol % catalyst, 4–12 h, 115 °C).¹⁴ Saudan et al. reported shortly thereafter that trans-[RuCl₂(Ph₂PCH₂CH₂NH₂)₂] and related ruthenium-based bifunctional catalysts hydrogenate a variety of esters with TONs as high as 9900 under practical conditions in the presence of base [0.01–0.05 mol % catalyst, 5 mol % NaOMe, 100 °C, 49 atm H₂, tetrahydrofuran (THF), 1–4 h].¹⁵ Many ester hydrogenation catalysts have been developed since these discovery reports. The more active catalysts in the literature contain Ru^6,16−23 or Os²⁴ metal centers, the most active hydrogenate esters with TONs ranging between 49 000 and 91 000 (typical conditions: 0.001–0.002 mol % catalyst, 2–10 mol % alkoxide base, 25–80 °C, 49–100 atm).²⁵⁻²⁸

Milstein and co-workers reported the first iron-based catalyst mer-[(2,6-(tBu₂PCH₂)₂C₅H₃N)Fe(H)₂(CO)] for ester hydrogenations in 2014.²⁹ This catalyst hydrogenated 2,2,2-trifluoroethyl trifluoroacetate with TONs up to 1280 (0.05 mol % catalyst, 1 mol % KO^tBu, 40 °C, 10 atm H₂, 1,4-dioxane, 90 h). Several groups have since then developed with earth-abundant metal centers, including Co,³⁰⁻³² Fe,³³⁻³⁶ and Mn.³⁷⁻⁴⁰ These systems are highly promising, with TONs ranging from tens to a few hundreds (30–50 atm H₂, 70–120 °C). Jones et al. have recently reported 3890 turnovers for the hydrogenation of the lactone γ-valerolactone with a cobalt catalyst [(HN(CH₂CH₂PCy₂)₂)Co(CH₂SiMe₃)]BAr₄^F (0.01 mol % Co, 54 atm, 120 °C, 72 h, neat).⁴¹

Another strategy to lower the cost and toxicity of homogeneous catalysts is to immobilize them by covalent bonding to polymers and dedrimers.⁴²⁻⁶⁶ There are a number of reports describing polymerized Ru catalysts for the hydrogenations and transfer hydrogenations of ketones.⁶⁷⁻⁶⁹ For example,⁶⁹ a mesoporous, polystyrene-based, self-supported RuCl(p-cymene) (TsDPEN) [TsDPEN = (1R,2R)-N-p-tosyl-1,2-diphenylethylenediamine] complex catalyzed the transfer hydrogenation of acetophenone over 14 runs (total TON ∼ 1300) without significant loss in activity or enantioselectivity [enantiomeric excess (ee) = 93–94%, 1 mol % Ru per run, 500 mol% HCOONa, 40 °C, H₂O, 1.5 h per run]. Our group previously reported a highly reusable Ru–BINAP [BINAP = 2,2′-bis(diphenylphosphino)-1,1′-binaphthyl] catalyst–polymer framework that was cross-linked with the Ru–BINAP monomer.⁷⁰ The framework was synthesized with an alternating ring-opening metathesis polymerization (alt-ROMP) and supported on BaSO₄. The BINAP ligand was modified at the 5 and 5′ positions with ROMP-active norimido groups, thereby allowing the Ru-(5,5-dinorimido)BINAP monomer to cross-link the polymer to build a framework. The solid catalyst operated continuously over 25 runs (TON = 1000 per run) for the enantioselective hydrogenation of 1′-acetonaphthone (0.1 mol % Ru, 2 mol % KO^tBu, 40 °C, 10 atm H₂ pressure, 2-PrOH, 21 h) without loss in activity (100% yield) or enantioselectivity (95% ee). The leaching of this catalyst in any representative run analyzed by inductively coupled plasma mass spectrometry (ICP-MS) was <0.016% of the total Ru loading. The reaction stopped after 34 runs because the solid catalyst accumulated on the roof of the bomb. Several alt-ROMP Rh^71,72 and Ru⁷³ polymeric systems have been reported by our group.

To our knowledge, there is one report of a polymer-based catalyst for the hydrogenation of esters. Kamer and co-workers immobilized the tridentate PNN ligand ((2-C₅H₄N)CH₂NHCH₂CH₂P(Ph)CH₂-polymer) on Merrifield resin and then metallated it by reaction with RuCl₂(PPh₃)₃.⁷⁴ This polymer hydrogenated methyl benzoate with ∼75 TON per run for three runs (0.9 mol % Ru, 10 mol % KO^tBu, 40 °C, 49 atm, THF, 2 h). The TON of the fourth run dropped to 50 TON.

Ruthenium complexes with tetradentate PNNP ligands are active catalysts for a variety of enantioselective reactions, including hydrogenations and transfer hydrogenations of ketones,⁷⁵ epoxidations of olefins,⁷⁶ Diels–Alder reactions,⁷⁷ and cyclopropanations.⁷⁸ Saudan et al.¹⁵ reported that the Ru–PNNP complex trans-RuCl₂{N,N′-bis({(2-diphenylphosphino)phenyl}methylidene)ethylenediamine} (1) is active for ester hydrogenations (0.05 mol % 1, 5 mol % NaOMe, 100 °C, 49 atm H₂, THF, 1 h, TON = 1980). Although 1 does not contain N–H groups, it is quite reasonable to expect that the imine groups are reduced to the corresponding amines during the initial stages of the hydrogenation.^15,79 We now report a facile synthesis of an ROMP-active Ru–PNNP analogue of 1, its polymerization by alt-ROMP, and its performance as a reusable catalyst for ester hydrogenations.

Results and Discussion

As shown in Scheme 1, we prepared an ROMP-active analogue of 1 by condensing the known diamine, trans-5-norbornene-endo-2,3-diamine (2),⁸⁰ with 2 equiv 2-diphenylphosphinobenzaldehyde to give the norbornene-containing PNNP tetradentate ligand N,N′-bis({(2-diphenylphosphino)phenyl}methylidene)bicyclo[2.2.1]hept-5-ene-2,3-diamine (3). Reaction between 3 and trans-RuCl₂(NBD)Py₂⁸¹ (4, NBD = 2,5-norbornadiene, Py = pyridine) gave the trans-dichloride 5 in 53% yield after purification. Figure 1 shows the solid-state structure of 5 as determined by X-ray diffraction. The solid-state structure confirms the trans-configuration of the chlorides, as is similar to the reported structures for these types of compounds.^15,75,79

Scheme 1. Syntheses of 3 and 5.

Figure 1.

Perspective view of 5 showing the atom-labeling scheme. Nonhydrogen atoms are represented by Gaussian ellipsoids at the 30% probability level. Hydrogen atoms are shown with arbitrarily small thermal parameters, except for aromatic group hydrogens, which are not shown.

We found that ROMP polymerization of 5 with RuCl₂(=CHPh)(PCy₃)₂ (6) as catalyst was very slow (5 mol % 6, 40 °C, CH₂Cl₂), presumably because there was too much crowding between adjacent monomers in the polymer. Instead, we employed alt-ROMP to polymerize 5.^70,73 Briefly, alt-ROMP occurs between a crowded, strained monomer and an uncrowded, unstrained monomer. The more reactive, strained monomer reacts with the metathesis catalyst, but does not polymerize because the resulting ruthenium alkylidene is too crowded. Instead, the akylidene reacts with the less crowded monomer, typically cis-cyclooctene (COE) to generate a new, uncrowded alkylidene. This uncrowded alkylidene then reacts with the crowded, strained monomer and so on.^70,73 We found that 4 equiv of COE smoothly underwent alt-ROMP with 5 employing 1 mol % 6 as catalyst (40 °C, 20 h, CD₂Cl₂). Figure 2 shows the structures and NMR spectra of 5 and the resulting linear polymer 7. The signals for the bridging methylene and olefin groups in 5, as well as those for the olefin group in COE were absent in the ¹H NMR spectrum of the polymer. The ³¹P NMR spectrum of the polymer contained major and minor broadened peaks, shifted ∼2 ppm from those of 5. We believe the slight shift resulted from ring opening during alt-ROMP, and the major and minor species arise from the trans- and cis-olefin geometries in the polymer.

Figure 2.

Alt-ROMP polymerization of 5 and COE; the NMR spectra of 5 and the linear polymer 7.

Attempts to immobilize the polymeric catalyst 7 by depositing it as film on BaSO₄, as we did previously with the highly cross-linked 5,5′-bis(norimido)BINAP–Ru polymer framework,⁷⁰ were met with mixed results. Specifically, we evaluated 7/BaSO₄ with the hydrogenation of methyl benzoate in THF (0.1 mol % Ru in 7, 10 mol % KO^tBu, 40 °C, 50 atm, THF, 4 h, eq 1)

1

THF is often the solvent in which homogeneous ester hydrogenations operate with the highest rates and yields.^25,26,28 Although the first run reached 95% conversion (TON = 950), the filtrate was deeply colored and the conversion of the second run was significantly lower (65%). Not surprisingly, the solubility of the linear polymer 7 limited its isolation and reuse by filtration.

We expect that it is difficult to prevent polymers from dissolving under continuous stirring at 80 °C in THF at 21 h per run. We therefore prepared the cross-linking agent 8 (Scheme 2) in one step by condensation of 2 equiv of cis-5-norbornene-endo-2,3-dicarboxylic anhydride with ethylenediamine in acetic acid. The ROMP polymerization of 8 catalyzed by 6 was sluggish in CH₂Cl₂ at 40 °C. Presumably, the crowding around the olefin in the endo-norbornimide groups hindered the polymerization. The cross-linker 8 did, however, undergoes alt-ROMP with 5 and COE. The optimized ratio of 5/8/COE/6 was 1:2:8:0.02. A 2:1 ratio of 5/8 was employed to form a highly cross-linked polymer to minimize dissolution in hot THF. This mixture underwent complete polymerization after 48 h in CD₂Cl₂ at 40 °C. NMR spectra were recorded, while the alt-ROMP was underway, which showed that the olefin groups in 5 and 8 polymerized at comparable rates. Specifically, 21% of 5, 9% of the norimido groups in 8, and 5% of COE had polymerized after 1.5 h. The corresponding conversions were 89, 61, and 21% after 20 h. Both 5 and 8 were almost fully consumed after 48 h, with ∼20% COE remaining in solution. The comparable rates for polymerization of 5 and 8 indicate that the resulting polymer more resembles an alternating copolymer rather than alternating blocks of 5 or 8. Further, it is highly improbable that the cross-linked product is a linear, ladder polymer. It is more likely a three-dimensional framework cross-linked by 8 and spaced by COE + 5. The ¹H NMR spectrum of the framework contained highly broadened peaks and did not contain signals from the olefin or bridging methylene groups arising from unreacted monomers. The peaks for 5 in the ³¹P NMR spectrum of the polymer (Figure 3) were broadened and shifted upfield by ∼2.1 ppm from those of the unreacted monomer. After ∼40 h reaction, the entire reaction mixture had converted into a uniform, static gel that could not be dissolved in CD₂Cl₂. This gel shrank to a compact, brittle piece of insoluble plastic upon evaporation of the solvent.

Scheme 2. Preparation of the Cross-Linked Polymeric Catalyst 9 Dispersed on BaSO₄.

Figure 3.

³¹P NMR spectra recorded during polymerization of 5, 8, and COE with 2 mol % 6 (40 °C in CD₂Cl₂). Spectrum of (a) initial, (b) 1.5 h, (c) 20 h, and (d) 48 h.

In previous reports,^70,73 we utilized an inert, high-surface-area material such as BaSO₄ to support the polymerized catalysts. The functions of the support were to improve mass transport, to engender mechanical stability for long-term batch reactions with stirring, and to act as a filtration aid. The cross-linked polymer gel 5/8/COE could not be diluted or dissolved after it formed, and so could not be deposited onto the BaSO₄ by simple solvent evaporation. Instead, we carried out the alt-ROMP polymerization of 5, 8, COE, and 6 directly in the presence of BaSO₄. Specifically, a solution of 5, 8, COE, and 6 was mixed with solid BaSO₄, with the reaction solution just covering the BaSO₄. The polymerization was allowed to proceed for 66 h heating at 40 °C. The resulting catalyst–polymer framework on BaSO₄ (9, Scheme 2) was ground with a mortar and pestle, washed with CH₂Cl₂, and dried under vacuum. An NMR study confirmed that ∼7% of the Ru leached into the wash, presumably as lower-molecular-weight oligomers.

We hydrogenated methyl benzoate with loadings of 40 000 equiv per run to evaluate 9 (0.0025 mol % Ru in 9, 10 mol % KO^tBu, 80 °C, 50 atm, THF, 21 h, eq 2). These conditions are in line with those utilized for high TON ester hydrogenations in the literature.²³⁻²⁶ The product mixtures contained methyl benzoate, benzyl alcohol, and the transesterification product benzyl benzoate (Table S1). The TON of the first run was 32 960, corresponding to 82% conversion. The isolated yield after filtration was 30 340 TON using an internal standard. Based on the pressure drop, the hydrogenation was very rapid in the beginning, completing ∼15 000 turnovers (TO) during the first 1 h, ∼20 000 after 3 h, and ∼33 000 after 21 h. Similar initial burst activities were reported for homogeneous ester hydrogenations.^16,26,39,82 We found that a control homogeneous hydrogenation of methyl benzoate with the monomeric catalyst 5 (methyl benzoate/KO^tBu/5 = 25 000:2500:1, 90 °C, 50 atm) proceeded with a similar burst followed by a slowdown (∼14 000 TO after 1 h, ∼16 000 after 2 h, ∼17 000 after 3 h). Although the structure of 5 is modified by the ring-opening polymerization, the similarity in turnover frequencies between 5 and the polymer indicates that mass transport does not significantly limit the rate of the polymerized catalyst. Presumably, the combination of the open framework structure of the polymer, the BaSO₄ support, and the 80–90 °C THF solvent minimized mass transport effects. We suggest this strategy be incorporated into future polymeric catalyst designs.

2

Adding more hydrogen and substrate after the homogeneous hydrogenation had no effect on the rate after the reaction had slowed. Thus, the slowdown is either due to product inhibition, a decrease in basicity of the medium as the hydrogenation proceeds, or catalyst decomposition. In contrast, the polymeric catalyst was easily reused after filtration, and the second run proceeded with 29 720 TOs and a similar initial burst in activity. Although some loss in activity occurred between the first and second runs, catalyst decomposition cannot explain the decrease in rate after the burst during the homogeneous and heterogeneous hydrogenations. Further, the hydrogenation of benzyl benzoate over 9 proceeds in a similar yield (74%) and TON (∼29 600) to that of methyl benzoate. Considering this similarity in reactivity, and that only small amounts of benzyl benzoate (∼3.8%) are in the product mixture from the hydrogenation of methyl benzoate, benzyl benzoate does not significantly inhibit the hydrogenation of methyl benzoate. The yield of the hydrogenation of benzyl benzoate with less base (1 mol %) was lower (40%). In our prior mechanistic studies,^82,83 we directly observed that alcohols inhibit ketone and ester hydrogenations by forming Ru-alkoxides, especially in the absence of base. We expect that this inhibition is stronger for ester hydrogenations for two reasons. The first is that esters are intrinsically less reactive than ketones. The second is that the hydrogenation of a ketone forms 1 equiv of a secondary alcohol, whereas the hydrogenation of a methyl ester forms 2 equiv of primary alcohols. Primary Ru-alkoxides are more stable than secondary alkoxides because of steric crowding.⁸⁴ We therefore expect that the buildup 2 equiv of primary alcohols partially explains the decrease in rate over the course of these ester hydrogenations.

In other research, we reported two direct observations that may explain how the excess base increases the rates of these ester hydrogenations. The first is that base removes alkoxide ligands from these types of Ru compounds by an elimination reaction involving deprotonation of an N–H group.⁸²⁻⁸⁴ We therefore expect that one role of the excess base in these ester hydrogenations is to minimize the inhibition by the 2 equiv of primary alcohol products through this elimination reaction. In another study, we showed that the reducing power of these catalysts is dramatically increased by deprotonation of a N–H group. For example, we observed that the deprotonated version of Noyori’s prototypal dihydride catalyst, (BINAP)Ru(H)₂(NH⁽⁻⁾CH(Ph)CH(Ph)NH₂), reduces imides on mixing at −80 °C in THF.⁸⁵ We propose that the combination of these two processes partially explains the need for excess base during active ester hydrogenations. For studies on the ester hydrogenation, see refs (86−90). Also, the basicity of these ester hydrogenations decreases as the medium changes from pure aprotic (THF + ester) to a protic medium as the 2 equiv of alcohol products form. This decrease in basicity would limit the alkoxide elimination, the catalyst activation, and the deprotonation of the N–H groups. We propose that the combination of primary alcohol product inhibition and the decrease in reaction medium’s basicity all contribute to the drop in rate after the initial burst in these ester hydrogenations. We suggest that the design of high-turnover-number ester hydrogenation catalysts in the future incorporates methods to minimize these types of inhibition.

Figure 4 plots TON versus run. The TONs of the first two runs were similar. The catalyst remained active in the subsequent runs, but gradually slowed. Run 5 proceeded with 14 760 TO, with a total over the five runs = 121 680, the highest TON reported for the hydrogenation of an ester with a molecular catalyst.²⁴⁻²⁷

Figure 4.

Turnovers per run, and total TON over five hydrogenations of methyl benzoate (0.0025 mol % catalyst, 10 mol % KO^tBu, 80 °C, 50 atm, THF, 21 h).

The amounts of Ru leached during runs 1–4 were 5.6, 3.8, 1.4, and 1.1%, respectively, of the total catalyst loading, as determined by ICP-MS. These results show that although leaching was reduced by the cross-linked polymer, it was not eliminated. These results demonstrated the heterogeneity of the catalyst. Specifically, if the activity resulted from dissolved Ru complexes leached into solution, run 4 occurred with an impossibly high ∼2 million TON. The Ru leaching can be attributed to either/both dissolution of lower-molecular-weight oligomers over 21 h stirring in 80 °C THF per run and/or some sort of active site decomposition. Further studies are required to study the mechanism of decomposition. Regardless, the total loss in Ru to leaching was only 11.9%, much less than the 55% drop in activity between run 1 and run 5. We speculate that the gradual drop in activity over the five runs was due mainly to some sort of decomposition of the catalyst that did not release Ru out of the polymer framework. This long-term decomposition was either innate to the catalyst under these conditions, or it was caused by trace amounts of systematic impurities.

Conclusions

We devised a convenient synthesis of a defined, dispersed Ru–PNNP cross-linked catalyst–polymer framework utilizing alt-ROMP. The highest reported total TON for the hydrogenation of an ester was obtained by reusing the catalyst. The catalyst had similar activity to a homogeneous analogue, indicating mass transport did not limit the activity. The study of this system provided insights into the mechanism of catalyst deactivation, how to minimize mass transport limitations, and guidelines for future catalyst design. Further research is underway in these laboratories to improve the TON, to further minimize leaching, and to prepare resolved versions of this chiral catalyst.

Experimental Section

General Information

Unless stated otherwise, all of the synthetic experiments were carried out under an inert atmosphere (dinitrogen or argon), using Schlenk techniques and a glovebox. High-pressure hydrogenations were carried out in a high-pressure Parr 4750 reactor with an extra coned pressure fitting for adding and filtering reaction mixtures. The gases N₂ (4.8), Ar (4.8), and H₂ (5.0) were supplied by Praxair. Deuterated solvents were purchased from Sigma-Aldrich and Cambridge Isotope Laboratories. All common solvents were used as received or distilled over proper drying agents under N₂ when it was necessary: CH₂Cl₂, CHCl₃, and hexanes were distilled over CaH₂; methanol was distilled over Mg/I₂; EtOH was dried over 3 Å molecular sieves and distilled over CaO; and THF was purified by distillation over Na/benzophenone. Triethylamine, pyridine, ethylenediamine, and piperidine were distilled over CaH₂. Norbornadiene and cis-cyclooctene were purified by simple fraction distillation. Methyl benzoate was obtained from Sigma-Aldrich and fractionally distilled over CaH₂ prior to hydrogenation. Benzyl benzoate was obtained from Acros Organics and distilled over CaH₂ under N₂ gas prior to use. The following chemicals were obtained from suppliers without further purification: trans-3,6-endomethylene-1,2,3,6-tetrahydrophthaloyl chloride, cis-5-norbornene-endo-2,3-dicarboxylic anhydride, and 2-diphenylphosphinobenzaldehyde were obtained from Sigma-Aldrich; NaN₃ was obtained from Fisher Scientific Company; NaBH₄ were obtained from BDH Chemicals. First-generation Grubbs catalyst was obtained from Strem Chemicals. BaSO₄ (white reflectance standard) was obtained from Eastman Chemical Co., Inc., washed with distilled CH₂Cl₂, dried under vacuum, and stored under nitrogen before use.

¹H, ¹³C, and ³¹P NMR spectra were recorded with a Varian Inova 500 MHz spectrometer, an Agilent/Varian 400 MHz spectrometer, or an Agilent VNMRS 700 MHz spectrometer. ¹H and ¹³C chemical shifts were recorded relative to tetramethylsilane using the signals from the residual protons in the deuterated solvent as the internal standard. ³¹P chemical shifts were reported relative to an H₃PO₄ external standard. Elemental analysis was carried out with a Flash 2000 Organic Elemental Analyzer made by Thermo Fisher. ICP-MS was carried out with a PerkinElmer ’s Elan 6000 ICP-MS.

Catalyst Preparation

trans-2,3-Diaminonorborn-5-ene Dihydrochloride (Racemate) (2)

This experiment was carried out in air, and it is a modification of a literature procedure.⁸⁰ The solvents were used as received, and the potentially explosive diazide intermediate was not isolated. Warning: this diazide may be explosive when concentrated. Do not isolate it by removing the solvent. Proceed immediately to the next step, as detailed in this procedure.

Acetone (7.5 mL) was added to an Erlenmeyer flask containing trans-3,6-endomethylene-1,2,3,6-tetrahydrophthaloyl chloride (3.085 g, 14.1 mmol), followed by a solution of NaN₃ (2.747 g, 42.2 mmol) in deionized water (7.5 mL). The mixture was stirred for 1 h at room temperature (rt). The aqueous solution was then extracted with toluene (3 × 25 mL), and the combined organic solution was washed with saturated NaHCO₃ (30 mL), brine (30 mL), and dried over Na₂SO₄. The solution was gravity-filtered into a round-bottom flask, refluxed for 2 h, allowed to cool to room temperature, and 120 mL of 8 M HCl was added. This mixture was refluxed for 4 h. Over this period, the color of the solution slowly turned light purple. After the solution cooled to room temperature, the aqueous layer was separated and washed with diethyl ether (2 × 50 mL). After removing the water under vacuum, the resulting brownish solid was washed with THF (2 × 20 mL) and then dried under vacuum to give 1.811 g of product with a yield of 65%.

¹H NMR (498.119 MHz, D₂O, 27 °C): δ 1.87–1.93 (2H, AB pattern, m, bridging CH₂), 3.16 (1H, m, CH), 3.21 (1H, m, CH), 3.29 (1H, m, CH), 3.88 (1H, m, CH), 6.34 (1H, m, CH, olefin), 6.53 (1H, m, CH, olefin). ¹³C{¹H} NMR (125.266 MHz, CDCl₃, 27 °C): δ 45.0 (CH₂), 45.1 (CH), 46.7 (CH), 55.6 (CH), 56.0 (CH), 134.1 (CH), 138.7 (CH). The HCl salt of the diamine was hydroscopic and required drying under high vacuum to obtain a satisfactory elemental analysis. Elemental analysis, calcd: C, 42.66; H, 7.16; N, 14.21. Found: C, 41.09; H, 6.91; N, 13.80.

N,N′-Bis({(2-diphenylphosphino)phenyl}methylidene)bicyclo[2.2.1]hept-5-ene-2,3-diamine (Racemate) (3)

This synthesis was carried out under N₂. Compound 2 (248.6 mg, 1.26 mmol assuming no water uptake) was weighed into a Schlenk tube along with 3 Å molecular sieves (565.6 mg, activated). The tube was degassed, then triethylamine (700 μL) and EtOH (5 mL) were added. 2-Diphenylphosphinobenzaldehyde (731.6 mg, 2.52 mmol) was dissolved in EtOH (12 mL) under gentle heating and transferred to the Schlenk tube. The reaction was stirred for 16 h at rt. The NMR spectrum of an aliquot showed that the aldehyde was present in 16% excess, so another 39.6 mg of diamine was added to the reaction. After another 24 h of stirring, the reaction mixture was filtered through Celite and pumped under high vacuum to dry. Dichloromethane (∼10 mL) was added to dissolve the compound, and the resultant solution was washed with water three times and then dried over Na₂SO₄. After filtration and drying under high vacuum, 641 mg of product was obtained with a yield of 76%.

¹H NMR (699.763 MHz, CD₂Cl₂, 27 °C): δ 1.43 (1H, m, bridging CH), 1.94 (1H, m, bridging CH), 2.38 (1H, m, CH next to olefin), 2.64 (1H, m, CH next to olefin), 2.76 (1H, m, CH next to imine), 3.35 (1H, m, CH next to imine), 6.04 (1H, dd, J = 5.64 Hz, 2.82 Hz, olefin proton), 6.08 (1H, dd, J = 5.64 Hz, 3.07 Hz, olefin proton), 6.82–7.79 (28H, aromatic), 8.43 (1H, d, J = 3.71 Hz, imine), 8.61 (1H, d, J = 4.10 Hz, imine). ¹³C{¹H} NMR (175.972 MHz, CD₂Cl₂, 27 °C): δ 45.9 (CH₂), 49.4 (CH), 51.7 (CH), 78.7 (CH), 79.0 (CH), 129.4–137.5 (multiple aromatic CH), 137.7–140.1 (multiple aromatic C), 158.8–159.3 (multiple CH). ³¹P{¹H} NMR (161.839 MHz, CD₂Cl₂, 27 °C): δ −11.6 (1P, s), −10.4 (1P, s). Elemental analysis, calcd: C, 80.82; H, 5.73; N, 4.19. Found: C, 79.01; H, 5.72; N, 4.18.

Dichloro{N,N′-bis({(2-diphenylphosphino)phenyl}methylidene)bicyclo[2.2.1]-hept-5-ene-2,3-diamine}ruthenium (Racemate) (5)

The complexation reaction was carried out by mixing 3 (387.0 mg, 0.58 mmol), RuCl₂(NBD)(Py)₂,⁸¹ and 4 (243.1 mg, 0.58 mmol) in chloroform (15 mL, distilled) under N₂. The reaction mixture was heated at 60 °C for 4 days, after which the solution was dried under high vacuum. The resulting solid was purified by a column of neutral Al₂O₃ (30 cm³ in bulk volume) with dichloromethane. The orange band on the column was collected and dried under high vacuum, and 257 mg of red powdery compound was obtained after recrystallization using dichloromethane and hexane, with a yield of 53%. Crystals for X-ray crystallography were obtained by a diffusion recrystallization in an NMR tube using dichloromethane and hexane.

¹H NMR (699.763 MHz, CD₂Cl₂, 27 °C): δ 2.28–2.31 (2H, AB pattern, m, bridging CH₂), 3.31 (1H, m, CH next to olefin), 3.34 (1H, m, CH next to olefin), 4.93 (1H, m, CH next to imine), 5.54 (1H, m, CH next to imine), 6.45 (1H, m, CH, olefin), 6.66 (1H, m, CH, olefin), 6.68–7.69 (28H, aromatic), 8.52 (1H, d, CH, J = 9.22 Hz, imine), 8.66 (1H, d, J = 9.34 Hz, CH, imine). ¹³C{¹H} NMR (175.972 MHz, CD₂Cl₂, 27 °C): δ 38.2 (CH), 42.7 (CH), 48.9 (CH₂), 78.5 (CH), 79.2 (CH), 127.3–127.7 (multiple aromatic peaks), 129.0–130.1 (multiple aromatic peaks), 131.1–132.1 (multiple aromatic peaks), 134.0–134.1 (multiple aromatic peaks), 135.3–137.0 (multiple aromatic peaks), 141.1 (CH), 158.3 (CH), 158.9 (CH). ³¹P{¹H} NMR (161.839 MHz, CD₂Cl₂, 27 °C): AB pattern, δ 51.6 (1P, d, J_pp = 27.5 Hz), 51.3 (1P, d, J_pp = 27.5 Hz). Elemental Analysis, calcd: C, 64.29; H, 4.56; N, 3.33. Found: C, 64.17; H, 4.77; N, 3.31.

1,2-N-Di(cis-5-norbornene-2,3-endo-dicarboximido)-ethane (8)

cis-5-Norbornene-endo-2,3-dicarboxylic anhydride (2171.8 mg, 13.2 mmol), ethylenediamine (396.9 mg, 6.6 mmol), and acetic acid (32 mL, distilled) were refluxed at 125 °C for 4 h under N₂. The solution was poured into 250 mL of water, and the product was extracted with dichloromethane (DCM) (2 × 150 mL). The combined organic layer was washed with saturated NaHCO₃ (3 × 200 mL) and water (200 mL) and then dried over Na₂SO₄. A crude product (2.54 g) was recovered after filtration and rotary evaporation. The crude product was recrystallized in boiling EtOAc (290 mL), and 1.52 g crystalline product was obtained with a yield of 65%.

¹H NMR (498.118 MHz, CDCl₃, 27 °C): δ 1.52 (2H, m, bridging CH), 1.71 (2H, m, bridging CH), 3.23 (4H, m, α CH to carbonyl), 3.34 (4H, m, CH next to olefin), 3.47 (4H, s, CH₂), 6.06 (4H, m, olefin CH). ¹³C{¹H} NMR (125.266 MHz, CDCl₃, 27 °C): δ 36.4 (CH₂), 44.7 (CH), 45.9 (CH), 52.3 (CH₂), 134.4 (CH), 177.6 (C=O). Elemental analysis, calcd: C, 68.17; H, 5.72; N, 7.95. Found: C, 68.16; H, 5.79; N, 7.93.

Alt-ROMP Polymerization of 5 and COE, NMR study

Catalyst monomer 5 (10.0 mg, 0.012 mmol, 1 equiv), COE (6.0 μL, 0.046 mmol, 4 equiv), and 6 (0.1 mg, 0.0001 mmol, 0.01 equiv) were mixed in CD₂Cl₂ (0.8 mL) in an NMR tube. The NMR was taken after 20 h of heating at 40 °C.

Supported Alt-ROMP Polymer of 5 and COE

Compound 5 (21.3 mg, 0.025 mmol, 1 equiv), COE (13.2 μL, 0.100 mmol, 4 equiv), and 6 (0.2 mg, 0.00024 mmol, in standard solution) were mixed in 0.8 mL of CD₂Cl₂. The polymerization was completed after 48 h at 40 °C. The resulting polymer was diluted to 2 mL with DCM and then transferred to a slurry of BaSO₄ (3.76 g) in DCM (20 mL). The slurry was stirred for 30 min, then slowly pumped under high vacuum to remove the solvent, resulting a pink powder. The powder was washed with MeOH three times and then dried under high vacuum. The MeOH wash was collected and analyzed using 1H NMR with 1,3,5-trimethoxybenzene as internal standard, indicating 53% of COE and 14% of Ru moiety dissolved in the MeOH. We suspect that these are lower-molecular-weight oligomers. The resulting supported catalyst had a loading of 0.0058 mmol Ru/g BaSO₄.

Hydrogenation with Supported Linear Polymer Catalyst (Two Runs)

The hydrogenation was carried out in a high-pressure reactor. For the first run, 1.88 g of methyl benzoate (1.35 mL, 1000 equiv) was added to the reactor through a cannula, followed by supported catalyst (0.0107 mol Ru, 1 equiv, slurry in THF) and KO^tBu (120 mg, 100 equiv, THF solution) through a cannula, with 12 mL of THF in total. The reactor was sealed and pressurized. After 4 h at 50 atm H₂ under 40 °C, the reaction was stopped and filtered through a cannula filter under H₂. The next run was set up by adding methyl benzoate (1.35 mL, 1000 equiv) and KO^tBu (120 mg, 100 equiv) in THF (12 mL) and the hydrogenation was carried out under the same conditions.

The filtrate from each run was pumped to dryness with a rotovap. Water (15 mL) was added to the mixture and the mixture was extracted with diethyl ether twice. The ether layer was dried over Na₂SO₄ and the ether was removed with a rotovap. The conversions were determined by the ratio between each compound in the ¹H NMR spectra.

Cross-Linked Polymerization Study in an NMR Tube

Compound 5 (11.0 mg, 0.013 mmol, 1 equiv), compound 8 (8.8 mg, 0.025 mmol, 2 equiv), and first-generation Grubbs catalyst solution (0.22 mg catalyst, 0.00026 mmol, 0.02 equiv, quantified with a standard solution in CD₂Cl₂) were placed in three separate NMR tubes under N₂. They were transferred into the same NMR tube after dissolving the solids in CD₂Cl₂; the total volume after mixing was 0.7 mL. cis-Cyclooctene (11.3 mg, 0.103 mol, 8 equiv) was added to the NMR tube, and NMR spectra of the reaction solution were recorded immediately. The tube was placed in a 40 °C oil bath, and the NMR spectra were recorded after 1.5, 20, and 48 h. The reaction solution slowly turned into a gel in ∼40 h, which could not flow anymore (Figure S1). However, the solution NMR spectra were still available with the gel.

Polymerization in the Presence of BaSO₄ To Make the Immobilized Catalyst (9)

The experiment was carried out under N₂. A polymerization solution was prepared first as described above. It contained the following chemicals: 5 (21.2 mg, 0.025 mmol, 1 equiv), 8 (17.7 mg, 0.049 mmol, 2 equiv), 6 (0.4 mg, 0.00049 mmol, 0.02 equiv, quantified with a standard solution in CD₂Cl₂), cis-cyclooctene (22.2 mg, 0.201 mmol, 8 equiv), and 1.3 mL CD₂Cl₂. An NMR spectrum was recorded to confirm the ratio between each compound. Most (90%) of this solution was transferred into a screwcap Schlenk tube with 4.0133 g of BaSO₄ inside, and 10% was left in the NMR tube as an indicator of the reaction process. Both the Schlenk tube and the NMR tube were placed in a 40 °C oil bath. The 10% solution in the NMR tube became a gel within 40 h. The reaction in the Schlenk tube was stopped after 66 h to make sure that the reaction had ended as it could not be monitored directly. The CD₂Cl₂ was removed under high vacuum, and then the solid was ground with a mortar and pestle in a glovebox. Big chunks were broken into powder, and the powder was ground as fine as possible. The resultant powder was washed with distilled DCM (3 × 20 mL), and 3.827 g of supported catalyst was recovered after drying (some could not be recovered from the mortar). A picture of the powdery supported catalyst is shown in Figure S2. The wash was collected, and the DCM was removed under high vacuum. The residue was dissolved in CDCl₃ with 1,3,5-trimethoxybenzene and triphenylphosphine as internal standards. NMR spectra confirmed that 0.0015 mmol Ru complex leached out into the wash, corresponding to 6.6% of the total Ru initially in the Schlenk tube. The loading of the resulting supported catalyst was calculated as follows: [90% × 0.025 mmol × (3.827 g/4.013 g) – 0.0015 mmol]/3.827 g = 0.0053 mmol Ru/g BaSO₄.

Catalytic Hydrogenation Experiments

Catalytic Hydrogenation with Reusing the Catalyst

The whole process was carried out under Ar or H₂, and all of the transfers were carried out using a cannula. The reaction was carried out in a high-pressure reactor with an extra pressure fitting where addition and filtration could be conducted. The reactor was assembled with a glass liner and a stir bar, then purged with Ar. The reactor was pressurized to 50 atm with H₂, evacuated, and purged with Ar again. Using Schlenk techniques, the supported catalyst 9 (0.2833 g, 0.0015 mmol Ru) was transferred with methyl benzoate (8.08 g, 7.48 mL, 59.4 mmol) as slurry to the reactor under Ar, followed by KO^tBu (665.4 mg, 5.9 mmol) solution in THF (16 mL). The reactor was sealed and pressurized gradually to 50 atm, and the valve on the reactor was closed to isolate it from the H₂ tank. The oil bath placed underneath was set to 80 °C, and the stirring speed was 600 rpm. (Note that the reactor would be repressurized if the pressure dropped below 40 atm as the reaction proceeded.) After 21 h, the reactor was cooled to rt and the pressure was slowly released. The reaction mixture was filtered through a cannula filter into a preweighed flask under H₂ atmosphere, and once bubbles appeared from the cannula tip, the filtration was stopped. To set up the next run, methyl benzoate (8.08 g, 7.48 mL, 59.4 mmol) was added to the reactor immediately after the filtration, followed by KO^tBu (665.4 mg, 5.9 mmol) solution in THF (16 mL). The reactor was pressurized and heated as described above, and this process was repeated for multiple runs. The collected filtrate was weighed with a flask, and the NMR spectra of a weighed aliquot of the filtrate were recorded in CDCl₃, with 1,3,5-trimethoxybenzene as the internal standard to calculate the conversion and yield. The details of each hydrogenation run are listed in Table S1. Figure S3 is an example of the ¹H NMR spectrum of this hydrogenation (run 1).

A white solid was found in the bomb on completion of the reaction. After the last run (run 5), the reaction mixture was cannulated into a side-arm flask under N₂, and a filtration was carried out to isolate the product. The remaining solid was washed with dry ether, and the organic compounds in the wash were combined with the filtrate of run 5. Next, the solid was pumped to dry under high vacuum, and 0.714 g of white powder was isolated. Its NMR results confirmed the solid to be potassium benzoate with the remaining catalyst/BaSO₄ powder (0.283 g BaSO₄ would remain theoretically), without anything else visible in the NMR spectra.

A hydrogenation process was carried out by the same procedure using benzyl benzoate as substrate (12.63 g, 0.059 mol). The reaction proceeded in 74% yield, ∼29 600 turnovers. Figure S18 shows the ¹H NMR spectrum of the product mixture.

Homogeneous Hydrogenation of Methyl Benzoate

Compound 5 (1.25 mg, 0.0015 mmol, 1 equiv), methyl benzoate (4.6 mL, 36.9 mmol, 25 000 equiv), and KO^tBu (419 mg, 3.71 mmol) were added into a high-pressure reactor with 16 mL of THF. The hydrogenation was carried out at 90 °C under 50 atm H₂. The approximate TON within each hour was monitored using a pressure gauge, which was ∼14 000 turnovers after 1 h, ∼16 000 after 2 h, and ∼16 700 after 3 h. The reaction was stopped after 3 h by cooling and releasing the H₂ pressure. A tiny aliquot was taken under H₂, and its NMR spectra confirmed a conversion of 72%, corresponding to 18 000 turnovers. Then, another 18 000 equiv methyl benzoate (3.3 mL) was added to the reactor. The reaction was brought back to 90 °C and 50 atm H₂. The pressure drop of the first hour after replenishing the substrate indicated that the TON was less than 1000.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsomega.9b01682.

• Crystal structure determination; NMR spectra of catalyst precursors, polymers, and hydrogenation reactions; tables showing extents of hydrogenation; and the leaching analysis by ICP-MS (PDF)

• Crystal structure of Ru–PNNP monomer 5 (CIF)

The authors declare no competing financial interest.