1 Introduction
In recent years, much attention has been paid to the search of new chelated ligands containing mixed functionalities as a potential tool to achieve more efficient and selective catalytic reactions. In particular, the use of the so-called hemilabile ligands 〚1〛 represents a step forward in catalysis as they may induce a higher reactivity of their transition metal complexes and allow the stabilisation of reactive transition metal centres. Due to the intrinsic nature of such ligands, the labile group can be displaced easily and reversibly from the metal centre, thereby allowing a catalytic reaction to occur. Such a potential has been widely used for bidentate ligands 〚2〛 but much less for tridentate ones 〚3, 4〛. However, Mathieu and al. reported on the use of ruthenium complexes bearing a tridentate P–N–O ligand in the catalytic transfer hydrogenation of ketones 〚5〛. More recently, the groups of Vrieze and van Leeuwen have described interesting reactivity properties for palladium complexes of a new hemilabile P–N–N ligand containing phosphine, imine and pyridyl donor groups 〚6, 7〛. These findings prompted us to synthesise a novel closely related ligand 2 and to investigate the coordination properties of 1 and 2 shown in Fig. 1, in the presence of rhodium (I) and iridium (I), as a first step towards a more detailed study of their catalytic properties.
Representation of ligands 1 and 2.
Herein, we describe preliminarily the synthetic routes to achieve this goal and report on the catalytic activity of the new complexes towards the hydrogenation of linear olefins, cyclic olefins and sterically demanding olefins.
2 Experimental
2.1 General considerations
All syntheses were performed under a pure dinitrogen or argon atmosphere using Schlenk techniques on an inert gas/vacuum manifold or in a dry box equipped with a dry-train HE-493 inert gas purifier. Solvents were dried by standard methods and distilled under nitrogen. AgBF4 and AgSO3CF3 were purchased from Strem and used as received. 〚Rh(COE)2Cl〛2 〚8〛, 〚Rh(NBD)Cl〛2 〚9〛, and 〚Ir(COD)Cl〛2 〚10〛 were prepared according to published methods. NMR spectra were recorded at ambient probe temperature on a Bruker AC-300 spectrometer operating at 300 MHz for 1H, 75 MHz for 13C and 121.5 MHz for 31P{1H}. Chemical shifts are reported in parts per million versus SiMe4 for 1H and 13C calibrated against the residue solvent resonance as an internal reference. 31P NMR spectra were referenced to 85% aqueous H3PO4.
The analyses of the catalytic reactions were carried out using a Hewlett Packard 5890 Series II Gas Chromatograph equipped with a flame ionisation detector (FID) and a methyl silicone gum column of length 10 m, diameter 0.53 mm and 2.65 μm film thickness.
2.2 General procedure for the hydrogenation studies
A typical experiment was set up using a stainless steel autoclave containing 21 μmol of catalyst in 15 ml of distilled and degassed solvent, and the substrate in excess (500 equiv). The apparatus was degassed and refilled three times with H2 and pressurized adequately. The autoclave that was equipped with a pressure gauge and an internal thermocouple was maintained in an oil bath at a given temperature. The hydrogenation reactions were followed by 1H NMR and GC analyses on samples that were taken out regularly. Except for the trans-stilbene, which, as a solid, was used without purification, traces of peroxides were removed from all the liquid olefins by passing through an alumina column. A trap-to-trap distillation followed by the classical freeze-pump-thaw technique for removal of oxygen allowed storage of the substrates under inert atmosphere.
In the catalytic runs, the catalysts were isolated prior to use as described above or generated in situ. In the latter case, a catalytically active solution was prepared as follows: after dissolution of 1 equiv of 〚Rh(COE)2Cl〛2 in the appropriate solvent, a solution of 2 equiv of the ligand 1 or 2 was added dropwise under nitrogen and allowed to react for 15 min, before addition of the substrate, and eventually the co-catalyst.
3 Results and discussion
In analogy with the synthesis by Vrieze et al. 〚6, 7〛, the new ligand 2 was obtained according to equation (1), by condensation reaction of 2-(diphenylphosphino)benzaldehyde and 2-(2-aminomethyl)pyridine in good yield (79%).
The complexation studies of 1 and 2 in the presence of rhodium and iridium have been carried out in a straightforward manner as shown in equations (2) and (3). Treatment of the rhodium precursor 〚Rh(COE)2Cl〛2 with 1 and 2 in toluene yields the neutral complexes 3 and 4, respectively. The cationic rhodium and iridium complexes 5–7 were prepared by halide abstraction with AgCF3SO3 or AgBF4, followed by displacement of the diene ligand by 1 or 2.
Ligand 2 and complexes 3–7 were characterised by multinuclear NMR, IR, FAB–MS and elemental analysis. All the data are fully consistent with the structural formula and will be reported and discussed in detail elsewhere. Furthermore, X-ray crystallographic analysis of 5 and 7 confirms the tridentate coordination of ligand 1 and indicates a bipyramidal geometry around rhodium or iridium.
In order to investigate the reactivity of the new complexes, we firstly explored the behaviour of 〚Rh(1)Cl〛 in the presence of H2. Interestingly, exposing a degassed C6H6 solution of 3 to H2 (10 bar), with stirring for 2 h, resulted in the formation of 〚Rh(1)H2Cl〛, as shown in equation (4).
The new species is characterised in C6D6 by its 31P{1H} NMR spectrum, which displays a doublet at 39.02 ppm (1JRh–P = 136.6 Hz) and by its 1H NMR pattern, which appears as two sets of resonances: a doublet of doublet at –17.35 ppm (1JRh–H = 17.1 Hz, 2JP–H = 9.4 Hz) and a multiplet at –15.60 ppm. The above data are consistent with two hydride ligands in a cis position.
These results prompted us to test the catalytic activities of the new complexes. Preliminary experiments with all the iridium compounds have indicated a low activity towards the hydrogenation of imines and olefins. Our efforts have then been concentrated in the study of the rhodium complexes as potential hydrogenation catalysts towards various olefins including an easily reduced linear olefin (1-hexene), cyclic olefins (cyclohexene, cyclopentene or cyclooctene), and sterically demanding olefins (1-methylcyclohexene, α-methylstyrene or trans-stilbene). The data relative to the catalytic properties of the neutral complexes 〚Rh(1)Cl〛 3 and 〚Rh(2)Cl〛 4 are listed in Tables 1 and 2, and those concerning the anionic compound 〚Rh(NBD)(1)〛(BF4) 5 in Table 3.
Hydrogenationa of 1-hexene and cyclic olefins in the presence of 3 and 4.
| Entry | Substrate | CatalystI/II | PH2 (atm) | Time (h) | Conversion (%) | t1/2 (h)c | iTOF (h–1)d |
| 1 | 3I | 5e | 70 | 99 | 14.5 | 50 | |
| 2 | 3I, b | 5 | 15 | 99 | 3 | 180 | |
| 3 | 3II | 2 | 11 | 99 | 4 | 190 | |
| 4 | 3II, b | 2 | 7 | 100 | 4 | 290 | |
| 5 | 3II | 5 | 5 | 99 | 0.75 | 360 | |
| 6 | 3II, b | 5 | 3 | 100 | 0.35 | 870 | |
| 7 | 4I | 2 | 1 | 99 | 0.35 | 620 | |
| 8 | 4II | 2 | 1.75 | 99.5 | 0.3 | 740 | |
| 9 | 〚Rh(PPh3)3Cl〛f | 2 | 1.7 | 99 | — | 600 | |
| 10 | 3I | 5 | 70 | 94 | 18 | 10 | |
| 11 | 3II | 5 | 6 | 100 | 2.2 | 770 | |
| 12 | 3II | 2 | 8 | 100 | 5 | 200 | |
| 13 | 3II, b | 2 | 8 | 100 | 3.5 | 30 | |
| 14 | 4II | 2 | 4.5 | 100 | 1 | 410 | |
| 15 | 4II | 5 | 2 | 100 | 0.8 | 690 | |
| 16 | 〚Rh(PPh3)3Cl〛f | 1.5 | 1.9 | 96 | – | 840 | |
| 18 | 3II | 2 | 1.25 | 100 | 0.5 | 710 | |
| 19 | 4II | 2 | 1.25 | 100 | 0.35 | 860 | |
| 20 | 3II | 2 | 4.25 | 100 | 2.2 | 310 | |
| 21 | 4II | 2 | 5.25 | 100 | 2.4 | 250 |
Hydrogenationa of sterically demanding olefins in the presence of 3 and 4.
| Entry | Substrate | CatalystI/II | PH2 (atm) | Time (h) | Conversion (%) | t1/2 (h) c | iTOF (h–1) d |
| 1 | 3I | 10 | 72 | 14 | – | 10 | |
| 2 | 3II | 10 | 23 | 100 | 2.5 | 90 | |
| 3 | 3II, b | 10 | 32 | 99 | 3 | 160 | |
| 4 | 4I | 10 | 48 | 98 | 18 | 100 | |
| 5 | 4II | 10 | 25 | 100 | 3.6 | 120 | |
| 6 | 4II | 5 | 24 | 100 | 3 | 90 | |
| 7 | 〚Rh(PPh3)3Cl〛f | 0.8 | 150 | 100 | – | 12 | |
| 8 | 3II | 2 | 47 | 100 | 24.5 | 50 | |
| 9 | 3II | 5 | 27 | 100 | 12.5 | 60 | |
| 10 | 3II, b | 5 | 24 | 100 | 12 | 5 | |
| 11 | 3II, e | 70 | 36 | 90 | 0.75 | 300 | |
| 12 | 3II, b, e | 70 | 38 | 93 | 0.4 | 840 |
Hydrogenationa of selected olefins in presence of 5.
| Entry | Substrate | Solvent | PH2 (atm) | Time (h) | Conversion (%) | t1/2 (h) | iTOF (h–1) |
| 1 | CH2Cl2 | 5 | 25 | 99.5 | 8 | <5 | |
| 2 | MeOH | 5 | 30 | 99 | 14 | 45 | |
| 3 | CH2Cl2 | 5 | 108 | 99 | 34 | 60 | |
| 4 | CH2Cl2 | 10 | 17 | 100 | 8.5 | 60 | |
| 5 | CH2Cl2 | 10 | 137 | 99 | 60 | 5 |
The catalytic reactions were carried out in an autoclave, usually under mild temperature and pressure conditions, using the catalyst (21 μmol), either previously isolated or prepared in situ, and combined with the substrate (500 equiv) in various solvents, although methanol was proven to be the best choice in many cases. The evolution of all reactions was monitored by 1H NMR and GC analyses on samples withdrawn regularly from the autoclave. Details are included in the experimental section.
On the basis of a comparative analysis concerning the behaviour of the various catalysts reported herein, it is clear that the cationic complex 5 is less active than the neutral ones 3 and 4, although conversion rates close to 100% have been observed in all cases. Moreover, in order to achieve the completion of the catalytic reaction, the reaction times appeared to be relatively long, especially with simple olefins such as 1-hexene. We thus have focused our attention towards the more promising catalysts 3 and 4, which were found to be very efficient catalytic systems for the hydrogenation of a wide variety of olefins. The linear and cyclic olefins undergo rapid reduction under mild conditions, with conversion rates of nearly 100%, as depicted in Table 1. Notably, in the case of 1-hexene, the classical isomerisation side-reaction leading cis and trans-2-hexene is disfavoured, which explains the unusual high yields in production of hexane up to almost 100%. The results obtained with the sterically demanding olefins, which are known to be difficult to reduce, are also remarkable in terms of total conversion. However, the trans-stilbene requires higher pressures and longer reactions times, although the initial rate remains generally high. In this case, the reaction slows down drastically after 75% conversion.
Several points of interest concerning the activities of 3 and 4 need to be pointed out.
1. The influence of the addition of triethylamine (5 equiv) as a cocatalyst can be summarised as follows. In the hydrogenation reaction of 1-hexene, it results in slightly increased initial and overall rates (entries 1–2, 3–4, 5–6 in Table 1). However, with cyclohexene as substrate, a negative effect is mainly detected on the initial rate (entries 12–13 in Table 1). In the case of 1-methylcyclohexene (entries 2–3 in Table 2), the conversion rate decreases, whereas with α-methylstyrene, the catalyst is slower (entries 9–10 in Table 2). The overall negative effect of a base added as cocatalyst is likely to indicate that during the catalytic cycle, no deprotonation reaction is involved.
2. In all cases, better activities and higher rates are obtained when the catalyst is generated in situ, instead of the preliminarily isolated catalyst. This difference is probably related to its sensitivity towards oxygen and/or to some instability of the complexes in solution as noticed in the course of their NMR characterisation.
3. Finally, the 〚Rh(2)Cl〛 catalyst was found to be significantly more efficient than the related 〚Rh(1)Cl〛 catalyst. Such a marked improvement in activity may be the result of the size of the chelate around the metal. As opposed to a 6-membered ring around the rhodium in 3, the constrained 5-membered ring encountered in 4 may allow a better accessibility to the substrate on the opposite side of the rhodium, yielding a better activity as observed.
Overall, complexes 3 and 4 are very active catalysts in hydrogenation reactions. In particular, the activity of 4 towards 1-hexene is comparable with that of the well-known Wilkinson’s catalyst 〚11, 12〛. However, if one refers to Crabtree’s iridium catalyst 〚13, 14〛, our system appears less competitive. In presence of cyclohexene, the same trend is observed, although, based on a higher conversion rate, our catalyst represents an improvement on the results obtained with 〚Rh(PPh3)3Cl〛. It is also noteworthy that the reactivity of our catalysts depends upon the size of the cycle of the olefin in the order cyclopentene > cyclohexene > cyclooctene. Furthermore, 〚Rh(2)Cl〛 is by far a better catalyst than 〚Rh(PPh3)3Cl〛 towards the hydrogenation of α-methylcyclohexene, but not as good as Crabtree’s catalyst, in spite of the conversion rate of 100% observed in our systems and of their robustness. Finally, 3 also exhibits a good catalytic activity towards α-methylstyrene and trans-stilbene, for which to our knowledge rare data have been reported in the literature until recently 〚15〛.
Although the detailed mechanism of hydrogenation of olefins in these catalytic systems is not fully understood, we propose the catalytic cycle presented in Fig. 2. The first step would involve oxidative addition of H2 on the active species 3 or 4, resulting in the formation of a dihydride compound (I). Convincing evidence for the existence of such a species was obtained by studying the reaction of 3 with H2 which also gives rise to a complex containing two cis-hydride ligands on the basis of the 1H and 31P{1H} NMR analyses. Olefin coordination will then be facilitated by the hemilabile character of the ligand, as represented in (II). The formation of an alkyl derivative (III) via olefin insertion into one of the Rh–H bond follows. Finally, the active species is regenerated via reductive elimination of the alkane.
Proposed mechanism for the hydrogenation of olefins in presence of 3 and 4.
In conclusion, the new neutral rhodium complexes 〚Rh(L)Cl〛 are remarkably active for the hydrogenation of olefins. Higher activities are generally observed for 〚Rh(2)Cl〛 compared to 〚Rh(1)Cl〛, especially if the catalyst is prepared in situ. In the presence of 1-hexene, our catalyst performance competes well with that of the worldwide known Wilkinson’s catalyst 〚11〛. When cyclic olefins are involved, reducing the size of the cycle leads to a higher activity, and also the conversion rates of 100% represent a real improvement with respect to 〚Rh(PPh3)3Cl〛. Finally, based on the overall rate and the conditions used for the reduction of hindered olefins, our systems may also be considered as better catalysts than 〚Rh(PPh3)3Cl〛.
This small contribution was meant to illustrate that, more than 30 years after the discovery of Wilkinson’s catalyst, John Osborn’s passion for organometallic chemistry has never been altered and his scientific roots have always been firmly anchored.
Acknowledgements
We are grateful to Maurice Coppe for his help during the preparation of this paper. We also thank the CNRS for funding and the French ‘Ministère de l’Éducation nationale, de la Recherche et de la Technologie’ for a fellowship to J.-P. M.