1 Introduction

Polydentate ligands combining sp²-hybridized nitrogen and σ³-λ³-phosphorus atoms continue to play an increasing role in homogeneous catalysis and have already found numerous applications in processes of synthetic importance such as allylic substitution [1–4], hydrosylilation [5,6], transfer hydrogenation of ketones [7–9], hydroboration of olefins [10–13], olefin/CO copolymerization [14–20]... Though many different types of heteroditopic PN ligands with various backbones have already been synthesized and evaluated [21–28], significant efforts have still to be achieved in the design of new structures that will allow us to rationalize and to finely tune the influence of steric and electronic effects. So far, most of the efforts focused on the derivatization of the nitrogen moiety and only a few variations have been achieved around the phosphine substituent, which, in most cases, is acyclic [20,29–33]. Recently we launched a program aimed at exploring syntheses and applications of mixed PN ligands incorporating a phosphole ligand and a nitrogen heterocycle. A first encouraging result was obtained during the study of the (2,5-diphenylphospholyl)-2-methylpyridine ligand 1 [34]. This easily available ligand, which exhibits a remarkable stability toward air oxidation, was found to be particularly efficient in the ruthenium-catalysed transfer hydrogenation of ketones (TON up to 20 × 10⁶).

This first result prompted us to investigate more thoroughly the field of application of these mixed (2,5-diphenylphospholyl)-based PN systems. Herein, we report on the synthesis and the catalytic activity of PdCl₂ complexes of this ligand and one of its oxazoline analogue.

2 Results and discussion

Ligand 1 readily reacts with [PdCl₂(COD)] in dichloromethane at room temperature to afford complex 2, which was isolated as an orange powder (Scheme 1). All NMR data (³¹P, ¹H, ¹³C) confirmed the formulation proposed. In ³¹P NMR, the coordination of the phosphole unit moiety is evidenced by a downfield shift of 36.5 ppm (δ(1) in CDCl₃ = –0.2 ppm).

Additional evidence of the structure of 2 was given by an X-ray crystal structure analysis. Suitable crystals of 2 were obtained by diffusing hexane into a dichloromethane solution of the complex. An ORTEP view of one molecule of 2 is presented in Fig. 1. Crystal data are summarized in Table 1 and selected bond lengths and bond angles are collected in Table 2. As expected for Pd complex having the d⁸ electronic configuration, the overall geometry around palladium is square planar and nitrogen and phosphorus atoms adopting a cis-arrangement.

The catalytic activity of complex 2 was tested in three cross coupling reactions: the Heck [35], Suzuki [6,36–38] and Miyaura processes [39]. Modest yields and TON were obtained in the two first transformations. Thus, in the Suzuki coupling between bromoacetophenone and phenylboronic acid (in toluene under reflux), a conversion yield of 61% was obtained after 2 h heating using 1.10^–3% of catalyst (TON = 61 × .10³). Although this performance is reasonable, this TON does not compare with the best systems reported to date involving palladacycles [40,41]. Using the same loading of catalyst, bromobenzene reacted with styrene in the Heck coupling to yield stilbenes (ratio between Z and E not measured) with a 31% yield after 15 h of heating at 80 °C (TON = 31 × 10³) in dioxanne in the presence of triethylamine as base. Other solvents such as NMP, DMF, MeCN were also tested but did not yield improvements in terms of TON. Here again the catalytic activity of complex 2 proved to be modest by comparison with other systems [40,41]. Most interestingly, catalyst 2 proved to be very efficient in the Miyaura cross-coupling reaction that furnishes arylboronic esters. Nearly complete conversion and high TON (up to 1 × 10⁵) were obtained in the coupling of pinacolborane with iodoaromatics (see Table 3) after 48 h of heating in dioxanne at 80 °C using Et₃N as a base (Scheme 2). The conversion of some bromoaromatics was also attempted. Though lower conversion yields and TON (up to 9 × 10³) were obtained under the same experimental conditions, catalyst 2 still proved to be highly active compared to reported systems [42–45].

As a logical extension, we then turned our attention to the synthesis of optically active ligands featuring the 2,5-diphenylphosphole unit. The availability of numerous functional oxazolines prompted us to investigate the synthesis of a 2-methyl substituted derivative analogous to ligand 1. It must be noted that some oxazoline–phosphole-based ligands have already been reported, but with different backbones [46–48]. The synthetic strategy used is similar to that employed in the synthesis of 1. The 2-chloromethyl-5R-oxazoline was reacted with one equivalent of the 2,5-diphenylphospholide ligand in THF at room temperature (Scheme 3). Compound 3 was isolated after conventional work-up as an air-stable yellow solid. The formulation of 3 was confirmed by NMR spectroscopy, elemental analyses and X-ray crystallography. An ORTEP view of one molecule of 3 is presented in Fig. 2. Crystal data are summarized in Table 4 and selected bond lengths and bond angles are collected in Table 5.

Ligand 3 was conventionally converted into the [PdCl₂(3)] complex 4 in 94% yield, by reaction with an equimolar amount of [PdCl₂(COD)] in dichloromethane at room temperature (Scheme 4). All NMR data support the formulation proposed. Like for complex 2, coordination of the phosphole ligand results in a ³¹P NMR downfield shift of Δδ = +29.6 ppm.

Single crystals of 4 were obtained by diffusing hexanes into a dichloromethane solution of the complex at room temperature. An ORTEP view of one molecule of 4 is shown in Fig. 3. Crystal data are summarized in Table 4 and selected bond lengths and bond angles are collected in Table 6. As expected, complex 4 adopts a square planar geometry and, apart from the N1–Pd1 bond length, which is slightly shortened (2.016(5) Å in 4 vs 2.078(3) Å in 2), metric parameters within the metallacycle are quite similar to those of complex 2 (bite angle in 2: 84.2(1)° and 4: 83.6(2)°).

The catalytic activity of complex 4 was tested in the asymetric Heck coupling process between iodobenzene and 2,3-dihydrofuran. All the experiments were carried out in benzene or N-methylpyrrolidinone (NMP) between 40 and 60 °C using 0.1 or 1% of catalyst in the presence of triethylamine as base to yield the 2-phenyl-2,5-dihydrofuran (2-phenyl-2,3-dihydrofuran was also detected as traces < 2% in each experiment). All these results are summarized in Table 7. As can be seen, shorter reaction times and good conversion yields were obtained when NMP was used as solvent but no ee was recorded suggesting that NMP can displace the oxazoline ligand from the palladium centre. More convincing results were obtained when benzene was used as solvent, but the transformation proved to be sluggish. Thus an ee of 89% was obtained (with a 67% yield) after 5 days of heating at 40 °C with 1% of catalyst. As expected, increasing the temperature results in a significant acceleration but to the detriment of the ee. A comparison of these results with literature data clearly show that performances of catalyst 4 are modest compared to other systems (Scheme 5) [32,49–52].

In conclusion, we have shown that the [PdCl₂(L)] (L = 2,5-(diphenylphospholyl)-2-methylpyridine) complex 3 efficiently catalyses the cross-coupling reaction between pinacolborane and iodo- and bromoarenes. On the other hand, the [PdCl₂(L)] (L = 2-methyl-5R-phenyloxazoline-2,5-diphenylphosphole) complex 4 yielded disappointing results in terms of enantioselectivity in the asymmetric Heck coupling reaction between iodobenzene and the 2,3-dihydrofuran. Further work will now focus on the synthesis of differently substituted oxazoline-based derivatives of this 2,5-diphenylphosphole and their systematic evaluation in catalytic transformations.

3 Experimental section

All reactions were routinely performed under an inert atmosphere of argon or nitrogen by using Schlenk and glove-box techniques and dry deoxygenated solvents. Dry THF, ether, dioxanne, hexanes and toluene were obtained by distillation from Na/benzophenone and dry CH₂Cl₂ and CDCl₃ from P₂O₅. Dry CD₂Cl₂ was distilled and stored, like CDCl₃, on 4-Å Linde molecular sieves. Nuclear magnetic resonance spectra were recorded on a Bruker Advance 300 spectrometer operating at 300 MHz for ¹H, 75.5 MHz for ¹³C and 121.5 MHz for ³¹P. Solvent peaks are used as internal reference relative to Me₄Si for ¹H and ¹³C chemical shifts (ppm); ³¹P chemical shifts are relative to a 85% H₃PO₄ external reference and coupling constants are expressed in Hertz. The following abbreviations are used: b, broad; s, singlet; d, doublet; t, triplet; m, multiplet; p, pentuplet; sext, sextuplet; sept, septuplet; v, virtual. Mass spectra were obtained at 70 eV with a HP 5989B spectrometer coupled to a HP 5980 chromatograph by the direct inlet method. In the Heck coupling reactions, the ee was determined with a Perichrom PR2100 gas chromatograph equipped with a Lipodex A column. Elemental analyses were performed by the ‘Service d’analyse du CNRS', at Gif-sur-Yvette, whereas [PdCl₂(COD)] [53] and triphenylphosphole [54] were prepared according to literature procedures. All other reagents and chemicals were obtained commercially and used as received.

Acknowledgments

The authors thank the CNRS and the ‘École polytechnique’ for financial support of this work.