Unexpected Vulnerability of DPEphos to C−O Activation in the Presence of Nucleophilic Metal Hydrides

Abstract

C−O bond activation of DPEphos occurs upon mild heating in the presence of [Ru(NHC)₂(PPh₃)₂H₂] (NHC=N‐heterocyclic carbene) to form phosphinophenolate products. When NHC=IEt₂Me₂, C−O activation is accompanied by C−N activation of an NHC ligand to yield a coordinated N‐phosphino‐functionalised carbene. DFT calculations define a nucleophilic mechanism in which a hydride ligand attacks the aryl carbon of the DPEphos C−O bond. This is promoted by the strongly donating NHC ligands which render a trans dihydride intermediate featuring highly nucleophilic hydride ligands accessible. C−O bond activation also occurs upon heating cis‐[Ru(DPEphos)₂H₂]. DFT calculations suggest this reaction is promoted by the steric encumbrance associated with two bulky DPEphos ligands. Our observations that facile degradation of the DPEphos ligand via C−O bond activation is possible under relatively mild reaction conditions has potential ramifications for the use of this ligand in high‐temperature catalysis.

Hydride ligands: Under mild conditions (25→90 °C), [Ru(N‐heterocyclic carbene)₂(PPh₃)₂H₂]/DPEphos or [Ru(DPEphos)₂H₂] undergo C−O bond activation of this catalytically important diphosphine ligand by attack of the nucleophilic Ru−H ligands.

Since their introduction ca. 20 years ago,1 wide‐angle phosphines such as xantphos and DPEphos (Scheme 1) have become indispensable ligands for a range of catalytic reactions.2 Their usage stems from two advantageous properties; firstly, the availability of highly flexible bite angles that allow cis‐ and trans‐, as well as hemilabile P‐O‐P coordination modes, to be adopted3 and, secondly, resistance to the types of P−C degradation reactions reported in tertiary phosphine metal complexes.4 This latter property has promoted the use of xantphos and DPEphos in reactions that require high temperatures.2c, 2g, 2l, 5

Scheme 1.

Structures of xantphos and DPEphos.

Any suggestion that such phosphines might be susceptible to degradative reactions, particularly under relatively mild conditions, could therefore have important ramifications for their applications in catalysis. While xantphos has been reported to be susceptible to P−C bond activation at room temperature,6 cleavage of DPEphos appears to be restricted to a single example of high temperature C−O bond activation reported by Weller and Willis.7 In the course of studies on [Rh(η⁶‐ortho‐xylene)(DPEphos)]⁺ catalysed carbothiolation of alkynes, they reported that heating the Rh complex together with ortho‐MeSC₆H₄C(O)Me at 120 °C in the absence of any alkyne led to C−O cleavage of DPEphos to afford a catalytically inactive Rh complex with chelating phosphine aryloxide and bidentate phosphine arylthioether ligands. Herein, we demonstrate that C−O activation of DPEphos can take place even at room temperature in the presence of ruthenium dihydride complexes. DFT calculations reveal that such processes involve attack of highly nucleophilic hydride ligands on the aryl carbon on the C−O bond.

In the course of studies to investigate the substitution chemistry of the all trans‐dihydride complex [Ru(IMe₄)₂(PPh₃)₂H₂] (1, Scheme 2),8 1 was treated with 1.1–1.5 equiv DPEphos in benzene. No immediate reaction was observed at room temperature, but upon heating to 90 °C for ca. 12 h, a single ruthenium‐containing product 2 (Scheme 2) was formed. An X‐ray crystal structure (Figure 1) revealed the presence of a phosphinophenolate ligand generated upon C−O activation of DPEphos.9 The P,O‐termini of the ligand were trans to PPh₃ and Ru−H respectively. The coordination sphere was completed by two mutually trans IMe₄ ligands, each of which displayed an N‐Me group with a short C−H⋅⋅⋅O contact to the phosphinophenolate ligand (Supporting Information). The trans H‐Ru‐O arrangement led to both a long Ru−O distance (2.2720(16) Å)10 and a low frequency (δ=−18.40 ppm) hydride resonance.11

Scheme 2.

C−O activation of DPEphos by 1 to give 2.

Figure 1.

Molecular structures of (left) 2 and (right) 4. Thermal ellipsoids are shown at 30 % level. All hydrogen atoms, except for Ru−H have been omitted for clarity.

The formation of 2 was achieved under even milder conditions, although at the expense of longer reaction times (e.g. 6 days at 70 °C), and a low (5 %) yield could even be formed at room temperature, albeit only over 3 weeks.12 No simple substitution product arising from replacement of the two PPh₃ ligands by DPEphos was observed under these conditions (vide infra). Treatment of 1 with the more‐electron rich cyclohexyl diphosphine Cy₂P(C₆H₄)O(C₆H₄)PCy₂ also resulted in C–O activation, although the reaction failed to reach completion, even after heating at 120 °C for 2 days. There was no evidence for C−O activation of xantphos by 1.13

Replacing 1 by the N‐Et substituted carbene derivative cis, cis, trans‐[Ru(IEt₂Me₂)₂(PPh₃)₂H₂] (3, Scheme 3) led to an even more unexpected reaction with DPEphos. Heating in toluene at 90 °C gave the phosphinophenolate complex 4 (Scheme 3), in which the {Ph₂P(C₆H₄)} moiety generated upon C−O cleavage had combined with a C−N activated IEt₂Me₂ ligand to generate a Ru‐bound N‐phosphino‐functionalised carbene ligand.14 The X‐ray structure of 4 (Figure 1) showed the presence of a distorted octahedral ruthenium centre with a cis‐arrangement of the two carbenes and two phosphines and the same trans H‐Ru‐OAr arrangement as in 1 (Ru−O=2.265(2) Å). The phosphino moiety appended to N3 exhibited a considerable cone‐tilt, with Ru‐P‐C_ipso angles ranging from 102° to 132°. In support of the C−N cleavage process, the ¹H NMR spectrum showed just three NCH₂CH₃ methyl and six NCH₂CH₃ methylene resonances. The Ru−H resonance (δ=−17.7 ppm) was coupled to the two inequivalent phosphorus nuclei (δ=59 and 55 ppm) with cis‐² J(H,P) coupling constants of 20 and 15 Hz.

Scheme 3.

C−O and C−N activation to yield 4.

C−N activation of a metal‐bound NHC ligand has been described previously,15 including in studies on Ru‐NHC complexes related to those employed here.16 However, this process has only rarely been observed alongside the activation of another ligand,17 and, certainly not as a route to the formation of a phosphinocarbene.18

The C−O activation of DPEphos was not restricted to NHC‐containing ruthenium hydride precursors. The reaction of [Ru(PPh₃)₄H₂] with DPEphos gave the isolable cis‐dihydride complex [Ru(DPEphos)₂H₂] (5; Supporting Information),19 which upon heating to 80 °C overnight underwent C−O activation of one of the DPEphos ligands to afford [Ru(DPEphos)(Ph₂PC₆H₄O)H] (6, Scheme 4).20 This was characterized by the presence of a quartet Ru−H resonance at δ=−14 ppm with a ² J(H,P) splitting (22 Hz) indicative of hydride cis to all three phosphorus nuclei and a ³¹P{¹H} NMR spectrum which showed a triplet at δ=77 ppm (² J(P,P)=30 Hz), together with a broad, featureless signal at δ=50 ppm. We attribute the latter to the intact DPEphos ligand switching rapidly between κ ²‐P,P and κ ³‐P,O,P coordination. At −15 °C, this signal resolved into two doublets, the two ends of the DPEphos ligand becoming inequivalent as a result of the oxygen now staying bound to Ru. Although an X‐ray structure of 6 proved elusive, crystals of the chloride derivative 7 were isolated from CH₂Cl₂/pentane solutions of 6, affording a structure (Figure 2) which confirmed the coordination modes at ruthenium.

Scheme 4.

Formation of the C−O activated DPEphos complex 6 and chloride derivative 7.

Figure 2.

Molecular structure of 7. Thermal ellipsoids are shown at 30 % probability. Cl1 is disordered with a hydride ligand in a 75:25 ratio. Hydrogen atoms and all minor disordered components have been for clarity.

DFT calculations21 have been used to explore the mechanism of the C−O bond cleavage reactions in 1 and 5 and the factors promoting them. For 1, no intermediates are observed experimentally and so all free energies are quoted relative to this species plus free DPEphos. PPh₃ substitution in 1 by DPEphos gives [Ru(IMe₄)₂(DPEphos)H₂], 8, for which the all‐cis isomer, 8_ccc (+3.6 kcal mol⁻¹), and the cis, cis, trans‐isomer, 8_cct (+4.2 kcal mol⁻¹) are most stable.22

The accessibility of the trans dihydride isomer 8_cct suggested a hydride nucleophilic attack mechanism may be involved, similar to that characterised for the hydrodefluorination of (hetero)aromatics at trans‐[Ru(NHC)₄H₂] complexes.23, 24 Figure 3 shows the computed reaction profiles for this process in 8_cct and 8_ccc. For 8_cct the trans hydride arrangement gives a long Ru−H¹ bond (1.70 Å) and NBO calculations indicate significant hydridic character (−0.21). Nucleophilic attack proceeds via TS(8‐2)_cct at +25.0 kcal mol⁻¹, with a short H¹⋅⋅⋅C¹ distance of 1.56 Å and Ru⋅⋅⋅H¹ stretching to 1.84 Å. The C¹−O bond also lengthens to 1.48 Å and elongated C¹−C² and C¹−C⁶ distances in the aryl ring suggest a Meisenheimer‐type structure consistent with nucleophilic aromatic substitution. Hydride attack is also accompanied by a conformational change in the 8‐membered Ru−P−C=C−O−C=C−P ring, from a distorted twist‐boat conformation in 8_cct to a boat conformation in the transition state,25 similar to the DPEphos fac‐κ ³‐P,O,P binding mode.26 IRC calculations confirm that TS(8‐2)_cct links directly to 2_cct in which H² is trans to the phosphinophenolate oxygen. The lowest energy conformation of 2_cct is at −31.5 kcal mol⁻¹.27

Figure 3.

Computed free energy profiles (kcal mol⁻¹, BP86(benzene, D3BJ)) for hydride attack in 8_cct and 8_ccc, with selected distances in Å. Energies are relative to 1 plus free DPEphos and NBO charges at Ru and H¹ are indicated in italics for dihydride precursors. For clarity, IMe₄ ligands are truncated at the C2 position (i.e. C⁷ and C⁸ in the Figure) and phenyl substituents at the ipso carbons. DPEphos hydrogens are also omitted. 8_ccc’ is a conformer of 8_ccc that lies directly on the pathway for C−O cleavage (see text for details).

The equivalent reaction of 8_ccc involves an initial conformational change of the Ru−P−C=C−O−C=C−P ring to form 8_ccc’ at +14.5 kcal mol⁻¹. C−O bond cleavage then proceeds via TS(8‐2)_ccc at +34.1 kcal mol⁻¹ with similar geometric changes to those described above for TS(8‐2)_cct. The shorter Ru‐H¹ distances in 8_ccc and 8_ccc’ (1.65 Å) and lower NBO charges (ca. −0.12) indicate that H¹ is now less nucleophilic than in 8_cct, and this reflects the change in the trans ligand, from a hydride in 8_cct to IMe₄ in 8_ccc. This also correlates with C−O bond cleavage being less kinetically accessible in 8_ccc. TS(8‐2)_ccc leads to 2_ccc at −25.2 kcal mol⁻¹, substantially less stable than 2_cct as this structure lacks the favourable trans‐H‐Ru‐O arrangement.28

C−O bond cleavage was also modelled for [Ru(DPEphos)₂H₂] and the most accessible pathway is shown in Figure 4. The all‐cis isomer, 5_ccc, reacts via 5_ccc’ and TS(5–9)_ccc at +29.9 kcal mol⁻¹ to give a phosphinophenolate product, 9_ccc, at −23.0 kcal mol⁻¹. The short Ru−H¹ distance in 5_ccc (1.60 Å) and low NBO charge on H¹ (−0.02) indicate reduced hydride nucleophilicity compared to 8_ccc, although the barrier in the bis‐DPEphos system is actually lower (see below). In stark contrast to 8_cct, the trans dihydride isomer of [Ru(DPEphos)₂H₂] 5_cct, has a large barrier of +48.5 kcal mol⁻¹. This difference is due in part to the higher energy of 5_cct (+13.8 kcal mol⁻¹) and the reduced charge on H¹ (ca. −0.08 cf. −0.21 in 8_cct). The latter result highlights how the NHC ligands also serve to enhance hydride nucleophilicity. Differential steric effects in the transition states may also be a factor, as probed by calculations on 5_ccc and 5_cct in which the PPh₂ groups were replaced by PH₂. This model system gave a similar relative energy for 5_cct (+12.6 kcal mol⁻¹), but a reduced barrier for the subsequent nucleophilic attack (i.e. from 5_cct to TS(5–9)_cct: 30.2 kcal mol⁻¹ cf. 34.7 kcal mol⁻¹ in the full system). In contrast, the computed barrier for 5_ccc with the small model is 38.6 kcal mol⁻¹, 8.7 kcal mol⁻¹ higher than the full model.

Figure 4.

Computed free energy profile (kcal mol⁻¹, BP86(benzene, D3BJ)) for hydride attack in 5_ccc, with selected distances in Å and computed NBO charges at Ru and H¹ in italics for the dihydride precursors. For clarity, phenyl substituents are truncated at the ipso carbons and DPEphos hydrogens are omitted.

Computed geometries show significant distortions in the full model: in 5_ccc the trans‐P‐Ru‐P angle is 142° with the bulky PAr₃ moieties tilting over the hydride ligands. As this angle is only 160° in the small model, we speculate that the greater distortion of the full model enables nucleophilic attack.

Comparing [Ru(IMe₄)₂(DPEphos)H₂] and [Ru(DPEphos)₂H₂] shows C−O bond cleavage via 8_cct (ΔG ^≠=25.0 kcal mol⁻¹) is more accessible than in 5_ccc (ΔG ^≠=29.9 kcal mol⁻¹) and this is consistent with the lower reactivity of the bis‐DPEphos system observed experimentally. Lower barriers are computed with higher trans influence ligands (H>IMe₄) trans to the hydride nucleophile. The mixed NHC/DPEphos systems appear particularly vulnerable to C−O bond cleavage as the strongly donating NHC ligands both enhance hydride nucleophilicity and render 8_cct, the key trans dihydride precursor, accessible. The hydride attack mechanism described here has similarities to the “asynchronous oxidative addition” pathway described by Crimmin and co‐workers where a Ru^II metal centre acts as a nucleophile prior to C−O bond cleavage.12 A similar asynchronicity is seen here, with C−H bond formation in the transition state being far advanced of either C−O bond cleavage or Ru−O bond formation.

In summary, we have characterised the surprisingly facile C−O bond activation of DPEphos ligands in the presence of nucleophilic hydrides. Ligand exchange of all‐trans‐[Ru(IMe₄)₂(PPh₃)₂H₂] with DPEphos results in the formation of phosphinophenolate complex, 2, while with cis, cis, trans‐[Ru(IEt₂Me₂)₂(PPh₃)₂H₂], C−O bond cleavage is accompanied by C−N activation of the NHC to form the N‐phosphino‐functionalised carbene complex 4. DFT calculations indicate that C−O activation involves a nucleophilic pathway in which a hydride ligand attacks the aryl carbon of the DPEphos C−O bond. This process is promoted by the accessibility of a trans dihydride intermediate that features highly nucleophilic hydride ligands. C−O bond activation also occurs upon heating cis‐[Ru(DPEphos)₂H₂], a process that DFT calculations indicate is promoted by the steric encumbrance of the mutually cis DPEphos ligands. This undesirable ligand degradation of DPEphos is of particular note given the wide use of this ligand in high temperature homogeneous catalysis. Indeed, degradation of the Rh‐DPEphos system described by Weller and Willis is also thought to involve nucleophilic attack, in this case by a thiolate ligand.7 On a more constructive note, the hydride nucleophilic attack mechanism proposed here has already been shown to operate in catalytic C−F functionalization,23c, 23d and so may also be an effective means of promoting C−O bond activation of the type required for the valorization of lignin and of its highly oxygenated monomers.29

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

M. K. Cybulski, N. A. Beattie, S. A. Macgregor, M. F. Mahon, M. K. Whittlesey, Chem. Eur. J. 2020, 26, 11141.